(but actually on how we use AI tools in media)

I.

“Our first task is neither to demonize nor idolize technological tools, but to utilize them on the basis of a fundamental principle, namely that truth is a common good and not the property of those with power or influence.” — Pope Leo XIV, Magnifica Humanitas, 2026

Our zeroth task is to determine whether artificial intelligence is simply a new technological tool or if it represents a new form of consciousness capable of transcending human thought. Is there something innately special about human consciousness that is impossible to replicate in a machine of our own creation?

It is safe to assume the Pope’s position on the matter, and we will certainly not discover the answer here, but it is worth considering the implications of the materialist perspective that we can arrange matter in a specific pattern to create consciousness.

If yes, then AI is not just a tool. We could theoretically create an artificial superintelligence capable of replacing the need for human intelligence altogether. In that scenario, AI is the next step in the evolutionary chain of conscious experience of the universe.

Solving the hard problem of consciousness is outside of the scope of this week’s introduction to Aemula’s media spotlight, but be sure to subscribe so you do not miss when we eventually get around to it.

However, we can at least question whether the current generation of AI tools are capable of truth-seeking, which we would define as the process of curiosity driving the continuous search for new information to update an understanding of reality based on novel discoveries that challenge the currently accepted set of facts.

Our basic human endeavor, as faulty creative creatures, is one of truth-seeking. We do not know everything (yet), but we leverage individual experience and divine inspiration to pose novel ideas to be considered for acceptance into our consensus reality.

But do novel ideas even exist? Are there unknown variables that are impossible to discover with current knowledge and observable reality?

An internal OpenAI model disproved the unit distance conjecture, one of the remaining unsolved Erdős problems in mathematics, which had eluded mathematicians for eight decades. It was an objectively novel solution, in the sense that it was not known beforehand, that advances our understanding of mathematics.

What do we make of this?

Sure, a sufficiently powerful computer can solve sufficiently difficult math problems, but does that same line of reasoning extend into reality? If understanding reality (discovering “the truth”) is just a difficult math problem, then throwing a big computer at it will result in a solution (or at least a partial solution).

Is math the map or the territory? Are LLMs just really good cartographers or are they the next generation of explorers?

It is safe to say that we can assume the Pope rejects materialist determinism. If we humans have free will and our understanding of reality is not just a hard math problem, then AI is just a tool to assist the human truth-seeking endeavor. Albeit, a very powerful tool.

We agree with the Pope in that current LLMs are extremely powerful tools (AI isn’t going to steal your job).

But not to be distracted by theology and philosophy of mind, this is supposed to be an examination of AI use in media. If AI is only a tool, then the question is not whether we should use it, but where it belongs in the creative process.

II.

“As knowledge becomes increasingly fragmented, it becomes difficult to grasp reality as a whole, to ask profound questions about meaning, or to develop authentic, critical and creative thought.” — Pope Leo XIV, Magnifica Humanitas, 2026

Digital platforms shape our collective imagination. We leverage our information environments to construct the worldviews through which we assess reality. Information is a tool in our truth-seeking endeavor.

Yet, there is more information available than we, as humans, can possibly comprehend. Naturally, this leads us to each hold unique subsets of the full set of available information. The specialized beliefs we develop from our unique perspectives result in the fragmentation of knowledge. We form distinct ideological communities that hold competing sets of consensus facts.

Through communication across these communities, we share ideas and challenge beliefs in our effort to arrive at the truth. In our modern era, this process of communication is “the news” or “digital media”. Some of it is purely for entertainment, some of it is to share newly discovered facts, and some of it is a combination of the two, but all of it is information that influences our beliefs and experiences.

Alternatively, LLMs are capable of holding the full set of human knowledge (if given enough computing power), compressed into an interface we can prompt for answers.

Yet, in their current form, LLMs flatten thought and homogenize expression. ChatGPT decreases idea diversity in brainstorming. While LLMs are incredibly powerful at aggregating and compressing existing information, they collapse viewpoints around the most statistically probable response to any given prompt.

Is the most probable response the truth?

If human knowledge is incomplete and truly novel ideas still exist, then the most probable response is simply the best approximation of truth based on current information. We do not know if it is an accurate representation of truth yet, but we do know that we cannot arrive at The Truth by solely relying on approximations from incomplete sets of facts.

To generate new information, we need ideological diversity. We need unique human experience and observation as an input to continue refining the most probable response on the path towards The Truth. Yesterday’s consensus does not inherently yield tomorrow’s innovation.

Yes, novelty can arise from the recombination of existing ideas, as demonstrated by OpenAI’s model disproving the unit distance conjecture, but recombinant innovation still relies on embodied observation and value judgement from humans. LLM model weights reflect a point-in-time snapshot of this human observation and judgement.

Within the framing of a specific goal, AI can brute force new combinations of existing information until it achieves success in that goal. Infinite monkeys hitting keys on a typewriter will eventually produce the paper disproving the unit distance conjecture, but its significance would unfortunately be lost on them (though the monkeys would probably evolve Homo sapiens level intelligence before they randomly typed the paper and could then understand its significance, but they probably would have questioned their existence by that point and stopped typing).

Similarly, an AI model may produce a valid result to a posed question, but transformational human creativity remains necessary to frame the question, interpret the answer, and update the goal. “To ask the right question is harder than to answer it”. We need humans in the loop to continue setting the path for progress.

It is for this reason that we believe a human-centric journalistic process is necessary for the continued refinement of LLMs. It is for this reason that AI research labs have committed over $3b across the next three years to license newsroom content.

And it is for this reason that we cannot replace the journalistic process with AI. Feeding AI-generated content back into the data aggregation pipeline only collapses our consensus “truth” on the most probable response that exists today. If the information we have today is incomplete, then the process of circular AI feedback does not seek the truth. The goal remains unchanged. Artificial intelligence finds itself racing towards a local minimum in a static landscape rather than progressing to The Truth.

III.

“The speed and simplicity with which information, complex analyses, media content and practical assistance can be accessed undoubtedly makes life easier. Yet they can also encourage excessive reliance and the search for ready-made answers, and weaken personal creativity and judgment.” — Pope Leo XIV, Magnifica Humanitas, 2026

Information is abundant. Attention is scarce.

More content is produced than we can possibly consume. To differentiate signal (quality) from noise (slop), we rely on complex algorithms to curate our news feeds.

Quality content is expensive to produce. It takes time and effort to harvest new information, cut it with unique perspectives, and package it in a format that is easy and entertaining to consume.

In the not-so-distant past, we could rely on a fairly simple heuristic to make our own judgments on slop versus quality. Short-form content was likely slop, either clipped or hastily recorded to be pumped out into the world to capture easy engagement, while long-form content such as news articles, podcasts, and movies took effort to produce and was more likely to be quality. Of course, there is variance within each category, but we roughly knew what we were getting when we chose to consume a specific form of content.

AI has changed that calculus. LLMs can produce coherent long-form articles, research papers, podcasts, and even movies without much human effort at all (though effort still exists in the considerable data aggregation, model training, computing resources, and energy required to produce the output from the LLM).

But again, the result is a statistical average response. One that often lacks the human judgment necessary to progress our search for truth.

It will only become easier and cheaper to produce slop that mimics quality content. It is for this reason that human consumers of information are skeptical of AI-generated content. Don’t ask why, but humans are eerily adept at identifying when they are interacting with something that isn’t human. If someone did not go through the time and effort to produce a piece of content, why would it be worth our time and effort to consume it?

The result is that AI-generated content has been largely rejected in long-form media, with public backlash against the use of AI-generated writing in news reporting or AI-generated animation.

However, AI remains an incredibly powerful tool to quickly research vast amounts of information. As long as the sources supporting the results of AI-assisted research are verifiable, then it should be utilized at some point in the journalistic process.

There are lines appearing in the sand pointing to the socially acceptable uses for AI in content creation.

Utilizing AI for brainstorming and idea generation results in a lack of truly creative ideas. The end consumer might not be able to tell, but the idea likely won’t contribute meaningfully to our media environment. We would hope that human-generated ideas would outcompete content discussing topics chosen by LLMs.

Utilizing AI to generate the actual content skips the valuable step of human input for introducing novel ideas, creative storytelling, and curating a specific experience valued by the target consumer. Go read a bunch of AI-generated writing and you will find that you rarely feel anything. We would hope that human-generated content would outcompete the creativity of LLMs.

However, utilizing AI to research and process vast amounts of data during the creative process can guide human thought and accelerate the production of valid output. We could increase the amount of quality content floating around our media environment so that it doesn’t get lost in the sea of slop. In this case, LLMs are the tool to assist the human-centric truth-seeking endeavor. This appears to be the internal policy of The New York Times.

AI cannot replace human thought, originality, and judgment, from which meaningful writing emerges and worthwhile goals are framed.

LLMs are tools to help humans find information faster. They can help us test our existing assumptions. They can help us see to the edges of what is already known. But they cannot seek the truth without us.

This week, we highlight writers discussing AI and its role as a tool in the creative process. We encourage you to explore their work and consider subscribing directly.

Media in the Media

Theme: Where do we draw the line on acceptable use of AI in the writing process?

What 370,000 College Essays Tell Us About A.I.’s Effects on Creativity — The New York Times

• Personal statement essays post-ChatGPT were rated as more creative by human judges, but were found to lack truly creative ideas

• “In a separate study, the team found that human-written essays offered up to eight times more new ideas than those produced by A.I.”

• ChatGPT decreases idea diversity in brainstorming

Theme: Data security, anonymity, and source protection are critical in protecting our information sources from external influence

Thousands of journalists’ data exposed to dark web — Semafor

• Proton discovered 116k instances of journalists at NYT, WSJ, and WaPo having their personal data, including emails, names, phone numbers, DOBs, and addresses leaked on the dark web

• The leaks appear to be due to your garden-variety data exposure inherent to using the internet rather than security breaches of the publications

• “‘Leaked passwords open doors to email accounts, internal systems, and communication platforms where source identities could be exposed,’ and the release of reporters’ personally identifiable information ‘creates opportunities for blackmail or targeted harassment campaigns designed to silence or discredit’ them.”

Trend: Companies are going all in on communication strategies to directly connect with their customers

Visa, the payment company, is hiring an Editor-in-Chief for their “Newsroom”

• $245k top salary range and 10+ years of experience in relevant editorial and journalism roles

• Visa’s global newsroom currently appears to be comprised of press releases and the @VisaNews social accounts

• The role description seems to imply an alignment with more of a traditional newsroom editorial role in setting editorial standards and narrative coherence

• Storytelling is an increasingly important aspect of market efforts for consumer businesses, and this role captures that trend through “ensuring Visa’s storytelling reflects evolving realities across markets and keeps the organization equipped to address both current and emerging narratives”

Spotlight

Young Money

Written by Jack Raines, author of the upcoming book Young Money, a writer, and venture capitalist whose work blends behavioral finance, technology, markets, and twenty-something life experience, drawing on an unconventional path from small-town Georgia to startups, investing, and writing online.

Young Money by Jack Raines

Audience Capture and Short-Form Slop

Welcome to Young Money! If you’re new here, you can join the tens of thousands of subscribers receiving my essays each week by adding your email here. And if you like my blog, you should pre-order my first book, Young Money, by clicking below…

Read more

5 days ago · 85 likes · 10 comments · Jack Raines

“Every platform is extracting from their audience rather than serving them. What’s the optimal ad load we can hit someone with without them churning? Short-form video has proved to be a digital fentanyl that can keep us scrolling for hours. Stretched truths garner far more attention facts, so those are going to be amplified, because, again, more time on screen = more ad revenue.

And the content creators themselves, in desperate bids to “keep up with the times,” are just as guilty, turning to more and more aggressive strategies for “hooking” a viewer so they too can make a living by monetizing this engagement with ads.

The result: very little content on the internet is “good,” as the entirety of the internet, from platforms to posters, is now optimized for engaging viewers where they are: passively scrolling for hours a day. Quality is sacrificed for the sensationalistic and the absurd, as anything that’s actually quality takes time to create, and the only way to stay top-of-mind is by churning out more and more content as quickly as possible. And we wonder why 99% of the timeline is ephemeral slop.”

Ghost of Electricity

Written by Alex Imas, a behavioral economist and Director of AGI Economics at Google DeepMind whose research examines how artificial intelligence is reshaping productivity, labor markets, decision-making, and creative work through the lens of economics and human behavior.

Ghosts of Electricity

What will be scarce?

Starbucks is a huge company (market cap of $112 billion) that sells one of the most standardized products in the modern economy. Making a cup of coffee or even one of the fancy specialty drinks is very easy to mechanize and reproduce. If the entire economy is soon to be automated, with labor being replaced with increasingly more sophisticated capital, S…

Read more

2 months ago · 1769 likes · 155 comments · Alex Imas

“The critical link to AI comes from new work with Graelin Mandel. We find that AI involvement undermines the perceived exclusivity of a good; objects with AI involvement are perceived as inherently reproducible and non-unique. People bid for physical art prints that varied in described AI involvement. Human-made artwork gained 44% in value from exclusivity (one copy vs. many), but AI-generated artwork gained less than half that, only 21%. The mere involvement of AI made the work feel inherently non-exclusive, as if it could always be reproduced, regardless of how many copies were said to exist.”

Shive Talkin’

Written by Kevin Shively, a writer and commentator whose wide-ranging background across music, technology, media, business, and religion informs an open-ended exploration of politics, culture, and the human experience, approached with curiosity rather than ideology.

Shive Talkin'

Keys to the Kingdom

In 1999, my entire career ambitions changed. I got a job working in an industry that didn’t even really exist yet. Actually, it was a media company, so that part of the industry obviously existed, but this was using a new technology that had no real infrastructure yet to sustain it. I went to work for an internet radio station…

Read more

3 days ago · Kevin Shively

Read on Aemula

“Artificial intelligence provides the tools for almost anyone to produce content on a scale that can compete with teams of professionals in a wide array of disciplines. While this is potentially threatening to the status quo, it also presents opportunities that could never have been considered prior to its arrival. It also presents potential harms to society including economic, environmental, and even potentially cultural, but each of these could be accounted for with proper foresight and regulation.”

Requests for Reporting

We are building out new features to allow our community to create requests for reporting, which serve as bounties to incentivize independent journalists to report on under-covered stories of interest.

• Comment or message to submit a request

• Requests for Reporting will be included here each week

• Publish an article on a requested topic through Aemula to earn a bounty

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

— Julian Assange, “Self destructing paper”, with pull-quote credit to Gwern

I.

We typically imagine the internet to be an infinite archive. Growing up with social media, it was responsible to understand that anything shared online should be considered to be permanently available to the public. Yet, digital information remains more transient than we believe.

We generate unimaginable amounts of data, building up rich histories of interactions with social platforms as we upload photos, chat with friends, and bookmark posts. Our digital footprint is a detailed reflection of our personal lives, which serves as a modern replacement for photo albums, journals, and letters as a way to memorialize our past. The difference now is that we do not control our digital twins.

A significant portion of my own memories are recorded as Instagram posts, Facebook albums, saved Snapchats, group messages, and iCloud photos. My phone was stolen and I lost all of my saved passwords and two-factor authenticators. I was locked out of all of these accounts. My recovery email was a college account I could no longer access. My backup recovery email was deleted by Google for inactivity. These memories were lost. They exist on Meta and Apple servers, gated behind accounts I cannot authenticate. They are not mine.

I am still bitter about Microsoft acquiring Mojang, the game studio behind Minecraft, in 2014, which blocked me from my save files after they migrated my data to a Microsoft account I could not access. Spotify removes songs saved to my playlists. Amazon updates and removes books from my Kindle.

Admittedly, these are relatively benign examples. I was able to recover a decent portion of my lost photos through friends and lengthy account recovery processes, but some memories were lost forever. However, the purpose of these examples is to show that we do not actually own our data if it lives on someone else’s servers. Its existence and preservation is at the whim of the actual owner, while we merely plea for visitation rights every time we attempt to sign in. If the company running the servers ceases to exist, changes their policies, or has an outage, your data is lost with it.

II.

A more noteworthy example of the internet’s impermanence came last week when ABC removed the FiveThirtyEight archive from their website. The founder and former editor-in-chief of FiveThirtyEight, Nate Silver, a renowned election forecaster who now writes Silver Bulletin here on Substack, lost years of hard work produced by him and his team. The internet lost valuable historical insight into previous reporting and analysis of election polling data because ABC refused to sell the rights of the content back to Nate Silver before deciding to delete the archives.

“Ownership” is a social construct, and we largely rely on institutions and systems to recognize and enforce our rights to access and control the things that we “own”, both physically and digitally. The vast majority of the time, this is not a problem. It is a convenience that is well worth the tradeoff in most cases. At least until the system fails.

We trust that Meta will keep our personal data secure, handling it in line with agreed upon terms, and protecting access so we are the only ones who can control our accounts. We do not worry about this until we find out about events like Cambridge Analytica’s mass extraction of personal social data from Facebook to create extensive voter profiles for targeted political ads outside of Meta’s data policies.

We trust that our bank balances represent real dollars that we own, while trusting the bank to ensure our money is safely held and conveniently available to use for purchases. We don’t worry about this until events like the collapse of Silicon Valley Bank risked customers losing their uninsured deposits until the Treasury, Federal Reserve, and FDIC invoked a systemic risk exception to bail out the failed bank.

Again, these systems are exceedingly reliable and work as expected the vast majority of the time. We accept everyday convenience in exchange for the low-probability risk of failure.

Do not take this as a recommendation to empty your bank account and bury gold in your backyard. Don’t delete your social media accounts in favor of analog communication like you’re Christopher Nolan. Sure, after losing my photos, I now keep redundant backups on physical drives, but that is overkill born from a stubbornness to not make the same mistake twice.

That said, you should still be aware of the risks that arise when you rely on trust in third-parties to endorse your ownership of the things you value.

This is especially true when you are building your livelihood on digital-native platforms operated by third-parties. Independent creators put significant time and effort into producing their work to be uploaded and shared online. Yes, you own the work and you have the right to the revenue share from its success, but can you actually say you own something that you don’t control?

III.

Similar to the control exerted by ABC in their deletion of FiveThirtyEight, independent platforms like Substack have the right to delete your publication. They do a commendable job of upholding the freedom of speech and not deleting accounts (arguably erring in the opposite direction), and I personally trust them to preserve my own content. But that can’t be guaranteed into the future.

When you publish on Substack, your post is stored on their servers. You might be able to export your mailing list, but you rely on their servers to send your emails, maintain your archive, and distribute your posts (you also rely on their servers to be able to export your mailing list). If they determine they don’t want you on their platform, or if Substack ceases to exist, your work will be lost.

Your revenue from subscriptions are paid through your Stripe account. It has your name on it, the number on your account balance represents dollars that you have earned, and your subscribers are paying you for the work you are creating.

Yet, that is not how the funds actually flow. Your subscribers are sending money into Stripe’s corporate settlement accounts, where they are held before being released to your connected bank account. Stripe has broad rights to delay, withhold, cancel, reserve, reverse, or offset your earnings at their discretion if they believe there is risk associated with your expressive content.

I am confident to say that you will not run into issues with these systems, simply because the risk is so low. The majority of Substack writers are posting casually, earning from engaged subscribers, and Stripe has no reason to investigate further (though you should backup your mailing list regularly).

However, if your goal is to make a living by publishing independently, you should hesitate to let the entirety of your work, all of the IP and value you have created, exist solely in a place that is outside of your control.

And this is more important now than ever.

IV.

As people increasingly rely on AI chatbots for information discovery, we are seeing a decrease in organic web traffic to media sites. Organic search referrals to news publications decreased 38% in 2025. The industry is shifting their focus to build information pipelines meant for consumption by LLMs. And they should, since that seems to be the direction we are headed.

AI is becoming an increasingly prominent consumer of information, through agentic research and compiled summaries in chat responses. In 2025, AI web traffic grew from 1 AI bot for every 200 human visits (0.5%) to 1 AI bot for every 31 human visits (3.2%).

This creates a new distribution channel and revenue stream for creators and reporters as long as they are able to ensure they are compensated when an AI bot accesses and references their work. As a result, we have seen a wave of new companies offering AI content licensing for independent creators, such as Parallel, TollBit, and DripStack (and Aemula). If bots are going to be the ones reading your work, you should at least get paid for it.

However, in a future where the majority of your earnings come from AI consumption, data ownership is critical. When you have human readers, you are building real connections with real people. If your data is lost or removed from a platform, at least people still know you exist. They can find you again or follow you to a new platform or continue subscribing as you rebuild your archive. Your data may have been lost, but your business can persist.

AI bots are different because they are indifferent to your existence. From their perspective, data is your only output. If that data ceases to exist, they will just find another source. The only value you are creating for a bot is the data you produce, not the relationships you form with an audience, the reputation you build over time, or the impact your work has had on your readers. If your data is lost, you are dead to them. The value you have created is gone forever.

Parallel, TollBit, and DripStack offer a monetization layer for when bots access your work, but they assume that you are relying on a third-party to store and persist your data. You can earn from your work as long as your platform of choice chooses to maintain it. But they can just as easily make it vanish.

If you are working independently to grow a media brand, publish digitally, and earn a living for your efforts, you need to ensure that you actually own your content. Not “own” in the sense that your digital existence relies on trust in some third-party, but “own” in the sense that you actually control your own data.

V.

I know we said we would be more optimistic in these newsletters. This hasn’t been the rosiest of messages. That said, we are actually quite optimistic on this topic. All of the fear mongering up until this point is not actually a problem at all since it can be easily solved.

This is not some doomsday prepper guide pushing for the digital equivalent of hoarding canned food and ammo in a bunker behind your house. I personally don’t believe we are on the verge of the singularity. AI isn’t going to Wall-E us any time soon.

Instead, we can acknowledge that AI agents are set to make up a significant share of revenue for digital creators in the next few years based on observations of recent trends. In that world, data ownership is critical for the stability of your business.

You wouldn’t say you owned a house if you didn’t have the keys or deed. Owning a digital media business follows the same logic. You should self-host your data (hold the keys) and have immutable digital records of your ownership (hold the deed). With today’s technology, this is relatively easy to set up, and it gives you actual ownership and control of your work.

However, we also realize that most people don’t want to self-host their content, since maintaining servers and payment infrastructure would be a distraction from your time spent actually creating.

The good news is that we have spent years designing a publishing protocol with the sole focus of ensuring true independent ownership of digital content. If you are building an independent media business on Substack and want to self-host your data and have immutable digital records of your ownership, you can set it all up in two minutes on Aemula. Just follow this link and follow the prompts.

Once set up, just continue publishing on Substack as you normally would, and we will create a copy of your post to be stored and served on a peer-to-peer network. “Peer-to-peer” means that rather than storing your content on a server we control, it is securely stored on a network of independently operated servers. If one server decides they don’t want to store your content, that’s fine. The others will continue to do so. You can even become a peer in the network yourself to ensure your content is always available without relying on anyone else.

That might sound complicated, but it is much easier than you may think. If you want to self-host, just let us know. We will send you a box you that you plug into your wall. It will automatically backup and store anything you publish through Aemula and continue to serve it to your readers as long as you have it plugged in.

Additionally, your ownership of your content is tied to an account that you control on a peer-to-peer network. If Aemula ceases to exist, your account persists. We can’t lock you out of it since we don’t control it. You hold the key and can use the account from anywhere on the network.

We also offer a monetization layer for bots to access you work, but since you own and control your content, you control this access. If you want to allow AI bots to license your work, you will get paid directly when they do. If you don’t want to allow bots to license your work, you can block them. Many AI bots scrape free content from Substack whether you want them to or not. Everything you publish through Aemula is encrypted, so AI bots can only access it if you allow them and they pay you for it.

As we head towards a future where AI agents make up an increasing portion of digital media consumers, the value of your work consolidates around the actual data you produce. Make sure you are the one who owns it.

This week, we highlight writers discussing content ownership and AI licensing. We encourage you to explore their work and consider subscribing directly.

Media in the Media

Theme: The freedom of the press

New York Times Files Second Lawsuit Against Pentagon — The Wall Street Journal

• The New York Times escalated their fight over rules that limit press access inside the Pentagon

• The new lawsuit targets the Pentagon’s interim press policy requiring journalists to have an escort inside the building

• A federal judge previously ruled in favor of the Times that an earlier version of the policy was unconstitutional

• The Times claims the policy is unconstitutional by restricting reporting on military operations and government accountability

• The Pentagon claims they are balancing transparency with security around the disclosure of sensitive information

Theme: The platforming problem of centralized media

Beehiiv founder gets pushback from writers after posting about right wing LA mayoral candidate — Revenue Rulebreaker

• Tyler Denk, co-founder and CEO of Beehiiv, tweeted an endorsement of LA mayoral candidate Spencer Pratt

• Users reacted with backlash about Denk’s political leanings and concerns about its effect on the operations of the platform

• Users related this to similar backlash against Substack for platforming far-right newsletters

Theme: Even in the age of AI, we still need humans to think freely and share their ideas

Anthropic co-founder Jack Clark speaking on the importance of reading actual source material and not just relying on AI summaries

Spotlight

Silver Bulletin

Written by Nate Silver, a statistician and renowned election forecaster, founder and former editor-in-chief of FiveThirtyEight, professional poker player, and author of The Signal and the Noise, previously featured in our spotlight “A Dying Industry”.

Silver Bulletin

Disney erased FiveThirtyEight

Last Thursday night, I was working late, trying to put some of the finishing touches on our forthcoming World Cup model — and actually looking up an article I’d written for FiveThirtyEight in 2014 about my previous soccer model, SPI. Although the quality of the archive has gradually deteriorated since Disney…

Read more

11 days ago · 166 likes · 8 comments · Nate Silver

“It would be nice if that work could be preserved for the public record. I don’t know what plans Disney has for FiveThirtyEight, if any. But I did approach Disney a year or two ago, through my agent, about acquiring the remaining IP. I’m probably the logical high bidder, though the value is rapidly depreciating as what’s left of the site falls into disrepair. At a minimum, we’d restore the archive, with prominent links to Silver Bulletin.

We were told to basically get lost: ABC was annoyed with my critical public comments about their management of FiveThirtyEight. It apparently wasn’t a long conversation, so I don’t have a lot more color to report than that.

Hanlon’s Razor states: Never attribute to malice that which can be adequately explained by stupidity. But honestly, I don’t know which explanation is better suited to ABC.”

Card Catalog

Written by Hana Lee Goldin, a library and information science researcher whose work explores AI fluency, information literacy, and the ways people can navigate digital systems with greater clarity, confidence, and critical thinking.

Card Catalog

So What if They Have My Data?

If you’d rather listen than read, I recorded an audio version of this essay for paid subscribers at the end of the post. Thank you for being here…

Read more

a month ago · 143 likes · 12 comments · Hana Lee Goldin, MLIS

“AI companies are training their models on personal data scraped from the web. Web crawlers pull content from blogs, social media profiles, online marketplaces, photo-sharing platforms, and anywhere else that isn’t behind a login wall. A 2025 MIT Technology Review investigation of a major AI training dataset found thousands of identifiable faces, identity documents, and job applications in a sample representing just 0.1% of the data, and estimated the full set contained hundreds of millions of images with personal information. Meta has said its AI models are partially trained on public Facebook and Instagram posts. LinkedIn began using member data to train its AI tools before updating its terms of service to reflect that change. Because AI companies have scraped content posted long before generative AI existed, the people whose data is in those training sets never had the opportunity to consent to that use. And there’s a feedback loop: we generate data by using AI systems, that data refines those systems, and those systems shape the information environment we navigate. Our data becomes part of the infrastructure that determines what information reaches us next.”

Realtime Techpocalypse Newsletter

Written by Émile P. Torres, a philosopher, historian, and journalist whose work examines human extinction, AI ethics, and the ideologies shaping Silicon Valley’s technological ambitions, with a particular focus on propaganda, existential risk, and the cultural narratives surrounding emerging technologies.

Realtime Techpocalypse Newsletter

The Guardian, Wired, TIME, and The Atlantic Are All Collaborating with OpenAI

Who’s Working With OpenAI…

Read more

14 days ago · 63 likes · 7 comments · Émile P. Torres

“I published a book in 2024 with Routledge titled Human Extinction. Routledge is owned by Taylor & Francis Group, which, according to an email from them, “entered into agreements with two AI companies to provide access to our content for training,” one of these being Microsoft. The email added that “standard publishing contracts do not enable authors to opt out from a specific use of their content (whether that is for AI or Text and Data Mining (TDM) or other types of licensing opportunity).”

(…)

This deal was apparently finalized just after my book was released, and I was never once informed that years of my intellectual labor would be sold to Microsoft to train fucking LLMs. Sorry for cursing, but this really upsets me. No one asked for my permission or consent, nor was I compensated financially.”

Requests for Reporting

We are building out new features to allow our community to create requests for reporting, which serve as bounties to incentivize independent journalists to report on under-covered stories of interest.

• Comment or message to submit a request

• Requests for Reporting will be included here each week

• Publish an article on a requested topic through Aemula to earn a bounty

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

The collective intelligence of humanity — our ability to solve complex problems and coordinate at global scale — is an emergent trait of our individual decisions. We work together to create something that exceeds the sum of its parts.

Whether we realize it or not, we all collaborate as one large neural network, achieving things that no single person could possibly achieve and no single entity could possibly coordinate.

Society has become incredibly complex, but this complexity has provided us with the knowledge and resources necessary to grow the global population and generally improve quality of life. As we continue this progression, we must constantly work together to solve increasingly complex problems. Progress is not inevitable, and the solutions will not be given to us.

Our ability to solve these problems is a function of our collective intelligence. People receive, process, and share information to collaborate at societal scale, conducting the global economy and dictating geopolitics. The power of this neural network is only limited by the number of people involved and the ease at which they can communicate.

We can communicate efficiently. Anyone, anywhere can exchange messages in an instant. But are we communicating effectively? Are we ensuring the necessary information is being routed to the people most capable of processing and acting upon it?

Currently, our information environment — the collection of systems tasked with ensuring effective communication — is plagued by ideological fragmentation. Traditional publications pander to the existing beliefs of their audiences, and social media algorithms sort us into our own bespoke realities. These rifts have severed communication across ideological divides, effectively lobotomizing our global neural network. We risk losing our ability to solve increasingly complex problems.

While this represents an existential threat to our current way of life and the future trajectory of civilizational advancement, this is not doomerism. As with any problem, this is solvable, and it is solvable with current technology.

Rather than relying on systems that gatekeep institutional knowledge, fragment digital communities, and stoke ideological division, we can maximize our collective intelligence by creating a more cohesive, open information environment.

Everyone is an observer of reality. Everyone has some area of expertise, be it through their knowledge, skills, or personal experience. We need this information to be available to anyone who needs to access it, and that requires the freedom of speech.

True freedom is permissionless. You do not need permission to think. You do not need permission to contribute to public discourse.

To be effective, our information environment must be open to all, fully diverse, and truly borderless. It must be permissionless. Our ability to discover and share perspectives should be determined by individual liberty, not the authority of some institution, platform, or government.

Though permissionless systems do not entail anarchy. These systems are built on sets of rules that are transparently defined, evenly enforced, and democratically governed. Collective moderation of harmful content is critical to the health of permissionless systems.

Decentralized, open information networks are capable of developing dense connections between individuals that bridge the divides between our ideological communities, allowing us to construct a more accurate understanding of the world around us and think more clearly as a society. These permissionless networks exist. It is now just a matter of adoption.

This week, we highlight writers discussing the limitations of our ability to freely contribute to public discourse. We encourage you to explore their work and consider subscribing directly.

Media in the Media

Trend: Increasing demand for independence in media

Writers are fleeing the Substack Tax — The Verge

• As independent newsletters build substantial audiences, platforms battle to attract top writers

  • Recently highlighted with Patreon poaching Substack publications

    • Most recently, Max Read departing Substack for Patreon yesterday

      • “[Patreon] was able offer me generous income guarantees to ease the risk inherent to a transition like this.”

• “The Ankler, one of Substack’s most popular publications, left for a platform that gives it more control over its site. Others who have departed Substack within the past year voiced similar complaints and cite the platform’s increased focus on social features”

• Top newsletters are choosing to migrate to platforms that charge a flat monthly fee (like Beehiiv and Ghost) to avoid Substack’s 10% platform fee

  • Substack’s value-add is growth, providing features to expand your audience and share your work

  • Once newsletters are past their growth phase, Substack’s native growth features are no longer worth the expense

• “The pricing on Substack isn’t the only pain point for creators, as critics argue that it also locks writers and their subscribers into a closed ecosystem.”

• “Platformer creator Casey Newton, who left Substack in 2024, says that while the publication is saving money on Ghost, ‘the more important thing is that we have a home on the open web that we control, and whatever anti-creator changes Substack is forced to make in the future to live up to its valuation we won’t be affected by.’”

Theme: Yes, narrative control is a powerful tool used by governments to influence public discourse

Calif. mayor Eileen Wang admits acting as Chinese spy, resigns after shocking plea deal — The New York Post

• Eileen Wang, former mayor of Arcadia, California, admitted to working with the People’s Republic of China to spread propaganda via the news website “U.S. News Center” she operated with her then-fiancé, Yaoning “Mike” Sun

• Sun was charged in 2024 with conspiracy and acting as an illegal agent of a foreign government

  • “Sun received and executed taskings from Chinese government officials, distorted our public discourse by disseminating Chinese propaganda, and surveilled groups in the United States that China viewed as threatening.”

• Wang was ordered by the PRC to post their official articles, including denials of human rights abuses against Uyghurs in Xinjiang

Theme: The effect of AI on the media

New York Times Issues Stern Warning to Its Freelance Writers About AI Use — Futurism

• “Using [generative AI] tools to create, draft, guide, clean up, edit, improve, or rephrase your writing is strictly prohibited” states the email sent by the Times to freelance contributors

• The warning comes after the Times issued a correction last week that revealed an article published on April 15th contained a quote hallucinated by AI

• Internal staff writers at the Times follow a separate set of rules for AI use in reporting, implying they are able to use generative AI as a tool throughout the process before passing articles through review by human editors

Spotlight

Persuasion

A project of the Persuasion Institute, a nonpartisan nonprofit, Persuasion is an online magazine focused on threats to the free society and reflections on how to fight back, previously featured in our spotlight, “The Lens Through Which We See”, with the below article written by Francis Fukuyama.

Persuasion

Getting to Denmark

Will you be in London on Sunday, September 6? Yascha Mounk will be interviewing Francis Fukuyama about his life and thought to mark the publication of his memoir In the Realm of the Last Man at the Sekforde at 5pm. Find out more and book tickets here…

Read more

20 days ago · 198 likes · 9 comments · Francis Fukuyama

“Technology has also contributed to this polarization. The spread of the internet and social media has changed the nature of social interaction. Citizens who used to rely on a small number of elite-controlled media channels can now get information from anywhere in the world. The kinds of filters that used to control the quality of information have been undermined, which has led to the appearance of parallel information universes in which there is no common understanding of empirical reality.”

Expression

Written by the Foundation for Individual Rights and Expression (FIRE), a nonpartisan nonprofit that defends free speech rights for all Americans, previously featured in our spotlight, “Many Birds”.

Expression

How McCarthy scared America silent

This year, the United States celebrates its 250th anniversary. To commemorate the occasion, FIRE is proud to present the limited series “Figures of Speech,” looking at the heroes and villains of free speech in American history. We begin with Joseph McCarthy, the senator who became our censor-in-chief and gave us a new term for political oppression: McCa…

Read more

17 days ago · 21 likes · 2 comments · David Josef Volodzko

“The most damaging feature of McCarthyism was not simply government investigation, but the culture of fear that made institutions abandon free expression without being forced to do so. Newspapers who fired journalists, studios who blacklisted actors.

The irony was this was also how the Soviet Union and the East Germany Stasi operated — through voluntary compliance far more than coercion. Once dissent itself becomes professionally dangerous, formal censorship is almost unnecessary.”

Anecdotal Value

Written by Hollis Robbins, a Professor of English at the University of Utah and former dean of humanities, whose work spans American and African American literature while examining how language, institutions, and emerging technologies like AI shape academic life and public discourse.

Anecdotal Value

When I hear “viewpoint diversity,” I think Diogenes the Cynic

Diogenes of Sinope (c. 412- 323 BC) was a Greek philosopher and a principal architect of Cynicism, a school of thought that rejected conventional desires for wealth, power, and social standing in favor of a simple life of virtue in harmony with nature. He lived in a large ceramic jar in the marketplace of Athens, critiquing the society around him throug…

Read more

9 months ago · 89 likes · 21 comments · Hollis Robbins

“Today less formal but still exclusive gatherings have replaced the civic forms Diogenes mocked. Assemblies, courts, and marketplaces of public speech are now conferences, podcasts, newsletters, journolists, social media circuits, Signal chat groups. Diogenes understood that belonging limits freedom. οἱ μὲν ἄλλοι κύνες τοὺς ἐχθροὺς δάκνουσιν, ἐγὼ δὲ τοὺς φίλους, ἵνα σωθῶσιν (“Other dogs bite their enemies; I bite my friends to save them,” DL VI.60).”

Requests for Reporting

We are building out new features to allow our community to create requests for reporting, which serve as bounties to incentivize independent journalists to report on under-covered stories of interest.

• Comment or message to submit a request

• Requests for Reporting will be included here each week

• Publish an article on a requested topic through Aemula to earn a bounty

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

Trust in our news sources has been declining steadily since the 70s. To reverse this trend, rather than attempting to rebuild trust in mass media, what if we developed a form of mass media that didn’t need to be trusted? To determine if that is possible (or what that even means), we need to understand the factors that have contributed to the declining trust in media in the first place.

This is part of the broader trend of trust erosion across society. The Trust Foundation outlines this in detail in their series on the Trust Apocalypse, which we highly recommend.

Trust Apocalypse

Part I: Where are we?

“Trust is like the air we breathe—when it’s present, nobody really notices. When it’s absent—everybody notices.” – Warren Buffett…

Read more

a year ago · 7 likes · Keiron McCammon

But to focus on media, specifically news, we need to question how our current institutions rose to prominence. We wouldn’t want GK Chesterton to start lecturing us about fences.

The “news” is really just information about our world that we did not experience directly. Outside of our immediate communities, we rely on strangers to report the news.

We incorporate this information into our worldviews, and it becomes the foundation for the personal beliefs that guide the choices we make. Given the critical function of the news, we need to ensure that it is accurately reported from trusted sources. If not, our misunderstandings can have destructive consequences as they affect the emergent behavior of society.

Historically, our institutional publications served as the trusted intermediaries in this process. They vetted sources, confirmed facts, and curated impactful stories, which they packaged and sold at scale. Our money was well-spent, since we became more informed as a result, and we were able to collaborate more effectively across the globe. As these intermediaries of information built track records of reliable reporting, they earned reputations of credibility. As they gained our trust, they also gained a larger audience.

Larger audiences meant more money, providing those newsrooms with more resources to reinvest into more support for journalists and more distribution. The best journalists competed to earn a role at these institutions to be able to conduct the best reporting, signal the value of their work through association with the institution’s brand, and get their message in front of as many people as possible.

This created a positive feedback loop based on trust, improving both the quality of journalism and the access readers had to new information.

However, reliance on trust is fickle business. This vague method of credibility accounting is at the whim of public opinion. We placed our trust in these institutions based on our faith in their ability to continue reporting the news honestly. But faith is fleeting, and public opinion can be quick to turn. Trust is difficult to earn and easy to lose (as it should be).

During the period between the mid-70s and the early 2000s, we saw two major shifts that disrupted our ability to rely on trust in institutions to drive the positive feedback loop of high-quality journalism.

First, we witnessed the transition into the Information Age. Historically, news publishers and broadcast networks controlled the methods of information distribution. The infrastructure to print and deliver daily papers or broadcast radio and TV was expensive to build, operate, and license. This created a barrier to entry, restricting distribution into the hands of a few major institutions. It was difficult for the average person to get a message in front of millions of viewers.

As the internet became widely adopted, access to distribution was democratized. With the open infrastructure of the internet, as well as the decreasing costs of transmitting information, more people could participate in spreading the news. Almost anyone could get their opinions in front of a large audience, meaning that institutional publications had lost their defensible moat of distribution. Instead, they were forced to compete for attention in the crowded modern media environment. More competition meant lower margins, but quality reporting is expensive to conduct. The only way major publications could stay in business was to consolidate even further, relying on economies of scale to drive down costs.

While mass media was becoming more centralized, audiences were discovering that there was much more nuance to the narratives they had been hearing. We now had the tools to coordinate ourselves into niche communities, learning new information from forums and blogs and independent creators. Was it credible information? Who knows. But we began to circumvent the institutional narratives being spun, breaking the illusion of trust we once held.

The second shift came with the end of the Fairness Doctrine in 1987, an FCC policy that required licensed broadcasters to cover issues of public importance and provide reasonably equal airtime to contrasting points of view. While this sounds good in theory and worked to an extent, the result was an oversimplification of the national narrative. It is impossible to present all of the relevant points of view on every relevant topic within the time constraints of a single broadcast. Instead, the narrative was constrained to binary debates on a select few topics of interest. The simplicity was easy for audiences to follow, making it seem as if we had a grasp of the world around us, but the reality was that we just didn’t have access to all the information. Ignorance was bliss.

Even this narrative we are spinning now is an oversimplification. There were countless contributing factors to the decline of trust in media, and it would be impossible to unravel the full sequence of events if we wish to publish this today. However, both of these shifts played a meaningful role in fragmenting our ideological communities, leading us to question the trust we once placed in mass media. Without trust, the positive feedback loop necessary to drive the success of our institutional publications unraveled.

This is a good thing.

The trust and cohesion once held by mass media was simply an illusion due to oversimplification. We need more nuance. We need more complexity. It allows us to form a better understanding of reality. Though, it is understandably jarring to peek behind the curtain and see all of the problems we face in full resolution.

The period of collapsing trust has been a paradox. Information is cheap and abundant. We can communicate efficiently across vast distances and different languages. We have the tools to develop a more cohesive vision of reality. If our conflicts are all just a misunderstanding, then we can surely leverage this newfound knowledge to come to a collective consensus. Yet, we have only become more polarized as we voluntarily sort ourselves into smaller, increasingly disparate groups.

Fortunately, we have used this period of time to observe and learn from the flaws in our current system. Over the past decade, we have developed new technology to build internet-scale systems of coordination that do not rely on trust. Rather than placing blind faith in centralized intermediaries, trustless systems provide each individual with verifiable proof. We no longer need intermediaries because we can observe credibility directly. We have full transparency into how the system operates. Everyone plays by the rules because it is impossible to break them. They are written in code and executed in the open.

Institutional publications served as a trust layer for news curation and distribution. Trust is now obsolete. We can coordinate the same professional-grade newsroom resources in a collaborative environment, achieving infinite economies of scale without hitting the bottleneck of oversimplification due to consolidation. Where the positive feedback loop driving high-quality journalism was once built on collective trust, it is now built individual incentives.

Using these tools, we can confidently explore the full nuance of the world around us. We don’t have to outsource the development of our worldviews to the institutions once tasked with curating our information environment. We can see everything with our own eyes.

This is why we built Aemula as a trustless system. We don’t need to ask you to have faith in the credibility of the information we share. We don’t even need to prove it. You can verify it independently.

This week, we highlight writers discussing trust in media and the contributing factors to its decline. We encourage you to explore their work and consider subscribing directly.

Requests for Reporting

We are building out new features to allow our community to create requests for reporting, which serve as bounties to incentivize independent journalists to report on under-covered stories of interest.

• Comment or message to submit a request

• Requests for Reporting will be included here each week

• Publish an article on a requested topic through Aemula to earn a bounty

Media in the Media

Trend: Is the traditional media playbook of consolidation running out of steam as demand for independent media and reduced site traffic from AI search results affects the landscape?

James Murdoch in Talks to Buy Vox’s New York Magazine and Podcast Division— The Wall Street Journal

• James Murdoch, son of Rupert Murdoch who controls News Corp (parent company of The Wall Street Journal and Fox), is said to be in advanced talks with Vox Media to purchase their New York magazine and podcast division through his Lupa Systems investment company.

• This comes after Vox Media announced they would be unwinding part of their media portfolio. They had acquired New York Magazine as part of an all-stock merger transactions with New York Media in 2019.

• Vox’s decision to unwind their media portfolio is driven by decreasing site traffic and advertising revenue as they face increasing competition from the new wave of independent lifestyle publications and reduced page visits from search engine results due to new AI summaries.

• Lupa Systems (stakes in Tribeca Film Festival, Art Basel, and The Bulwark) will need to compete with Versant Media Group (spin-off of Comcast, parent to MS NOW, CNBC, USA, E!, Rotten Tomatoes, Fandango, and more) for interest in Vox’s podcast business (Kara Swisher and Scott Galloway’s “Pivot”, Brené Brown’s “The Curiosity Shop”, Adam Grant and Maria Sharapova’s “Pretty Tough”, and more).

Vice News Is Being Resuscitated With Fresh Ambitions — The Hollywood Reporter

• “equal parts creator-driven news outlet and brand partnership vehicle”

• Not hiring additional staff, but instead it seems like they will be platforming existing content creators

• Soft brand relaunch on Wednesday with a new Vice News site

Writer Spotlight

Let’s Address This

Written by Qasim Rashid, a human rights lawyer and author whose work draws on more than a decade of advocacy to provide clear, evidence-based analysis of pressing human rights issues, amplifying marginalized voices, and holding power accountable.

Let's Address This with Qasim Rashid

The Rise of Independent Media — And Why the Powerful Are Terrified of It

Yesterday we basked in appreciation of you, our readers, for making possible the successful two-year anniversary of Let’s Address This. This weekend, I am attending two events that—if you had proposed them to me even five years ago—would have been dismissed as fantasy…

Read more

a month ago · 296 likes · 16 comments · Qasim Rashid, Esq.

“The newsrooms that once covered city council meetings, school board decisions, local elections, and the stories that directly shaped people’s daily lives have been gutted, consolidated, or eliminated entirely.

Into that vacuum have stepped billionaires—purchasing media outlets not to inform the public, but to protect their interests, sanitize their reputations, and shape the information environment in ways that serve their bottom line. I have written about this in detail before. The pattern is unmistakable and the consequences are severe.

But something else has also stepped into that vacuum. Something the billionaires did not anticipate and cannot fully control.

Independent media.”

New Cartographies

Written by Nicholas Carr, a bestselling author and Pulitzer Prize finalist whose work examines the impact of technology on society, drawing on decades of writing for outlets like The Atlantic, The New York Times, and The Wall Street Journal to explore the intersection of technology, business, and culture.

New Cartographies

The Myth of the Informed Citizen

My latest book, Superbloom: How Technologies of Connection Tear Us Apart, comes out in paperback this week. To mark the occasion, here’s an essay adapted from the book…

Read more

25 days ago · 49 likes · 5 comments · Nicholas Carr

In 1919, Lippmann wrote a despairing essay in the Atlantic Monthly titled ‘The Basic Problem of Democracy.’ Democracy’s founding ideal—that of a well-informed citizenry capable of making reasoned judgments about national problems and plans—had come into being in a much simpler time, he argued, when most concerns were local and people had direct experience of them. The assumptions of America’s founders, a small, insular, largely agrarian elite, held little relevance to the bustling modern world, with its urban and industrial energies and lightning-quick communications. Society was much more complex now, and people’s sense of it came not from their own first-hand observations but through information received ‘at second, third, or fourth hand.’”

The Pugilist

Written by Alisa Valdes-Rodriguez, an investigative journalist and former staff writer for The Boston Globe and Los Angeles Times, whose work spans literary nonfiction and fiction to examine culture, politics, and everyday life with an emphasis on depth, context, and understanding.

The Pugilist with Alisa Valdes-Rodriguez

Please Stop Telling Independent Journalists They Should Send Their Stories to Legacy Media Outlets

There is no free press in America. There used to be. I’m old enough to have worked in it. But it was dismantled deliberately, over the past thirty years, by the same people and the same ideology that dismantled unions, deregulated Wall Street, and turned healthcare into a profit center…

Read more

2 months ago · 1168 likes · 141 comments · Alisa Valdes-Rodriguez

“The institutions you grew up trusting have been hollowed out and replaced by something that serves a very different set of interests, follows a very different set of incentives, and produces news that is not neutral, not comprehensive, and not, in any meaningful sense, free. This was by design. It is not that the press isn’t doing their job, as so many seem to think. It’s that the press was killed and some of you keep expecting it to breathe.”

Did you catch the change in the subtitle? “Media” spotlight instead of “writer” spotlight? We experimented with some new sections to cover the media industry more broadly while still spotlighting independent writers. Let us know what you think.

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

“We witnessed a similar dynamic with the loss of trust in traditional newsrooms as people turned to social media and private social forums as a replacement.”

— Aemula Writer Spotlight - 4.23.26

The day after the White House Correspondent’s Dinner shooting, The New York Times reported on the “information void” left in the wake of the attack. Despite eyewitness accounts coming directly from journalists in attendance, the reporting from institutional newsrooms was stuck trailing the rumors circulating on social media. Even though the reporting gap was measured in minutes, the public’s attention demanded updates and veered towards speculation. When people can’t get the information they seek from a trusted source, they turn to less credible alternatives.

Reporting from the event was quick to enter the time-intensive process of verifying facts prior to print in traditional publications.

“‘Getting out the truth and establishing facts and reliable information takes time,’ said Amanda Crawford, associate professor at the University of Connecticut who has studied media coverage of mass shootings and conspiracy theories.”

All the while, speculation spread rapidly across social media, as influential accounts sought engagement by joining the discourse without bringing any new verified information to the table. As Dr. Cliff Lampe, a professor and associate dean for academic affairs at the School of Information at University of Michigan, states in the Times article,

“Rumor moves very quickly, and then it often takes a very long time to correct those errors”.

In this case, the rumors included stories of multiple casualties, the involvement of foreign actors, and that the whole ordeal was staged. While some brief research and a bit of cross-referencing could allow readers to discern fact from fiction, social media algorithms continuously surfaced false narratives. And we should not be surprised, since these algorithms are geared to promote engagement, not truth, which leads to fabricated stories earning a prominent place in the timeline.

The fundamental structure of social networks is engineered to increase the speed and virality of posts. Our online communities exist in a state of criticality, making it easy for information to cascade across the network without passing any checks of credibility. Even our social proofs fail on social media, since follower counts and upvotes are correlated with platform success as driven by the same incentives that promote inflammatory content and clickbait headlines. Without accountability, rapidly developing stories can become consumed by erroneous speculation on social media, whether intentional or otherwise.

Yet, among all the noise, the true information was out there and publicly available. Videos and audio from inside the room, posts from credible reporters at the event, and open source intelligence from astute observers. The shooter circulated his “manifesto” immediately prior to the attack, we had reports of shots fired within minutes, confirmation of the suspect in custody and no fatalities within the hour, and the name, background, and socials of the shooter soon thereafter.

With reasonable accuracy, we could theoretically have established the timeline and motive in real time based on available information. The constraint is not the speed of communication, but our ability to coordinate the information that emerges. Fortunately, this is a solvable problem.

Note that a solution does not need to involve jumping to conclusions. In any event like this, it is imperative to report accurately and not make false accusations. If rumor is difficult to correct, it is significantly more difficult to correct errors in reporting originally framed as factual. As a result, traditional publication pipelines responsibly err on the side of caution.

A decentralized newsroom can significantly improve the efficiency of this journalistic process. With frictionless, internet-scale collaboration, stories can be pieced together at the speed of social media, under the scrutiny of many credible reviewers. To quote Linus’s Law, in reference to Linus Torvalds’s philosophy on building open source software with many contributors:

"given enough eyeballs, all bugs are shallow"

With enough trusted journalists working in the open, we can conduct accurate reporting in real time.

Importantly, this process relies on verifiable reputations, accountability, financial incentives for contributors, credentialed individuals at relevant events, a network of trusted sources, and professional tools for efficient research. Historically, this relied on the work of institutional publications, fragmented newsrooms, and coordination behind closed doors. All of which works for investigative reporting that takes place over the course of months, but in the age of the modern internet, this structure cannot keep pace with the information cascades of social networks.

Through decentralization, we can provide all of these tools to independent reporters, working at a scale and speed that cannot be achieved within a single newsroom.

The key, though, is independent reporters, since we cannot rely on speed alone. While decentralization avoids the risks of narrative control and censorship that arise within sufficiently large institutional newsrooms, it also solves the core problem that perpetuates the public’s acceptance of conspiracies.

In centralized publications, no matter the accuracy of their reporting, people can always discredit the reality being portrayed by questioning the credibility of the messenger. If there is suspicion of powerful people controlling the narrative behind the scenes, there will always be skeptics who believe they are being manipulated by some grander scheme.

For example, consider the recent survey conducted by the Manhattan Institute where 49% of respondents believe the 2024 assassination attempt against Donald Trump at the campaign rally in Butler, PA was staged. Despite the Wikipedia page for the shooting claiming that very little information was released to the public, the same page continues on to outline extensive detail about the shooter’s background, the plotting of the assassination attempt, and the events of the day, all citing credible reporting and investigation documents.

Lack of information is not the problem. Instead, people simply distrust the sources of reporting and the people conducting the investigation, providing a reasonable doubt to support the belief that the event was staged.

Now consider what Isaac Saul of Tangle states in his piece on the White House Correspondent’s Dinner shooting:

“I don’t believe — and have never believed — that these stories are all performances, which would require all reporters down to the last one to ignore the most incredible stories of our lifetimes.”

If every reporter is working independently, they need only answer to themselves and consider their own reputations. The incentives compel independent reporters to prefer uncovering genuine conspiracies rather than suppress them, since exposing a scheme to stage an assassination attempt would be a bigger story than the assassination attempt itself. There are no centralized points of weakness where powerful actors can force control of the narrative to cover a plot. There is no collusion between backroom marionettists. Diffuse networks of individuals are resilient to conspiracy because it is too difficult to perfectly conspire.

Admittedly, discrediting conspiracies and competing with the speed of the social media rumor mill are difficult problems to overcome, but Aemula is intent on experimenting with solutions to exhibit the ability of decentralized newsrooms to optimize for both speed and credibility. To start, we are now developing tools for collaborative, real-time reporting and integrating new resources for open source intelligence gathering.

Many past attempts at collaborative reporting have failed. However, we now have the technology to make this possible in theory. In practice, success relies on a large, diverse community of credible reporters. If you are writing independently, join our community of journalists to be part of making these solutions a reality.

This week, we highlight writers discussing the rumors and reporting arising from the White House Correspondent’s Dinner shooting. We encourage you to explore their work and consider subscribing directly.

Tangle

Written and curated by the editorial team of Tangle, an award-winning, reader-supported news organization offering clear, non-partisan coverage of major U.S. political stories by presenting arguments from the left, right, and center, followed by their own transparent analysis, as previously featured in our spotlight, “Patriotism”.

The White House correspondents’ dinner shooting.

“Trust in the media continues to fall, and in some cases that distrust is well earned. I built an entire media business based on my view that news consumers were living in bubbles, and that major media organizations and social media platforms were part of the problem. Journalists too often inject their own bias into stories sold as neutral reporting, advertising-first models incentivize sensational headlines, and Americans are desperate for a more diverse set of views when they open their phones or computers. I think those criticisms are valid; in many ways, I bet my career that they are.

Yet I don’t believe — and have never believed — that these stories are all performances, which would require all reporters down to the last one to ignore the most incredible stories of our lifetimes.”

Ken Klippenstein

Written by Ken Klippenstein, a former reporter for The Intercept and DC correspondent for The Nation prior to going independent to report more openly on national security, alongside Editor-in-Chief, William M. Arkin, a George Polk Award-winning and Pulitzer Prize-nominated journalist with previous work for The New York Times, The Washington Post, LA Times, and NBC News, and investigative reporter Dan Boguslaw, with work for publications such as Rolling Stone, Vice, and The Intercept. Previously featured in our spotlight, “The Lens Through Which We See”.

Ken Klippenstein

Assassin Wasn’t on FBI’s Radar, Sources Say

I’m grateful to be able to get the truth out about shootings like these and to push back against the attempts to politicize or obscure them. But I can’t keep doing this without your support - please become paid subscriber…

Read more

a month ago · 1252 likes · 45 comments · Ken Klippenstein

“Now, rumors are already swirling that Allen is some anti-Christian zealot aided by a network of left-wing compatriots. Acting Attorney General Todd Blanche says the FBI is looking into whether Allen was operating alone. But who was he, exactly? And what does he represent?

The answer, drawn from a copy of his resume I obtained and interviews with people who knew him, is unsettling in its ordinariness.”

Today’s Edition Newsletter

Written by Robert B. Hubbell, a civic commentator writing to offer hope, perspective, and steady analysis for readers committed to preserving American democracy.

“To be clear, the simplest explanation is likely the correct interpretation: 40% of Americans live in households with access to guns (Pew Research), and roughly 1 in 20 Americans own an assault rifle (The Independent). The alleged shooter texted friends and family before the attack and revealed his plans and motivation, including intended targets (The Guardian). Video shows the attack in progress, as planned by the shooter.

(…)

There is no need to conjure complicated conspiracies to explain the shooting. Guns are ubiquitous. The alleged shooter harbored grudges against Trump and administration officials. He telegraphed his plans.”

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

The above graph, based on the Financial Times’s analysis of data from the Co-operative Election Study (CES), examines the political viewpoint distribution of three different populations.

The distribution of political views for the “general population” is constructed using CES survey responses from real humans.

The “social media content” distribution is constructed from those same survey results, but selected for humans that have indicated they post political content on social media.

The “AI chatbot conversations” distribution is constructed based on synthetic survey responses from “simulated personas”, where LLMs are prompted to mimic Americans with different beliefs when answering survey questions.

Notably, the data does not actually represent the effect of AI chatbot conversations on users’ political beliefs. It does not measure whether AI use actually makes people less polarized, whether it plays a role in shifting political viewpoints, or the real-world conversational dynamics between humans and AI.

The actual takeaway from the survey data should be that social media over-represents fringe voices, since those are the only ones passionate enough about their beliefs to spend time posting about them on social media. Active participants in political discourse on social media are also affected by the feedback loop of algorithmic rewards for outrage and virality, becoming more entrenched in their belief systems.

However, we should not be surprised by the results of the synthetic survey responses generated by LLMs since there are structural explanations for why these outputs appear more moderate. Language models are trained on large amounts of data, and their task of predicting the most probable response leads them to converge towards the statistical center of the data. Additionally, reinforcement learning from human feedback during the model alignment process tends to favor balanced framing while avoiding extreme positioning.

While we don’t know the exact methodology used in creating the synthetic persona prompts, we typically see that demographic encoding fails to capture the emotional intensity held by real humans. While our current AI capabilities are powerful, they are not yet capable of accurately modeling real-world behavioral dynamics due to the fundamental structure of LLM training. As synthetic survey results gain prominence with a new wave of companies specializing in the process, we should be skeptical of the accuracy of results from simulated personas. Though, as these companies work to address these problems, the results could improve over time.

Yet, the discussion around AI’s impact on political polarization raises interesting questions. Can LLMs play a central role in our truth-seeking endeavor to build a more cohesive shared reality?

In their current state, that does not seem to be the case. The natural sycophancy of LLMs telling users what they want to hear is shown to affirm a user’s beliefs in conversation 49% more often than other humans, and LLMs validate user behavior even when harmful or clearly wrong. The result is that more users are convinced they are right after chatting with an LLM, increasing their confidence in their beliefs without increasing accuracy. The sycophantic distortion of belief formation leads to confirmation bias, polarization dynamics, and users being less empathetic to other beliefs.

If you disagree with these claims and do not believe this to be the case with current LLMs, note that we used ChatGPT to support our above claims, so if we are incorrect, our claims are proven true.

AI is adept at finding (or hallucinating) sources to support preexisting beliefs, and users are capable of steering chatbot responses to produce the desired output. Anecdotally, it is pretty easy to convince an LLM to support your point of view, even when it is wrong. As a relatively innocent example, it only took a brief conversation to get ChatGPT to support our claim that peanut butter is a soup:

We are not convinced that ChatGPT actually believes peanut butter is a soup, but it eventually came around to supporting our perspective. Condiments being classified as soups is not necessarily a harmful result, but the same process of users prompting their desired answers out of an LLM happens across the political spectrum.

If we want AI to serve as an arbiter of truth, it will need to be able to stand its ground when attempting to support its own claims. However, even as models have improved their ability to push back against user inputs, we are starting to find that users will change models to use ones more aligned with their beliefs. Different models have access to different training data and different alignment procedures, resulting in different political personalities, so users can easily switch between them depending on their desired behavior. Further fragmentation of the epistemic landscape, coupled with the confirmation bias already present in AI interactions, only risks polarizing us further.

However, this does not mean that LLMs are fundamentally incapable of supporting our truth-seeking endeavor. We just need to have a realistic understanding of the role they play.

Language models exist downstream of the data they use for training. We can attempt to scale alignment efforts and the use of reinforcement learning, but if the underlying training data exhibits the characteristics of our polarizing information silos, then we can expect LLMs to fall into similar patterns. Inherently, their training is all they know.

With Aemula, not only do we intend to improve the overall health of the underlying information environment, but our decentralized, independent model allows us to support the possibility of individualized LLM conversations for users on their own truth-seeking mission.

Utilizing AI as a centralized arbiter of truth risks users rejecting the monolithic “truth” and searching for answers from less credible sources to support their existing beliefs. We witnessed a similar dynamic with the loss of trust in traditional newsrooms as people turned to social media and private social forums as a replacement. Instead, we should not seek to pursue the construction of an untrusted oracle, but rather focus on presenting users with AI responses that are the closest representation of the truth that a user is willing to adopt into their current beliefs. We should aim for the discovery of truth over time, not attempt to immediately force truth where it won’t be accepted.

LLMs are more than capable of handling this task. They can support existing beliefs where accurate, expose users to new perspectives where necessary, and manage an individualized truth-seeking journey over the long-term. All they need is the context of where user beliefs align and how they develop over time. During this process, we should also not rely on LLMs to infer the credibility of external sources used in citation, but we should give them access to trusted, human-generated credibility scores for each independent source.

If you have played around with our new Explore page, you will understand how Aemula is able to provide this context to LLMs through our mapping of ideological communities along with credibility scores for each article.

Now, the remaining piece to the puzzle is to allow for independent journalists to license their work for use in LLM responses, while ensuring that sensitive user information remains private and owned by each user.

With that goal in mind, Aemula just released our encryption update, meaning that journalists now have more control over gating access to their content. As more readers discover news sources through chatbot responses rather than search or editorial curation, it is more important now than ever for independent journalists to have the ability to leverage that channel of distribution and be compensated accordingly.

With encryption, each individual journalist has the ability to opt in or out of LLM content licensing. If they don’t want their work accessible by LLMs, they can ensure that it is encrypted and cannot be illegally accessed. If they do want to monetize the citations of their work in chatbot responses, they now have the ability to license their work at point of citation to ensure they are compensated any time it is referenced. A feature that was previously only been available to journalists through bespoke content licensing deals with institutional newsrooms (making up an increasing portion of newsroom revenue), is now available to independent reporters.

As AI continues to impact how we all interact with our information environments, it is important that we consider it as a core part of our design for rebuilding trust and reversing polarization in media. If we intend to leverage LLMs in our truth-seeking endeavor, it is critical that the underlying information used to fuel their responses is decentralized, credible, and independently owned.

This week, we highlight writers discussing AI and its impact on our mission to discover a shared consensus reality. We encourage you to explore their work and consider subscribing to them directly.

Notes from a Strategy Scholar

Written by Nicolai J Foss, a Professor of Strategy at Copenhagen Business School whose extensive research in organizational economics and the theory of the firm examines how incentives, ownership, and individual behavior shape the performance of organizations and markets.

Notes from a Strategy Scholar

AI Isn’t a “Rationalization Machine”—It’s a Motivation Amplifier

What happens when our natural tendency to seek confirmation meets a tool that can generate instant, persuasive arguments for almost anything…

Read more

a month ago · 6 likes · 1 comment · Nicolai J Foss

“At the heart of both the social media debate and the current AI debate lies the same psychological foundation: humans are biased reasoners. We search for supportive evidence, interpret ambiguity in self-serving ways, and remember information selectively.

These tendencies long predate any particular technology.

People once relied on newspapers and television. Later, they used search engines and social media. Now, they can use AI. Each technological shift has reduced the friction involved in finding—or constructing—confirming information.”

Agents of Influence

Written by Renée DiResta, an author and Associate Research Professor at Georgetown University whose work examines online propaganda, influence operations, and the dynamics of information ecosystems, drawing on experience at the Stanford Internet Observatory and advising governments and institutions on disinformation and digital threats, previously featured in “The Aemula Bookshelf”.

Agents of Influence

Source Wars and Bespoke Realities

Control what Wikipedia considers reliable, and you control what machines—and then people—learn about the world…

Read more

6 months ago · 29 likes · 7 comments · Renee DiResta

“Back in 2011, Eli Pariser warned that search engines were quietly personalizing what users saw based on their clicks, location, and profiles. Two people could search the same term and get different results without ever knowing it. But those filter bubbles still floated atop a mostly shared basis of fact.

Now, the fight has moved from which stories appear in a list of results to which sources are digested by an answer engine.”

Cosmos Institute

Written by the Cosmos Institute, an academy dedicated to developing philosopher-builders who can guide AI toward human flourishing through research, fellowships, and interdisciplinary collaboration, with this piece written by Alex Chalmers, a staff writer focused on AI, policy, and the role of technology in advancing autonomy, decentralization, and truth-seeking.

Cosmos Institute

AI won’t fix central planning

In 1962, Victor Glushkov pitched the Soviet authorities on a nationwide cybernetics network to solve the oldest problem in socialist economics: how to allocate resources without private property and market prices. Washington was alarmed enough for the CIA to create a special taskforce…

Read more

3 months ago · 104 likes · 24 comments · Alex Chalmers

“Eventually, we get stuck in a loop. Even if we could train an AI on these interactions, what would the training data consist of? Someone has to decide to record certain things and not others. For example, we may include the transcript of the conversation, some financial metrics, and the outcome of the loan, but not the handshake or some of the pauses between words. The data is already a selective compression of the interaction, shaped by prior human decisions about what matters.

The tacit knowledge that made those framing decisions is invisible to the system trained on their outputs. No dataset encounters raw reality. It arrives pre-shaped by decisions about what to measure and what to discard. When you tell the system to look for indicators of trustworthiness, you’ve already decided what the relevant features are, which is precisely the judgment you were hoping the system would replicate.”

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

We would wager that you were well aware of these issues long before we outlined them here. If not from previous reporting on the subject, then from your own personal experience of interacting with these forms of media on a daily basis. It is more likely that you found us in your search for an answer rather than learning about the problem from us.

Yet, if you went back and read our 84 previous newsletters, all you would find is more discussion of these problems. In our effort to prove the value in what we are building, all we did was paint a fairly grim picture of the world around us. As it turns out, we are not immune to the negativity bias that plagues most of today’s media.

And what would be the point of continually discussing these problems? If you are already aware they exist, we are just continuing to reinforce your existing beliefs. The people we are trying to reach with this information are not likely to be the ones who have sought us out. While we sit here and complain about the audience capture that occurs in traditional publications, we find ourselves just as guilty.

So rather than continuing down this path of pessimism, we are choosing to shift our focus. We have discussed the problems extensively, so instead of rambling on about them, we have memorialized some highlights from our newsletter archive by problem category. This way, those who discover us in their search to learn more about the problems facing the media can easily read about them. But going forward, we want to turn our attention to the positive.

You might not expect it based on our pessimistic track record, but we are actually quite optimistic for the future of media. We wouldn’t be doing all of this if we weren’t confident that these problems can be solved.

In fact, we are already solving many of these problems. Not just in theory, but in practice. Take the above issue of audience capture as an example.

When publications only write for an audience that already agrees with them, they get strong engagement, but what is the actual impact of their work? The benefit of sharing information should be to help other people form a more accurate understanding of the world. If people already have that understanding, then the benefit is lost. Your target audience should not be one that already agrees with you, but should be comprised of the people who would benefit the most from hearing what you have to say.

Unfortunately, that strategy does not yield impressive engagement metrics on traditional platforms that aim to maximize clicks, views, and retention. The algorithmic incentive is to feed information to the people that already accept it. The rewards value audience capture, not information transfer.

With Aemula, we don’t rely on pure engagement to determine an article’s reach or earnings. Instead, an article’s earnings is based on its actual impact. An article that receives support from a small, diverse group of new readers will receive a higher payout than one that receives relatively more engagement from a tight-knit community of existing readers. If readers support an article that would have historically conflicted with their worldview, then we know that the article presents high-quality information from a unique perspective. The transfer of new information has a positive impact on the community and should be rewarded.

As readers support or disagree with articles on Aemula, we can create a map of who typically agrees with whom. We don’t know who the readers are or what the articles say, but we can begin to understand the ideological communities that exist (whatever those ideologies may be). We can then measure an article’s impact by observing its ability to bridge the gaps between these communities, and we can reward the author of that article accordingly. With more lines of communication across ideological divides, everyone is able to experience more nuanced perspectives and develop a more accurate view of the world.

This might sound abstract when reading about it, but it is much easier to experience for yourself! Our updated Explore page is now live on Aemula, where you can see how you align with the rest of the Aemula community, discover new authors and articles based on their ideological positioning, and search topics to see articles based on how closely they align with your current beliefs.

If you want to see the new features we have launched over the past few weeks but you have already used your free trial, send us a message and we will grant you a new one!

This represents just one of the many reasons why we are optimistic about the future of media. These days, online discourse often makes it feel as though we are headed in the wrong direction (we have contributed to that discourse). The prevalence of AI threatens creators, professional newsrooms are continuing to reduce headcount, and social media algorithms have divided us into our own bespoke realities.

Yet, we believe we are at the start of a new golden age for media. Taste is quickly becoming the new differentiator, distribution has been democratized, demand for independent media is at all-time highs, and the traditional industry is giving way to journalists who speak directly to their readers. A renaissance is forming, and with access to new technology, we now have the tools to set ourselves on a better trajectory in our age of information. We have discovered the problems of the past and developed the solutions necessary to address them.

The future is bright.

This week, we highlight writers who are discussing optimistic visions for the future of media as we navigate the problems of today, with this selection specifically focused on AI’s impact on writing. We encourage you to explore their work and consider subscribing directly.

Richard Hanania’s Newsletter

Eponymously written by Richard Hanania, a political commentator and president of the Center for the Study of Partisanship and Ideology, whose work draws on a background in law and international relations to examine politics, culture, and ideology through a contrarian lens.

“I think we should be more optimistic about AI raising the collective intelligence of society as a whole, which would be the opposite impact of social media. If you care about truth and are an intelligent person, there’s no way that immediate factchecking and more access to information won’t improve your work.”

Metatrends

Written by Peter H. Diamandis, MD, an entrepreneur and investor named one of Fortune’s 50 Greatest Leaders, whose work across XPRIZE, Singularity University, and multiple ventures informs his exploration of long-term technological and societal trends shaping the future.

“Technology is the single greatest force for uplifting humanity on this planet. Full stop. It democratizes access. It demonetizes what used to be scarce. It did this with information. It did this with communication. It did this with computing power. And now? It’s about to do it with intelligence itself.”

The Media Copilot

Hosted by Pete Pachal, a veteran technology journalist and media executive whose decades of experience in newsrooms inform his analysis of how generative AI is reshaping journalism, media, and the future of content creation.

The Media Copilot

The Writing Renaissance: Tony Stubblebine on Medium’s Human Future in the Age of AI

In this episode of The Media Copilot, host Pete Pachal interviews Tony Stubblebine, CEO of Medium, on the shifting economics and ethics of the internet’s creative engine…

Listen now

7 months ago · 1 like · Pete Pachal

“As AI systems learn to write, summarize, and remix content, the question is no longer whether creators should adapt, but how.

Stubblebine’s answer reframes the role of technology in creativity. He sees a path forward where writers use AI not as a shortcut, but as a catalyst for deeper thinking. Medium’s forthcoming AI-powered editor aims to help creators stay in the zone, cut through distraction, and build meaningful relationships with their readers.

This is not the death of the open web, Stubblebine argues, but its reinvention. The future belongs to platforms that can make technology feel personal again, where every story, idea, and insight still begins with something only a human can do—write.”

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

Polarization of the Press

Centralized Media’s Control of the Narrative

The Enshittification of Media

Bespoke Realities, Fragmentation, and Audience Capture

The Unsustainable Independent Model

The Decline of Trust in Media

The Complexity of the World Around Us

The Emergent Effects of Media Deterioration

As Matt frames it:

“In a famous 1980 paper ‘On the Impossibility of Informationally Efficient Markets’ I’ve visited in recent months, Sanford J. Grossman and Joseph E. Stiglitz puzzled over a theoretical contradiction in how valuable information is produced and disseminated in a marketplace.

The Grossman-Stiglitz Paradox goes like this, highly simplified: Important or valuable information can be costly to obtain or produce, so not everyone does it. However, the information gets transmitted to the rest of the marketplace when the information producers can exploit a market opening and recoup the costs of information production. But the more widely distributed information becomes in a marketplace, you’re likelier to lose your money if you keep investing in producing high-quality information. So everyone stops producing new information, and strange scenarios emerge: Everyone is either perfectly informed — or perfectly uninformed. Ultimately, Grossman and Stiglitz concluded: ‘There is a fundamental conflict between the efficiency with which markets spread information and the incentives to acquire information.’”

Taking the simplification even further, discovering new information takes effort. In the age of the internet, information spreads quickly, so once you share or act on that information, everyone else learns it without expending much effort. The more informed everyone becomes, the more effort it takes to discover new information. At a certain point, the rational decision is to just not put in the effort at all, and everyone chooses to stay uninformed.

We appreciate this connection between informationally efficient markets and the current state of our media ecosystem, since we are intent on creating an informationally efficient media environment. As such, we outlined our response to this idea in the comment section of Matt’s post, and it felt fitting to carry that discussion into this letter.

Matt’s point is an important connection to make, since observing the underlying incentives that drive information discovery and distribution can tell us a lot about why we face the issues we currently face in media. However, recent research and technological progress has provided us with solutions to this paradox.

With the Grossman-Stiglitz Paradox, the assumption is that participants are only profiting by exploiting an informational advantage (as they do in financial markets by trading assets with inaccurate prices that don’t yet reflect all information). The difficult work of acquiring an informational advantage is rewarded by being able to exploit that advantage through trading. Yet, with more informed participants, it is more difficult to gain the advantage. The potential reward is reduced, and the work is not worth it.

In both traditional and social media, the difficult work of information discovery is rewarded based on distribution (ad-supported revenue based on engagement metrics, as well as direct subscriptions). The more engagement with a piece of information, and the greater its reach, the more you can expect to earn from its consumption. As the internet and digital media democratized the distribution of information away from the hands of the publishers, the reward from distribution fell. People became more informed as they gained access to more sources of information, and that information spread more efficiently. The professional newsrooms that were doing the difficult work of high-quality, original reporting now had to compete with free-to-use social media platforms.

To reduce costs of professional journalism, newsrooms consolidated in an effort to achieve economies of scale, laying off staff and decreasing the number of voices contributing to the information environment. With less nuance in original reporting, the second-order commentary across social media also lost nuance. The health of our overall media environment deteriorated because the reward of distribution decreased. It doesn’t make economic sense to do the hard work of discovering accurate information.

We refer to this as the paradox of abundance and distrust. With greater access to cheaper information, we would expect there to be a clearer picture of consensus reality. Yet, we have fragmented into distinct ideological communities that operate within their own bespoke realities, and distrust in the information shared across these groups has only increased since the 70s.

Fortunately, we no longer have to rely on distribution as the reward for producing high-quality reporting. Instead, we can now incentivize reporting based on its quality, accuracy, and merit, so journalists are actually rewarded fairly for the work that they produce. Without relying on distribution as the driver of earnings, we don’t need to gatekeep access to information and paywall each individual source. Readers can pay a single subscription to the environment as a whole, and community-governed payout algorithms can determine how journalists are compensated for their work. All of the engagement in the ecosystem is already monetized, and it can now be distributed based on merit rather than reach.

The only way to change an equilibrium outcome is to change the underlying incentive structures that produce it. We have discovered the equilibrium for the current business models in media (bad). It is time we address the fundamental problem at hand and create a new incentive structure entirely. One where the equilibrium results in maximal informational efficiency, rather than voluntary ignorance.

This week, we highlight writers that discuss the impact of the economic incentives behind the scenes of our current media environment. We encourage you to explore their work and consider subscribing directly.

Matt Pearce

Eponymously written by Matt Pearce, President of the Media Guild of the West and former reporter for The LA Times, previously featured in our spotlight, “Influence”.

Matt Pearce

The simple economic paradox wrecking the news

The “media,” and all of us interested in it, are like travelers on a huge, busy freeway. It’s a secular communion of strangers in the same place at the same time, but who took different on-ramps and are driving to different exits, usually at different speeds. S…

Read more

2 months ago · 8 likes · 2 comments · Matt Pearce

“The ‘media,’ and all of us interested in it, are like travelers on a huge, busy freeway. It’s a secular communion of strangers in the same place at the same time, but who took different on-ramps and are driving to different exits, usually at different speeds. Some of us are in sedans, SUVs, motorcycles, Ferraris, and buses; some are just walking along the side of a busy road hoping to hitch a ride. Some are going to work, some to the hospital, some are just joyriding for fun. It’s a big road that delivers knowledge, entertainment, and, somehow, democracy. There’s no speed limit, and there are many accidents.”

Trend Mill

Written by Stephen Moore, a writer and editor whose unconventional path from hospitality and shopfitting to tech journalism informs a sharp, skeptical examination of the latest trends in business and technology, previously featured in our spotlight, “The New Internet”.

Trend Mill

I’ve Been Writing Online for 10 Years Now. Somehow I've Scraped a Living the Entire Time

Apologies, this newsletter is a little light on the usual tech-focused snark. Rest assured, that will be back shortly. This year marks the 10th year I’ve been writing online. I can’t actually remember the exact date, but it’s somewhere around now, so I’m feeling a little reflective…

Read more

2 months ago · 13 likes · 4 comments · Stephen Moore

“The lesson you might take from this is that there’s no point in trying to write good. That’s the wrong lesson. What I’m saying is that you shouldn’t expect your best work — or the work you’ve put the most effort into — to be reciprocated with the same level of returns. There’s too many other factors at stake; platforms shady working, algorithms, changing directions, and a lot of noise.”

Derek Thompson

Eponymously written by Derek Thompson, a bestselling author, longtime Atlantic writer, and host of the Plain English podcast, whose work explores abundance, technology, and the shifting social fabric of modern life with a mix of curiosity, skepticism, and clarity, previously featured in our spotlight, “Keep it Simple”.

Derek Thompson

How ‘Zombie Flow’ Took Over Culture

The late psychologist Mihaly Csikszentmihalyi was troubled by a paradox of progress. People alive today have more sophisticated machines, medicines, and systems for organizing the world. So why haven’t these advances made us happier? “The gods of the Greeks were like helpless children compared to humankind today and th…

Read more

2 months ago · 284 likes · 11 comments · Derek Thompson

“If you’re so smart, why aren’t you happier?”

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

Granted, presenting TBPN in this light does not do them justice. The show, which streams live on weekdays from 2-5pm ET, targets a niche audience of tech elites, capturing a significant mindshare in Silicon Valley. With their focus on cultivating a high-impact following by attracting an impressive slate of guests, they have been able to establish valuable partnerships with tech startups to advertise to the tech leaders who control the levers of the startup ecosystem. These partnerships are reportedly projected to draw in $30m in revenue for 2026, and with only 11 employees, the company’s reporting of profitability is quite believable.

With this understanding, OpenAI’s strategy behind the acquisition starts to become clear. The stated reasoning is that OpenAI wants to “create a space for a real, constructive conversation about the changes AI creates”, and “rather than trying to recreate” that space themselves, they are bringing on TBPN to do that for them. TBPN will report to Chris Lehane, OpenAI’s Chief Global Affairs Officer, and contribute to the company’s comms and marketing strategies. Despite the new responsibilities of the TBPN team, the classic TBPN show will continue, with OpenAI emphasizing that TBPN will retain “editorial independence”.

To lead off by playing our own devil’s advocate, the acquisition does significantly expand TBPN’s access to resources, and their success is well-deserved. They earned their influence through a long track-record of credible, entertaining reporting while providing thoughtful commentary with integrity. People familiar with the TBPN founders have only nice things to say about John Coogan and Jordi Hays.

Our concern, however, is with the pledge of “editorial independence”, since unfortunately, there is no such thing.

The only time you hear of editorial independence is from the mouths of those who have their independence questioned. When there is a clear conflict of interest in ownership or funding, newsrooms and media companies pledge the independence of their reporting to ensure that readers can trust that they are not influenced by the incentives that drive their underlying business. However, when there is no structural separation of these incentives, independence is impossible. Whoever controls the money has the ability to control the content.

For example, consider Jeff Bezos’s acquisition of The Washington Post in 2013. The high-profile acquisition raised questions about the billionaire’s potential influence over the newsroom’s reporting. How could the editorial board ensure their neutrality when covering the other businesses in Bezos’s portfolio? With the Post losing money, their survival was reliant on a steady flow of funds from their new owner. Don’t bite the hand that feeds.

Despite assurances of their editorial independence, the curtain was soon pulled back. Most notably with Bezos spiking the Post’s endorsement of Kamala Harris ahead of the 2024 US presidential election. Note, in this example, it was the timing of spiking the endorsement that was suspect. Avoiding political endorsements to maintain an unbiased newsroom is sound logic if implemented ahead of the election cycle and before a piece on the endorsement is already awaiting publication.

However, in less public examples, the Post failed to disclose their conflict of interest when reporting on matters where Bezos held a financial interest. In three instances over a span of two weeks, the Post aligned their reporting to the benefit of Bezos’s other ventures without noting the conflict of interest to their readers.

Of course, this was the point all along. Why else would Bezos pay $250m for control of the paper when that valuation could not be supported by the paper’s financials or comparable transactions? Consider that the Post had sold its ownership of Newsweek magazine three years prior for a single dollar. It is difficult to imagine that a billionaire, who spends time launching rockets to space and vacationing on superyachts, would overpay for the opportunity to lose money managing a national paper just for the love of the game. He certainly didn’t make the acquisition as an attempt to preserve a newsroom of national importance. Instead, the acquisition can be justified because control of the public narrative is an incredibly valuable tool.

This scenario with The Washington Post is a stark example, but the impossibility of editorial independence is even exposed under morally sound circumstances. Last week, The Liberal Patriot, a non-profit newsletter with 50,000 subscribers, announced they would be closing their doors. As a non-profit, their funding was largely reliant on donations, specifically from donors associated with the Democratic Party.

While they used their platform to support contributors discussing a “mix of economic nationalism and cultural moderation”, they inevitably lost support from their donor base. Rather than sacrificing the integrity of their reporting to protect their funding, they stayed true to their editorial mission. Yet, in their own words, “upsetting the partisan applecart on multiple issues is not a particularly good fundraising or business model”. For added context from their executive editor:

It is an unfortunate reality that you can never escape the incentives that underlie your business model. When we see claims of editorial independence, it should draw our attention to the conflicts of interest that exist. Once we understand the incentives, we can better identify potential conflicts in reporting.

With OpenAI, where might their incentives lie? Perhaps, controlling the narrative around the AI industry via a trusted source of direct content distribution among Silicon Valley elites would be an incredibly valuable tool to wield as they gear up for a potential IPO in the fourth quarter of 2026. It could definitely benefit the company after a wave of bad press. Of course, this is only speculation, but it certainly makes more logical sense than acquiring a talk show to help them scale in exchange for them to give OpenAI’s multi-billion-dollar marketing team tips on comms strategies.

While we pick on OpenAI and TBPN due to the recency of their announcement, deals like these are prevalent throughout the media industry. Influence over public perception is powerful, which is why it is so common for individuals and organizations to spend significant sums of money to exploit control over media outlets. With rapid news cycles and saturated public attention, most acquirers feel they can wave their hand, promise “editorial independence”, and hope that readers lose sight of the conflicts of interest behind the scenes.

However, readers lose trust in news sources when promises of independence are violated. Trust in traditional publications is at an all-time low. Reputations are difficult to build, but quick to be lost. Even if a professional newsroom conducts credible reporting the vast majority of the time, they can lose the trust of their audience with a single piece of conflicted reporting. As readers lose faith in the journalistic process, they turn to alternative sources, largely on social media, which can stray significantly farther from the integrity of a professional newsroom.

The only way to rebuild trust in media is to not rely on faith, which is why Aemula relies on proof. We are a verifiably neutral platform, utilizing open source algorithms and community governance to remove the possibility of narrative control. In a fully transparent environment built on immutable reputations, accountability, and independent ownership, readers can see for themselves that there is no outside influence over the news they read. We don’t have to promise editorial independence because there is no editorial board, and independence is built into the foundation of the platform itself.

This week, we highlight writers discussing editorial independence and the alignment of incentives in media. We encourage you to explore their work and consider subscribing directly.

TBPN

Written by Brandon Gorrell as part of the team at TBPN.

TBPN

OpenAI acquires TBPN

Today, we’re announcing that OpenAI has acquired TBPN…

Read more

2 months ago · 53 likes · 12 comments · Brandon Gorrell

“TBPN will continue to operate as an independent platform for companies across industries to share news, launch products, and interact directly with one of the most engaged audiences in tech.

The TBPN team will continue to operate the show and make all editorial decisions while working closely with OpenAI to expand its reach and impact.”

The Liberal Patriot

Now archived, but formerly written by the team at The Liberal Patriot along with frequent contributors.

The Liberal Patriot

No Learning Please, We’re Democrats!

Last July, I wrote a piece asking, in the wake of Democrats’ catastrophic defeat in the 2024 election and the obvious need for serious party-wide change, “Is Our Democrats Learning?” At the time, I saw little evidence that Democratic learning was, in fact, taking place…

Read more

2 months ago · 1073 likes · Ruy Teixeira

“Funding for our modest enterprise, always precarious, has now completely dried up. Our view that the party has neither solved its problems nor is even very close to doing so has tanked our appeal among partisan Democratic donors, even reform-oriented ones, who now tend to regard us with suspicion. A little heterodoxy is fine but there’s a limit! Hence: no money.

So we are forced to close our doors. The Liberal Patriot, alas, will be no more.”

The Free Democrat

Written by Erik Carter a lawyer, financial planner, and political activist who has provided political and financial commentary on numerous media outlets such as MTV, Time Magazine, Forbes, Bloomberg, and Fox Business.

The Free Democrat

The Warning in The Liberal Patriot’s Final Message

In my last post, I wrote about several conferences: climate advocates gathering in the Northeast, the abundance movement meeting at WelcomeFest, civic bridge-builders at Braver Angels, and liberal reformers at the New Liberalism conference. Those events represent something hopeful. They are spaces where people are still willing to debate ideas openly. B…

Read more

2 months ago · 2 likes · 3 comments · Erik Carter

“Ironically, the very forces that made sustaining a heterodox publication difficult are also creating a demand for new spaces of debate. Policy reformers are discussing housing and abundance. Climate advocates are pushing bipartisan solutions. Civic organizations are rebuilding cross-partisan dialogue.

These spaces suggest something important: the appetite for better arguments has not disappeared. What remains uncertain is whether the Democratic Party will embrace that culture of learning or continue to resist it. The editors of The Liberal Patriot delivered their warning before the next political cycle unfolds. That means the warning still has time to matter.”

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

“As the country became more polarised, the national media followed the money by serving partisan audiences the versions of reality they preferred. This relationship proved self-reinforcing. As Americans became freer to choose among alternative versions of reality, their polarisation intensified.”

— James Bennet, former opinion editor at The New York Times

It is no secret that we use this newsletter to discuss themes of censorship, narrative control, and ideological polarization in media to highlight the benefits of Aemula’s decentralized model. However, when we share examples of these issues in our current media ecosystem, as we did last week, we are typically referring to explicit cases of manipulation. In these instances, when platforms ban accounts and governments pressure publications to suppress specific stories, it is easy to understand the effects these actions have on the deterioration of our information environment.

However, these are also the instances that are easiest to defend against. When we can pinpoint the levers of control and observe the stories being suppressed, we can more easily determine the parts of the narrative we need to explore in order to construct a more accurate understanding of the world.

As Thomas Karat puts it in the article spotlighted below:

“Soviet censorship — the kind with gulags and disappeared journalists — produced a distinct pathology: everyone knew the official narrative was false, because the enforcement was visible enough to reveal the gap between what was said and what was real. The propaganda was so heavy-handed it became its own kind of transparency.

The subtler variants are far more dangerous, precisely because they leave no obvious gap.”

We often feel that fears of censorship, propaganda, and narrative control are overblown, since we can only recall a handful of examples, and in these examples, we discovered the suppressed stories regardless of the manipulation. The suppression itself drew attention to what information we were missing.

However, the more prevalent, dangerous manipulation is that which you do not recognize. We all lead busy lives, and news consumption is a rather passive activity. Cable news runs in the background on muted TVs, and articles merely graze our attention while we scroll social media or digital news sites. As headlines stream past us, we make split-second extrapolations of the content within a story.

Newsrooms understand this behavior, and they study it religiously since it drives their core business. The majority of Americans discover news through social media, and a study of 35 million Facebook users found that 75% of article shares were made without the poster clicking past the headline. An American Press Institute survey of journalists found that 72% of journalists believe most Americans passively come across news stories, and only 10% of journalists believe users actually read into the details of the story. 91% of journalists think that people only skim the headlines.

The media has optimized for passive headline skimming and social media distribution, which opens the door for the invisible manipulation of the narrative. Newsrooms can leverage headlines to frame events in line with their preferred narratives without needing to resort to spreading misinformation or suppressing specific stories. Public discourse rides the tide of headlines as they circulate between partisan online communities, while politicians leverage the progression of our collective attention to further their own agendas.

While we rarely explore the full nuance behind the headlines we read, we are not at fault. There is simply not enough time in the day to accurately assess the barrage of information in which we find ourselves. The only way we can survive in the modern media environment is to rely on vague understandings of the general narrative, and this reliance is being exploited.

To understand this, let’s review recent reporting from The New York Times. Note that we use The New York Times as our example since it is the largest paper by circulation, the largest newsroom by headcount, one of four papers of record in the US, and rated as reliable without a meaningful bias. They are one of the best examples of professional journalism in America, and they capture a significant percentage of the original reporting mindshare.

The rated bias of the Times skews left, so the below example of headline manipulation naturally skews left. However, publications that skew right are just as guilty of the same practices. The below example is meant to serve as evidence of the existence of headline manipulation, not commentary on the specific events discussed in the example. We share opposing views to these events in our spotlighted articles.

Headline: “T.S.A. Tipped Off ICE Agents Before Arrests at San Francisco Airport”

Context: This headline, published on Wednesday, 3/25, was part of an editorially curated section on the Times front page about ICE deployment to airports to support the understaffed TSA during the Department of Homeland Security funding dispute.

Nestled between headlines “Republicans Offer to Fund Homeland Security Without ICE Enforcement” and “What It’s Like to Stand in an Airport Security Line for Hours”, any rational reader skimming past this section would reasonably construct the narrative that Senate Republicans were blocking DHS funding by coupling it with disputed ICE funding, the unpaid TSA was facing challenges due to the block on DHS funding, ICE was deployed to airports to support the TSA, and ICE used this deployment to make arrests in airport security.

The Actual Story: However, when you read the story, you discover that the ICE arrest at SFO was conducted during the course of their normal operations on Sunday, 3/22, prior to ICE deployment to airports to assist the TSA on Monday, 3/23.

Additionally, SFO uses a private security company for screening, not federal TSA employees, and no ICE agents were deployed to SFO as part of the effort to support the TSA during the funding freeze.

The TSA provided information to ICE on the preceding Friday after flagging that a passenger on an upcoming flight had an outstanding deportation order. The arrest then took place in the terminal, not in the security line.

At the bottom of the article, which takes about 5 minutes to read, the Times notes:

“The encounter was not part of the Trump administration’s effort to use ICE to staff airport security lines while T.S.A. agents are going unpaid because of the government shutdown.”

Nothing stated in the article is untrue, and it is a solid piece of reporting that explores the scope of data sharing practices between the TSA and law enforcement, as well as the procedures used by ICE when conducting arrests.

Yet, the casual observer scrolling headlines on the Times front page would not have gained that understanding. Instead, they would reasonably construct the narrative curated by the editors of the Times before continuing scrolling.

And the casual observer can’t be blamed, since we simply don’t have time to devote 5 minutes to reading every article like this that we come across. When we are presented with headlines in a curated section, we feel that we easily understand the scope of the story given its framing. We don’t need to devote our time to getting the details, since we can already tell that the fears around ICE deployment to airports came true and arrests were made in the TSA line. Yet, this was not the case, and we are left with an incorrect understanding of events despite factual reporting from the Times.

There is no shadow power controlling this behavior. This type of behavior in media is not reliant on the collaboration of state actors, billionaire owners, and empowered editorial boards conspiring to manipulate the public. These are merely symptoms of the incentives in which our newsrooms operate. They need to pander to the existing beliefs of their audience in order to retain readers. Just look at the fall in ratings at CBS as Bari Weiss attempts to rework the narrative of the newsroom. Modern distribution and stretched attention spans require optimized headlines that bait engagement, while the expectation is that the nuance of the story will be glossed over.

Rather than play these games, Aemula addresses these problems at the root level by adjusting the underlying incentives. Reach and monetization on Aemula are based on the diversity, reputation, and support of the readers who actually read a post. Clickbait headlines, pandering to a partisan community, or farming engagement through inflammatory rhetoric do not earn you success on Aemula.

Instead, reporters are incentivized to share an accurate interpretation of the facts. Readers can then explore these perspectives while trusting that no conspiracy to control the narrative exists, since Aemula is verifiably neutral, open source, and governed by the community. Independent journalists own their content and their reputations, ensuring accountability in the event that they attempt to breach the trust of subscribers. Only by adjusting the incentives can we protect our information environments from these types of manipulation that plague the centralized media.

This week, we highlight writers who are discussing the soft narrative control deployed by traditional publications. We encourage you to explore their work and consider subscribing directly.

When the New York Times lost its way

Written by James Bennet and published in The Economist following his resignation as editorial page editor at The New York Times.

Read Article

“For now, to assert that the Times plays by the same rules it always has is to commit a hypocrisy that is transparent to conservatives, dangerous to liberals and bad for the country as a whole. It makes the Times too easy for conservatives to dismiss and too easy for progressives to believe. The reality is that the Times is becoming the publication through which America’s progressive elite talks to itself about an America that does not really exist.”

Thomas Karat’s Substack

Eponymously written by Thomas Karat, a behavioral analyst and former corporate executive whose research on deception and negotiation informs his interviews and analysis of global conflicts, offering a critical lens on power, narrative, and the forces shaping international affairs.

Read on Aemula

Thomas Karat’s Substack

The Censorship You Don’t Notice Is The Only Kind That Works

There’s a scene in Robert Cialdini’s Influence where he describes a waiter named Vincent — the highest-earning server in an expensive restaurant, by a significant margin. Vincent’s secret wasn’t charm or efficiency. It was that he occasionally talked customers…

Listen now

2 months ago · 64 likes · 1 comment · Thomas Karat

“The most effective historical examples of information control have almost never been the overtly brutal ones. Soviet censorship — the kind with gulags and disappeared journalists — produced a distinct pathology: everyone knew the official narrative was false, because the enforcement was visible enough to reveal the gap between what was said and what was real. The propaganda was so heavy-handed it became its own kind of transparency.

The subtler variants are far more dangerous, precisely because they leave no obvious gap. When journalists self-censor, the coverage that gets produced still looks like journalism. It has bylines and sources and the visual grammar of independent reporting. The absence — the story not pursued, the angle not taken, the question not asked — is structurally invisible. Nobody can point to it. Nobody can protest it. The reader has no way of knowing what they’re not reading.”

Brian Tyler Cohen

Eponymously written by Brian Tyler Cohen, a political commentator, YouTuber, and podcast host known for his clear, accessible breakdowns of US politics and interviews with major political figures, aimed at helping a broad audience better understand current events.

Brian Tyler Cohen

What They’re Hoping You Don’t Notice in Congress This Week: TSA Standoff, SAVE Act, and More

On Tuesday, I took a look at the TSA shutdown. Now, the whole debacle is ballooning faster than airport security wait times…

Read more

2 months ago · 972 likes · 46 comments · Brian Tyler Cohen

“It looks like chaos, but there’s a strategy. It’s meant to overwhelm us so that when we’re not looking, power shifts, and the structures safeguarding democracy get a little bit weaker.”

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

In Part II, we discussed how the mechanics of that structural misalignment creates chaos in media.

Now that we have identified the root cause of the problem and its consequences, we can define the solution. Last week, we established that avoiding both chaos and centralized control requires a decentralized model. What we did not adequately explain is why decentralized networks are actually more stable in practice.

In our current media system, you subscribe to individual entities, whether they are traditional publications or independent reporters. You follow people on social media, with the platforms themselves operating in their own siloed environments. As information flows from the original source of reporting to the end consumer, it filters through these points of control as it migrates across publications and platforms, unevenly amplified and reframed at each step along the way. The result is the power-law distribution of influence that concentrates control over the flow of information into the hands of a select few.

Alternatively, a decentralized network connects every consumer directly to the source of original reporting. There is no fragmentation since every individual, platform, and publication interacts with the network as a whole, rather than via influential points of control. Readers are able to freely explore without running into paywalls or relying on intermediaries.

The resulting network is dense and evenly distributed, meaning that no individual participant holds outsized influence. This structure naturally avoids the pitfalls of our current media systems, since no single point of control exists to meaningfully distort the flow of information. Whereas networks formed through preferential attachment are fragile, easily manipulated, and collapse into cascades of misinformation and negativity bias, decentralized dense networks create stability through redundancy, allowing competing sources of information to exist in parallel, with participants freely shifting between them.

However, up to this point, we have discussed these ideas in the abstract, which begs the question of what a decentralized solution looks like in practice.

The Aemula Protocol

Aemula is a publishing protocol that runs on decentralized infrastructure. To understand what this means, it is worth dissecting the component parts of that statement, starting with decentralized infrastructure. To solidify the concept of decentralization, it is helpful to consider centralized and federated alternatives.

Centralized and Federated Systems

Take Substack for example. When you log in to your Substack account to access the site from your browser or the Substack app, you are requesting your account information from servers that Substack operates. You provide your password, and they give you access to the information for your account. Your subscriptions, the accounts you follow, your contact information, and your preferences all exist only on computers that Substack controls and operates.

Similarly, when you publish an article on Substack, the content of that article is stored and served through Substack’s servers. They grant creators the ability to export their email subscriber lists (an amazing feature), but the existence of your work and the management of your data is completely controlled by Substack. It is unlikely, but they have the ability to delete your work, deplatform or shadow ban your account, remove your followers or notes, or even delete your subscriber list before you have a chance to export it. If Substack ceases to exist, so does your work.

This is the same architecture of most major social platforms, such as Facebook, X, and TikTok, as well as the digital side of institutional news publications like The New York Times and The Wall Street Journal. These platforms are simply programs that run on corporate servers that hold full control over every user interaction and every piece of media. You are at the mercy of the rules they enact at their sole discretion.

More recently, federated platforms like Mastodon and Bluesky have emerged with a different architecture. They attempt to distribute control by allowing multiple, independently operated servers to share and exchange data in a standardized schema. However, each server still operates under its own authority. Users have the freedom to migrate between servers, but each individual server runs the risk of falling into the same traps as a centralized platform.

As a result, federated networks either remain fragmented, with users primarily interacting with small subsets of different servers in the federation, or they concentrate around a small number of dominant servers. Users might have the ability to migrate, but they remain dependent on individual server operators that determine the rules of the ecosystem.

If the goal is a cohesive, censorship-resistant information network, neither of the above models is sufficient. Potential points of control must be removed entirely, rather than simply distributed.

Decentralized Systems

The Aemula Protocol is built on Ethereum and the InterPlanetary File System (IPFS). These are peer-to-peer networks that utilize a consensus source of truth without relying on centralized ownership.

Ethereum, otherwise known as “the world computer”, acts as a shared execution layer. Similar to federated networks, anyone can establish their own server as part of the Ethereum network, called a “node”. However, rather than operating this server node in a vacuum, each node within Ethereum must coordinate and collectively agree on the state of the Ethereum network as a whole. This means that everyone is operating in one cohesive ecosystem, without any single entity in control. It is an open, permissionless, global environment that executes online programs.

Centralized platforms, such as Substack and Facebook, are simply programs that run on servers they control. With Ethereum, programs run on the world computer. Each individual user controls their account and data directly. You can export and migrate your account anywhere within the Ethereum ecosystem at your sole discretion, without relying on an intermediary, and it remains the same. No one can gate your access.

Additionally, once a program is running on Ethereum, it is open source and can only be modified based on the rules enshrined in the program itself. This level of transparency builds trust, allowing everyone to participate with full knowledge of the rules of the game.

IPFS is a peer-to-peer content delivery system. Rather than storing content on a single server, files are distributed across a network of nodes. Similar to Ethereum, anyone can establish and operate their own node in the network. Content, such as articles, podcasts, and videos, is distributed across the network without needing to rely on any single server for availability. Whereas Substack can remove a user’s post at their discretion, if any single IPFS node attempted to remove a post, the content could exist elsewhere in the network and continue to be served without issue. Availability is not tied to a single provider.

Together, using Ethereum as a decentralized execution layer and IPFS as decentralized storage and content delivery, these systems autonomously operate The Aemula Protocol without centralized control. The rules are public, the infrastructure is open, and no single entity can gate access or remove content.

Why This Matters in Practice

In the vast majority of cases, centralized systems feel sufficient for our use cases. Most people casually read and explore these platforms, and they are happy with the ease at which they can discover and interact with content.

However, when it comes to news media, reporters are building their livelihoods on these platforms. Their income is tied to their ability to reach their audience and monetize their work. In the normal course of business, there is no issue with their reliance on centralized platforms.

Yet, when things start to break down, and the freedom of speech matters most, pressures begin to build on these systems. From information platforms to payment processors, corporations begin to restrain their users. It is not a question of whether individual companies are acting in good faith, but is instead a question of whether powerful external actors can compel companies to act on their behalf.

Freedom of speech is vital to the basic human endeavor of seeking the truth. The churning of ideas in public discourse, as well as our ability to question the status quo, is critical in developing an accurate understanding of the world around us, while progressively pushing forward the boundaries of our collective intelligence. It is our right, and one that must be upheld now that our public squares exist in the digital realm.

It may seem paranoid to worry about censorship pressure and narrative control when they rarely appear as issues to the passive observer, but they are more prevalent than most would expect.

To pull examples of incidents with major platforms and payment processors over just the last few years:

• TikTok — TikTok has repeatedly faced allegations claiming that they suppress certain political content, with the US later passing the divest-or-ban law, ultimately leading to TikTok voluntarily halting service in the US for a day, as well as the subsequent creation of a majority American-owned US joint venture (2024 - present), with the Trump administration set to receive roughly $10b in fees tied to the US TikTok deal

• PayPal — PayPal added the ability to issue “misinformation fines” and reject monetization for users at their sole discretion before reversing course and stating the update was made in error (2022)

• Meta — Major social platforms were pressured by the federal government to suppress discussions on the efficacy of COVID-19 policy (2021), Meta later stated they regretted the moderation decisions they made under this pressure

• Discord — Discord banned servers discussing the GameStop short squeeze event while Robinhood restricted investors’ ability to purchase specific securities (2021), with Discord quickly beginning to work with the community to create a replacement server after public backlash

• Patreon — Patreon adopted policies allowing them to remove and demonetize users based on “off-platform behavior”, expanding the scope of their moderation beyond user activity on their platform (2021)

• Twitter — Major social platforms were pressured by government officials to suppress the Hunter Biden laptop story as reported by the New York Post ahead of the 2020 election (2020), with former Twitter executives later saying that blocking the story was a mistake

• Apple — Apple removed apps related to the Hong Kong protests after criticism from Chinese state media (2019)

• YouTube — YouTube’s automated moderation systems demonetized and removed war reporting conducted by journalists and NGOs (Syrian Archive, Bellingcat) threatening digital evidence relevant to Syria and other conflicts (2017 - present)

Centralized control leaves platforms exposed to the influential power of external actors as they exert their will over user activity. Controlling public discourse is an extremely valuable tool, which has been repeatedly exploited throughout history.

The only protection against these external forces is to remove the pressure points entirely. When a platform is decentralized and democratically governed by the community, these pressure points no longer exist. The above cases could not have taken place on a decentralized platform.

Aemula as a Publishing Protocol

The Aemula Protocol defines a set of rules for publishing, accessing, and interacting with content. This ruleset is then executed on Ethereum and facilitated using IPFS, removing any points of centralized control.

When a creator publishes a piece of media, the protocol records ownership and broadcasts the event to the network. The content itself is stored on IPFS and controlled by the creator.

Consumers access content through a subscription to the protocol. A fixed payment to the protocol (currently $10 in USDC) automatically grants you a month of access to decrypt and consume any content within the network. This removes the need for fragmented paywalls, while also allowing for a permissionless payment process, meaning no single entity can block your access.

Subscribers can also interact with published media via the protocol’s defined mechanisms. Currently, users can passively interact, support or disagree with a piece of content, or report content for moderator review. Anyone can become a moderator to examine and vote to pause circulation of reported content, sending it back to the author for correction or removal.

These interactions drive the reputations of all users, the credibility of each piece of media, and the earnings that creators receive from the subscription pool. All of which are determined by open-source algorithms. The entirety of this process can be recreated from scratch using publicly available information by anyone who wishes to verify, replicate, or operate the protocol.

The rules defined by the protocol are democratically governed by the community, with voting rights becoming progressively distributed with scale. Any modifications of the ruleset requires collective agreement rather than unilateral control, as defined by the voting processes publicly outlined in the protocol itself.

The Aemula Platform

If you have used Aemula, you have interacted with aemula.com, which is distinct from The Aemula Protocol. The Aemula site is the gateway that we (as a company) offer to users so they can easily interact with The Aemula Protocol.

We run our own servers to execute open-source recommendation algorithms and efficiently cache and serve content. We help users set up Ethereum accounts, sponsor their gas fees, and provide easy onramps to fund their subscriptions. These are all quality-of-life services, provided with the intention to make Aemula as convenient as any other digital media platform.

Yet, anyone can set up their own gateway to the protocol. Everything on aemula.com can be independently recreated by utilizing public information provided by the protocol. Anyone who subscribes to Aemula can access the protocol through any gateway, even one of their own creation. Journalists can even store and serve their own content to ensure it can never be removed by someone else (if you are a journalist interested in doing this, we will help you set it up).

If you don’t like the Aemula platform services, you can create your own platform based on The Aemula Protocol. Anyone can. The right to exit or migrate platforms is fundamental to decentralized systems, and absolutely critical to building trust and providing censorship resistance in news media. As a user with ownership of your own account, you are free to choose which platform suits you best.

Importantly, additional platforms don’t fragment the network, but expand the cohesive ecosystem that is enabled by the protocol running on decentralized infrastructure. The benefits of decentralization increase with the number of platforms and participants. Whereas centralized and federated networks run into problems of control and fragmentation as they scale, Aemula can scale infinitely without issue.

Since Aemula does not unilaterally control content or gate access, we can never be externally pressured to encroach on someone’s freedom of speech. We simply can’t, since the mechanisms don’t exist. That right is held by the individual. The protocol empowers these individuals to coordinate independently, share freely, and conduct the journalist process in a cohesive, transparent ecosystem.

In fact, if you want to set up your own platform built on The Aemula Protocol, all the information you need is right here.

Note that The Aemula Protocol and platform are still in beta testing. Aemula currently retains ownership of governance rights to have the ability to implement fixes and modifications to the protocol. Prior to launch, Aemula will define decentralized mechanisms that programmatically distribute governance rights to the community based on well-defined benchmarks of scale and community coordination.

This way, decentralization of control will be guaranteed as the community grows and is capable of self-governance by a sufficiently diverse user base. The strategy is based on the safe distribution of control to avoid governance collapse prior to reaching stability, while ensuring the path towards decentralization cannot be revised by Aemula once incentives shift.

The first 5,000 paid subscribers at time of launch will receive voting rights in the initial distribution of governance. Subscribe today to reserve your spot.

This week, we spotlight writers discussing the risks of censorship and the pressures facing centralized platforms. We encourage you to explore their work and consider subscribing directly.

Expression

Written by the Foundation for Individual Rights and Expression (FIRE), a nonpartisan nonprofit that defends free speech rights for all Americans, previously featured in our spotlight, “Many Birds”.

Expression

Censoring Iran war news sets a dangerous precedent

This essay was originally published by UnHerd on March 16, 2026…

Read more

2 months ago · 14 likes · 2 comments · Aaron Terr

“Throughout his tenure, Carr has conflated the ‘public interest’ with the president’s interest. But that’s not the law. The law is clear that the First Amendment bars the government from dictating news coverage or punishing outlets for publishing what the president claims is “false.” The American people are entitled to uncensored news about what their government and military are doing. No matter who occupies the White House, war does not justify the government erecting barriers between the people and the press. To the contrary, the life-and-death stakes of war make public scrutiny and accountability more important than ever.”

Noahpinion

Written by Noah Smith, an economics PhD and former assistant professor of finance, now writing independently about economics after leaving a role as an economics columnist for Bloomberg Opinion in 2021, previously featured in our spotlight, “Lost Nuance”.

Noahpinion

Save us, Digital Cronkite!

I’ve been writing some pessimistic things about AI recently, so I thought I should try to balance those out with some optimistic takes. One way I think AI could really help our society is by injecting reasonableness and moderation into our public discourse…

Read more

2 months ago · 262 likes · 41 comments · Noah Smith

“But if you think social media has pushed society too far in the direction of anarchy, then you’ll welcome a bit of a push back in the direction of consensus. A country can’t get anything done if everyone is always at each other’s throats. Nor did fragmentation and polarization ‘democratize’ our information space — they marginalized the silent majority of moderate normies, and handed control of our thoughts to some of the worst extremists in our society. In a way, by giving voice to the center of the distribution, AI may be a more truly democratizing force in our discourse than the internet itself ever was.”

The Preamble

Written by Sharon McMahon, a #1 New York Times bestselling author and educator known as “America’s Government Teacher,” whose work explains politics and history with clarity and calm, helping readers stay informed, engaged, and hopeful without the noise of modern news cycles.

The Preamble

How the FCC is Trying to Control the News You See and Hear

On August 18, 1933 — seven months after Adolf Hitler became Chancellor of Germany — Nazi Minister for Public Enlightenment and Propaganda Joseph Goebbels stood before a crowd of journalists and radio officials in Berlin and gave a speech he called “Radio as the Eighth Great Power…

Read more

2 months ago · 499 likes · 17 comments · Sharon McMahon

“The people who built the FCC understood what it looks like when a government decides that the press exists to serve power rather than to check it. They had watched it happen in Germany, in Italy, in country after country where the machinery of communication was slowly turned against the people it was meant to inform. They wrote the prohibition of censorship into law because they knew that without it, “the public interest” would become whatever the man in charge decided it meant that week.

Brendan Carr is the man in charge this week. And what he has decided is that it means broadcasters who cover a war in ways the president dislikes should be warned — publicly, with the weight of federal licensing authority behind the warning — that there are consequences for telling you what is happening.”

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

Last week, we explored how the current, centralized model of original news reporting has deteriorated our trust in media. As we attempt to create “new” media systems to replace institutional publications, we risk reinventing the same problems. Scaling the new systems begins a cyclical migration. The rise of monopolistic institutions, subverted by revolutionary new platforms, gives rise to the next generation of monopolistic institutions.

These new media systems provide attractive alternatives when they exist as walled gardens. Niche corners of the internet where cultures can percolate uncontested, similar to how most of today’s major platforms and publications earned their beginnings. But they can’t grow to societal relevance without giving up the traits that made them unique, and they can’t avoid falling into the same traps that captured the current media institutions when they follow the same model.

We continue this discussion by examining the power-law dynamics that contribute to this cycle, which exist on Substack and other independent media platforms. A small number of winners gain outsized influence over the network, while the vast majority of active publications produce great content despite struggling to gain traction.

Power Laws

When examining the success cases on Substack, we should pay attention to the probability of success on the platform. The distribution of outcomes on Substack follows a power-law distribution, meaning that the top few publications achieve astounding success while the vast majority of publications only have a handful of paid subscribers each.

To use some quick math with the most recently available public data, there are roughly 20,000 active Substacks publishing at least monthly, and there are roughly 5m paid subscribers on the platform averaging $10/month per subscription. This implies there is approximately $600m/yr to be split among the 20,000 Substacks. So how are these earnings distributed?

The top 27 Substacks collectively earn $22m annually, and the top 50 Substacks earn over $1m/year each. This leaves $550m (at most) for the remaining 19,950 Substacks, or approximately $27k/yr per Substack (~225 paid subscribers per Substack). Yet even for the modest numbers of the bottom 99.8% of Substacks, most of these remaining earnings are concentrated in the top few hundred Substacks. The majority of active Substacks do not get rewarded relative to the quality of work they produce.

This result is not surprising, since many systems naturally result in a power-law distribution of outcomes, as seen with common examples of the Pareto Principle in applications from economics to computer science.

However, since power laws are exponential, as Substack scales the platform, the benefits will concentrate in the select few top publications, without having a measurable impact on the success of the vast majority of contributors. If the goal is to save the media, then the expectation is that the new media system will grow to replace our current, broken system. Yet, if the result of this replacement is that influence and success remains concentrated in the hands of a few powerful voices, then we are left with the same result. The cycle continues.

Again, the ability to utilize the democratizing forces of the internet to empower independent voices to share their perspectives is the key to breaking away from the trend of distrust and polarization in media. However, the direct, trusted relationships between subscriber and journalist on Substack are only beneficial when the scale is small. As we have seen in practice (and as fundamental principles predict), when that relationship scales, we approach the same result of our current system. Substack is just reinventing the same broken system from first principles.

If we wish to actually fix the media, we have to address the fundamental mechanism that results in the concentration of influence in the hands of a select few.

Preferential Attachment

To understand what creates the power-law distribution of influence, we have to understand how information networks form. In media, for both social platforms and traditional publications, we tend to use popularity as an approximation for credibility. We look at follower counts to determine if a social media account shares interesting content, and we look at success and subscriber counts of publications to understand their credibility. Intuitively, we expect a correlation between these metrics and an account or publication’s track record of success.

The result is that publications with a large number of subscribers are more likely to gain additional subscribers.

Think of a new person joining an information network. In the case of Substack, when you create an account, you are met with a list of recommended publications to subscribe to, which is representative of the leading Substacks in different categories based on subscriber count. The biggest, fastest-growing accounts are more likely to capture these new subscribers. Additionally, big publications with large audiences are more likely to have their work referenced in other publications or reposted in Notes or on other social platforms, so you are more likely to come across the big publications in all cases. They can then leverage the scale of their distribution to attract more resources and stronger partnerships, creating competitive advantages.

The same phenomenon exists across social media platforms, as well as traditional media publications. It is a simple mechanic of network effects, where the influence of a publication grows exponentially with the number of subscribers it has. Anyone who has built a following or grown an online community knows that the first subscribers are the hardest to get, and each additional subscriber becomes increasingly easier to attract.

This mechanic is called preferential attachment, meaning that as the network grows, new members are more likely to “attach” to others that already have a large number of connections. When you look at the distribution of influence in a network formed by preferential attachment, it follows a power law since a small number of influential members end up with an extremely high number of connections, while most members only have a few.

Chaos

Networks with power-law dynamics that form through preferential attachment exist in a state of chaos and maximal complexity. In practice, we can observe the consequences of this network structure by looking back to Hamish’s quote on social media dynamics.

“Conflict supersedes consideration. Speed overwhelms verification. This system is, in many senses, a marvel, with massive democratization potential. Anyone can have a voice, and your idea, if the winds blow just right, can reach billions of people in an instant.”

In these systems, a small number of members have outsized influence over the flow of information. News stories and misinformation can quickly propagate across the entire network in a few steps, gaining virality while avoiding having to pass through the review of many individuals. Human nature and negativity bias promote the spread of inflammatory content and shocking clickbait rather than positive, well-informed pieces, since bad news triggers stronger reactions of outrage, fear, and moral judgement.

We don’t consciously want to interact with this type of content though, and we are fatigued by the firehose of infinite-scroll, negative information. Fortunately, inflammatory clickbait is not a requisite to forming an information network. It is simply the result of the incentives used to form our current network structures.

Social Media Craves Chaos

With social media, where advertisers are the customers, platforms aim to keep their systems reliant on these power-law dynamics. When you optimize for engagement, speed, and reach in an attempt to maximize ad revenue, you want your network to exhibit these traits since they have been proven to increase clicks and time spent on platform.

But Substack doesn’t have ads in their Notes feeds (yet), so why would these incentive structures still apply? Even without external ads, the people posting on Substack are advertising themselves. They are attempting to maximize engagement to convert free clicks to paid subscribers. The incentives are the same, since they also want reach and speed and a small shot at virality. When independent writers are forced to compete for the finite attention of paid subscribers in a saturated environment, they will have to resort to the same growth-hacking mechanics used by traditional media to sell their content.

And again, as subscriptions scale, it becomes less of a social media game and even more of a traditional media game. Success in traditional media requires maximizing your paid subscriber base, while also subsidizing subscription revenue with ad sales. This has been the business model of news media since the 1800s. The incentives promote a system formed by preferential attachment and characterized by power-law dynamics. As publications compete for attention, they consolidate in an attempt to save costs through economies of scale and concentrate their influence, all while retaining the virality mechanics that are perpetuated by the state of chaos in which these ad-driven information networks thrive.

In reality, these systems are even more complex than described above, since traditional publications and social media platforms are constantly competing for mindshare within one holistic attention economy. As we all individually attempt to navigate our own information environments, we constantly face the same power-law dynamics formed by the incentive structures underlying the media system we have created.

The Progression of Media

Fortunately, these incentive structures are not required to form an information environment. Think of the joy you feel when discovering a new, independent writer creating high-quality content that aligns with your interests. The magic of Substack is that they have shown how much quality writing exists among the noise of our current media environment, and they have proven the significant amount of demand readers have for this type of content.

But the power-law distribution of success does not align with this diverse distribution of quality. It is clear that the success dynamics are more attributed to the mechanics of the network and not the quality of the underlying work. Consider the advice shared by many successful publications as they attempt to monetize their knowledge of how to grow a subscriber base. Many attribute their success to persistence and luck, not sheer talent. This is what you would expect from a chaotic, power-law network.

Our media systems naturally formed into this network structure as we progressed through technological discovery in the age of information, but that does not mean that we have to stop where we are and accept that this is how it was meant to be. We can continue our progression and incorporate new technologies to solve the problems we have discovered as we continue to scale the information networks we created with the advent of the internet. We can promote quality content while correlating success to merit rather than luck and persistence. We can allow independent journalists to focus on their reporting and earn based on its quality, rather than their ability to growth-hack social media followings.

Looking back at the progression of our information systems, we started with a centralized model of distribution. A handful of national outlets, relying on their ownership of the channels of distribution (printing, airwaves, networks, etc.), held control over the public narrative. This centralized structure was stable, but the stability was at the expense of accuracy. Overly simplified narratives and selective reporting did not provide the nuanced perspectives necessary to understand the complexity of the world around us. The wool was over our eyes.

As distribution was democratized with the internet, our information networks evolved into a structure characterized by preferential attachment. We had more options as consumers, allowing us to align our subscriptions with our existing beliefs, creating bespoke realities through which we filtered all of our information from the outside world. We were skeptical about our new digital environments (especially the security of online payments, which now feel second-nature), so the free-to-use platforms reduced the risk to consumers as they tested the waters of digital media.

While this was a critical step in promoting the transition to democratized information distribution over the internet, the ad-driven networks created perverse incentives that maximized chaos to optimize clicks and engagement. We traded the stability of our centralized networks for the diversity of our current online communities, but the subsequent chaos pigeonholed us into tightly-clustered information silos. The newfound diversity did not drive nuance and understanding of complexity, but instead fragmented our media environment, resulting in a multi-decade trend of increasing polarization and decreasing trust.

Decentralized Media

Now, we finally have the ability to make the next logical step in this progression. We no longer need to solve for early user adoption of the internet. We can now solve for stability and quality, all while retaining the full diversity of perspectives we achieved through the democratization of distribution. The final piece to the puzzle was the development of new technology capable of enabling fully decentralized information networks, which was not a practical reality prior to 2024.

With a decentralized information network, consumers can subscribe to the network as a whole. Without ads and without forcing competition for subscriptions between independent journalists, we remove the perverse incentives that drive chaos and open the door to outside influence. Subscribers can freely explore the nuance and complexity of diverse perspectives with a single subscription, and the network can optimize to promote quality and reward journalists based on the merit of their work, not just luck and persistence.

Critically, decentralization creates this structure without concentrating control within the network itself. The journalists own and earn from their work all without risk of censorship or narrative control at the platform level. The community democratically controls the incentives and has the ability to investigate and understand the mechanics of distribution in a fully transparent environment. There is no longer a tradeoff of stability for accuracy. Independence, democratization, quality, speed, and reach can all coexist in one cohesive ecosystem.

Next week, we will continue with Part III to share how Aemula is building a decentralized information network as described above and how it rebuilds trust, reduces polarization, and supports true independence in media.

This week, we highlight writers discussing the chaos of our current media environments and the progression we have taken to arrive at our current state of digital media. We encourage you to explore their work and consider subscribing directly.

Dwarkesh Podcast

Hosted by Dwarkesh Patel, a writer and interviewer known for long-form conversations with leading thinkers across artificial intelligence, economics, history, and science, exploring the ideas and technologies shaping the future.

Dwarkesh Podcast

Why Leonardo was a saboteur, Gutenberg went broke, and Florence was weird – Ada Palmer

Listen now

3 months ago · 100 likes · 12 comments · Dwarkesh Patel

“Not only did Gutenberg go bankrupt in the 1450s (after inventing the printing press), but so did the bank that foreclosed on him, and so did his apprentices. This is because paper was still very expensive, and so you had to make this big upfront CAPEX decision to print a batch of 300 copies of a book - say the Bible. But he’s in a small landlocked German town where only priests are allowed to read the Bible - so he sells maybe 7 copies. It’s only when this technology ends up in Venice, where you can hand 10 copies to each of 30 ship captains going to 30 different cities, that it starts taking off.

Speaking of which, the printing revolution wasn’t just one single discrete event, just as the computer revolution has been this whole century of going from mainframes -> personal computers -> phones -> social media, each with different and accelerating social impact. Books came first, but they’re slow to print, and made in small batches. The real revolution is pamphlets - much faster, much harder to censor. Pamphlet runners are how you can have Luther’s 95 Theses go from Wittenberg to London in 17 days.”

Hot Takes

Written by Adam Singer, a longtime commentator on tech, marketing, and digital culture whose newsletter delivers sharp, occasionally snarky takes on philosophy, innovation, and creativity, drawing from years of experience across startups, agencies, and global brands, previously featured in our spotlight, “Threat of Fractured Reality”.

Hot Takes

The media's obfuscation of reality continues

Just the other weekend, two ISIS-inspired terrorists came within a hair’s breadth of mass casualties near Gracie Mansion. They packed their devices with bolts and screws and tried to hurl them into a crowd of protesters. On…

Read more

3 months ago · 15 likes · 6 comments · Adam Singer

“This is a textbook specimen of what I’ve been calling the subprime attention bubble, and its most corrosive variant. Not only do they publish clickbait, but what we have here is institutional framing-as-product, where legacy outlets with genuine reputational capital to protect spend it in service of ideological narrative management. The facts were not hidden, they were available to anyone who read past the headline. But many people consume news in headline form, not article form. The framing is the product. And the product here was deliberate misdirection.

The subprime mortgage crisis happened because people confused packaging for value: the logos, the ratings, the prestige all intact while the underlying asset quietly rotted. The same dynamic is playing out in media. The mastheads are still there, the checkmarks are still there, the inheritance of decades of institutional credibility is still being spent. But many people haven’t marked it to market yet.”

Optimally Irrational

Written by Lionel Page, a professor of economics and a behavioral scientist with over 50 publications in economics and other behavioral sciences, arguing that many of the cognitive biases and puzzling behaviors discussed in the behavioral economics literature can be understood as good solutions to the problems we actually face in real life, with research covered by major media outlets including The New York Times, The Wall Street Journal, the Financial Times, The Times, The Daily Telegraph, The Independent, ABC, SBS, The Australian, and Der Spiegel, previously featured in our spotlight, “Divergence”.

Optimally Irrational

The game theory of cooperation

In this series of posts, I am offering a deep explanation of what human morality is. I started this series by pushing back on many popular views about morality, such as the idea that it comes from religion or that there are absolute moral truths “out there…

Read more

3 months ago · 63 likes · 20 comments · Lionel Page

“The conclusion is that high levels of cooperative payoffs can be reached in repeated interactions, even in situations where no cooperation is actually stable in single interactions, if players care enough about the future. Players can adopt conditionally cooperative strategies, and it is the conditionality of these strategies that enforces the behaviour of the players in the present.

In short, the folk theorem says that once interactions are repeated, players can adopt a shared understanding of the type “I’ll be nice as long as you are nice”. This is sustainable and individually rational (no need to assume that people are altruistic saints).

An equilibrium here takes the form of a commonly agreed rule of behaviour. This rule determines how to act in each period. A given rule leads to an average payoff in the long run.”

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

This is Part 1 of How to Fix the Media.

How to Fix the Media - Part II

The media is broken.

Traditional publications have lost public trust as they consolidate through rounds of mass layoffs and compromise their integrity with ownership structures that lead to partisan allegiances and centralized points of weakness.

Social media, once envisioned to democratize access to information, relies on perverse incentive structures that promote clickbait and inflammatory rhetoric as they optimize to sell ads.

Everyone feels the negative effects of these problems, but the issues seem so overwhelmingly complex that any effort to solve them feels futile. We have grown tired of the discourse.

This is why we respect the optimism Hamish McKenzie, co-founder of Substack, brings to the discussion. In his recent post, “How to Save the Media”, he lays out his vision for addressing the chaos of modern media while announcing his upcoming book of the same name.

And we agree with his analysis of our current media ecosystem as he outlines in his referenced work, “From the temple to the garden”.

“For decades, we lived in a world dominated by traditional media, characterized by top-down control, centralized authority, and official opinion. It was a stable system, and, though it probably offered some false comforts, it communicated a sense of security in its predictability”

The apparent cohesion of pre-internet media was merely an illusion, where a simple narrative was spun by the few voices that controlled public perception. We traded nuance and understanding for stability, leaving our media systems vulnerable to manipulation by outside influence and those in positions of power.

“Traditional media’s rigid order has been replaced by mayhem. Conflict supersedes consideration. Speed overwhelms verification. This system is, in many senses, a marvel, with massive democratization potential. Anyone can have a voice, and your idea, if the winds blow just right, can reach billions of people in an instant.

(…)

Of course, there are social effects too. We’re no longer tuning in for rituals and rites, but instead subjecting ourselves to a parade of loudmouths and lunatics, where common sense struggles to find a voice. We’ve gone from catechism to cacophony. Trump, Musk, Kardashian, Cuban, AOC, Hawk Tuah. We are now preeners, yellers, credulous fools, conspiracy theorists, and moralizers. Sometimes it’s fun!”

With the internet, the democratization of information has also resulted in the mayhem that comes with social media’s cultural influence, specifically with the role it plays in sharing news. However, the pros and cons of Hamish’s above points cannot be separated, since the conflict, speed, and viral potential of social networks are the fundamental features of these systems. This mayhem is what makes it fun and engaging. This is what captures attention to sell ads.

Social media platforms are incentivized to protect these aspects of their networks. There is massive democratization potential, but it will continue to be overwhelmed by the viral, inflammatory slop that maximizes engagement metrics to sell ads as long as those incentive structures exist.

“Our media systems don’t just convey information—they shape how we think and behave. Political culture responds accordingly. We’ve seen how this cultural interplay unfolded with the arrival of the printing press, which in the mid-1400s broke the literate elite’s monopoly on information and subverted the Catholic Church’s dominance. We saw it with the introduction of radio, which allowed political leaders, for the first time, to speak directly to citizens in their homes, placing new emphasis on emotional appeal. And we saw how television elevated image and personality above nuanced policy discussion.”

Importantly, we need to recognize the consequences of our chaotic media environment. The information we receive and interact with informs the beliefs we hold, the decisions we make, and the politicians we elect. It is a constant feedback loop that reverberates around the system, amplifying the chaos of our media into our cultural zeitgeist. Geopolitics emerge from the interactions caused by decisions we individually make based on our own understanding of the world around us.

We have always agreed with Substack’s mission, observing the same problems that plague our media system while understanding the critical importance of finding a solution.

Where we disagree is on the solution.

“How to Save the Media is about this era of transition, where old gatekeepers give way to new opportunities and direct relationships reign supreme.”

Yes, the solution relies on the independence of our media. We no longer need to rely on institutional publications to control our flow of information. And yes, this romanticized vision of the future where independent journalists have direct relationships and fireside chats with their audiences sounds like a wonderful alternative reality.

But the logical conclusion of scaling this solution to societal relevance results in us ending up right back where we are today.

Consolidation of Traditional Media

Consider the current state of traditional media outlets, where large, institutional publications have outsized influence over original reporting. Their scale and access to resources grant them a competitive advantage in producing high-quality news, earning millions of subscribers paying premium prices. However, their centralized control creates a single point of weakness that can be exploited in an effort to push specific narratives, or for powerful individuals to buy influence over the output from these newsrooms.

The outside influence over our media systems does not even need to be explicit, but can exist in the incentive structures behind the scenes of the daily course of business for systemically important media institutions.

“American media has not imploded. It has consolidated. Fewer companies now control more of what you read, watch, and scroll past before sunrise, and those companies are engaged in the ordinary but consequential work of large corporations—negotiating mergers, defending lawsuits, seeking regulatory approvals, trimming budgets, reshuffling leadership—activities that rarely make headlines but shape the environment in which headlines are written.¹

When a parent company needs a federal agency to approve a transaction, that creates leverage whether or not anyone names it. When executives decide a lawsuit carries unacceptable financial or regulatory risk, that decision becomes precedent, and precedent quietly reshapes future choices. No editor has to receive a call. No anchor needs a memo. The shift happens at the level of institutional risk calculation, and coverage adjusts at the margins.²”
— Bill Southworth, “Just the Facts, Ma’am”

The consolidation and centralized weaknesses of institutional publications have led to an increasing share of Americans believing that the government has a great deal of influence over US news outlets, growing from 49% to 57% between 2024 and 2025 (with 86% of Americans believing that political interests have some influence over news outlets). This is why we have seen increasing mentions of people referring to “state media” in the US. It is no wonder that Americans are losing faith in the freedom of the press.

Now consider Substack’s direct subscription model. With individual reporters, their small scale and independence protect them from the pitfalls of institutional publications, and the ability for readers to form trusted, direct relationships with these reporters while supporting their work financially is a beautiful model. But what does success look like in this model?

When a reporter gains a large following (or brings a large following with them from the institutional publication they were at prior to launching a Substack), their growing audience starts to demand more out of them. And for good reason, with readers paying an average of $10/month to individual Substacks, that is the equivalent of the average monthly subscription price paid to The New York Times. Certainly readers would expect some sort of parity in the level of coverage.

So successful independent reporters start to build independent publications. They hire additional journalists to fill out their newsrooms, they set up partnerships to get more resources to improve their reporting, and they create social media campaigns to attract new subscribers. They can grow large, successful businesses from their independent newsrooms, to the point where they gain enough influence to capture the attention of the institutional publications they set out to disrupt. Look at Bari Weiss’s The Free Press, which started as an independent publication on Substack until it was acquired by Paramount and CBS News for $150m.

The success case for an independent publication on Substack is to become part of the consolidating, institutional media environment. As you scale the solution, it just becomes a new version of the same problem.

And why would you expect it to be any different? The fundamental business model of an independent publication on Substack is no different from that of an institutional publication. The only difference is scale.

Subscribers pay to gain access to a single source of information, and that source has to compete to maintain that subscription by playing all of the same attention-economy games that a traditional publication has to play.

Note that The Free Press is just a single example, and there are many variations of the success case for publications on Substack. For instance, Tangle successfully scaled their Substack into an independent newsroom, and they continue to do great reporting. However, they are an exception, not the rule, as they work hard to fight against the underlying incentive structures at play when scaling a publication.

But instead of scaling a few large publications to spin out of Substack, what if we were able to scale Substack as a whole, with paid subscribers dispersed equitably across the full spectrum of diverse perspectives? Certainly that would result in a more nuanced, independent environment than today’s media ecosystem. Or are there structural reasons why scaling the platform results in only a select few winners with the vast majority of publications never gaining traction?

Next week, we will continue our discussion on how to fix the media by exploring the power law dynamics that determine success on Substack and other social media platforms, as well as their impact on the quality of our information environment.

Read “How to Fix the Media - Part II” here.

This week, we highlight writers who are discussing the current state of media in the age of AI, tying in some of the themes discussed last week in our spotlight, “AI isn’t going to steal your job”. We encourage you to explore their work and consider subscribing directly.

The Intrinsic Perspective

Written by Erik Hoel, author of The Revelations, an essayist with work featured in The Atlantic, and theoretical neuroscientist known for work on causal emergence and the Overfitted Brain Hypothesis, previously featured in our spotlight, “Life Imitates AI Art”.

The Intrinsic Perspective

Bits In, Bits Out

When I was ten years old I visited the ruins in Cornwall where King Arthur had been conceived and born, at least in legend. There at Tintagel Castle, surrounded by the ocean air and the jagged rocks, I separated from my mother and sister and made my way down to the beach. And on the beach of Tintagel, right near Merlin’s cave, I spotted…

Read more

3 months ago · 136 likes · 32 comments · Erik Hoel

“If LLMs were a true source of intelligence to rival humans, then discovering them should be like discovering oil. And if we were climbing the curve of an intelligence explosion their surplus intellect would be improving our civilization’s text as a whole in noticeable ways. If LLMs are tools, then we should expect their impacts to be a mirror of us, and concern efficiency and scale, rather than quality, and depend strongly on how people use them.

So let me ask you: if you took an observer from 2016 and teleported them a decade ahead to our time, and then showed them your social media feed or your emails and other media in general, what would their main response be? Would it be ‘Wow, everything is more intelligent now!’ Or would it be ‘Why is everyone writing like a pod person now?’”

Conspicuous Cognition

Written by Dan Williams, an academic philosopher with a PhD from the University of Cambridge, a lecturer at the University of Sussex, and an associate fellow at the Leverhulme Centre for the Future of Intelligence at the University of Cambridge, previously featured in our Spotlight, “I, Sandwich”.

Conspicuous Cognition

How AI Will Reshape Public Opinion

Epistemic status: highly speculative, big picture, maddening…

Read more

3 months ago · 157 likes · 34 comments · Dan Williams

“Of course, by influencing who speaks, communication technologies also influence what gets said. A media environment regulated by elites will marginalise information that threatens elite belief systems. But the medium also shapes the message in other ways. Print permits careful, detailed argumentation. Television favours confident sound bites. As I discuss below, social media often rewards division, conflict, and negativity.

These forces impact how audiences attend to and interpret reality, the ‘pictures in their heads’ that guide which leaders, movements, and policies they support and oppose. But they also influence how easily people organise around shared pictures. If gatekeepers block widespread views from a society’s communication channels, people will struggle to learn how widespread they are.”

Church and Main

Written by Dennis Sanders, an ordained minister in the Christian Church (Disciples of Christ) and longtime communications professional, whose work explores the intersections of faith, politics, culture, and everyday life in the modern world.

Church and Main

The Big Tent is Dead. And We are the Poorer For It.

Note: This is something I wrote nearly four years ago. I am planning on writing about how mainline Protestant denominations have gone from liberal to liberationist and how that has made them less hospitable to people who aren’t progressive. This is looking at this cultural segregation from a larger perspective…

Read more

3 months ago · 3 likes · Dennis Sanders

“When I look back at those days in my youth, I realize how much the mass culture and the big tent no longer exist. We don’t live in a world where institutions want to welcome a wide swath of the population. In fact, those institutions are breaking down. Our modern culture has created small like-minded communities where people can join certain groups and not have to deal with others that are different from us.”

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

The easy solution was to generate this newsletter using an LLM. Certainly that would be passable for one week to keep our streak alive. Especially since the bull posting around AI advancement has picked up in recent weeks after the launch of OpenAI’s Codex and Anthropic’s Claude Opus 4.6 model. Seemingly, we will all be out of jobs soon anyway, so why not see if an LLM could do my job for a day?

The narrative around the approaching singularity reached a local maximum this week with the publishing of Citrini Research’s piece forecasting how AI will disrupt the global economy in the coming years (included below). Despite starting their piece by saying “this isn’t bear porn or AI doomer fan-fiction”, it certainly reads as such. Citrini Research was little known prior to the post, but after going viral and earning millions of views across platforms, public markets traded down heavily on the stocks singled out in the piece. The impact of the piece was so great, that Substack’s Chris Best even gave it a shoutout, using it as an example of the platform’s cultural influence.

However, I am yet to be convinced that Monday’s public-market drawdown was a direct result of the Citrini Research Substack piece. If traders were actually trading the thesis of the piece, we would have seen different results in markets. But Citrini’s bear case definitely fit the cultural narrative growing around AI’s potential for disruption.

The traditional bear case is that if AI is a bubble and we don’t achieve AGI (the singularity) in the next few years, then the massive bets placed on AI research labs and data center fabrication will all go belly-up, dropping us into a recession. When GDP growth is driven entirely by data center capital expenditures, it certainly feels as though the burst of the AI bubble would not be good news for the economy. Additionally, the Blue Owl/CoreWeave debacle and liquidity crunch from last week did not help to ease those concerns. Fortunately, yesterday’s strong Nvidia earnings Flex Sealed the leaks in public sentiment, pushing off any second-guessing of the AI bubble into future quarters.

However, Citrini proposed an alternative downside scenario, where AI is not a bubble, and we do achieve AGI capabilities in coming years. In this case, the disruption to our current economic structure results in widespread layoffs and a dystopian vision of the future of work. My hesitation on the impact of this piece is that people must have already considered this scenario if they have been worrying about AI’s impact on markets in the near-term. This has always been the result of the bull case for AI, so why would the expectation of economic disruption not already be priced in? My guess is that the Citrini article was just in the right place at the right time.

While I would love to continue expanding on these topics, Aemula’s Writer Spotlight is not the best forum for my personal predictions for AI’s impact on financial markets (yet, here we are…), and I admittedly don’t have the energy to form a coherent argument right now.

Though what I do know is that I tried to write a simple newsletter using the flagship models from OpenAI and Anthropic, the best ones available to the public, and they were, quite frankly, terrible.

I figured the task should be relatively straightforward. Using our past 10 newsletters as reference, I could just prompt the LLM to review the reference newsletters to understand their structure and style, then do a web search to find a relevant, trending topic that we haven’t covered, then draft the piece.

Using Anthropic’s Opus 4.6 with extended reasoning, it chose the topic of news avoidance, where readers are overwhelmed by the pace of content and negative sentiments in news coverage so they choose to opt out of following the news altogether. Admittedly, this is a good topic choice, but it hallucinated facts and made up sources to support the argument (see below). The output was also clearly AI-generated, following the tell-tale LLM syntax of “its not x, its y”. I don’t want to subject you to the full piece, but here is an excerpt if you’re curious.

you probably know someone who has stopped reading the news

maybe it is you

it is not that they do not care. it is not that they have given up on the world. they are tired. overwhelmed. they open an app and are met with a wall of urgency, each headline competing to be the most alarming. so they close it. they check the weather instead. they scroll past the politics and settle into something that does not make them feel like the floor is collapsing

this is not apathy. it is self-preservation

the Reuters Institute recently published their annual trends report for 2026, and the numbers tell a story we have been watching unfold for years. News avoidance has climbed to 40% across global markets.

Here is the actual Reuters report for reference, where news avoidance isn’t mentioned.

Despite attempts to refine this draft through prompting, I cut my losses and turned to ChatGPT 5.2, which came up with the topic of AI search summaries leading to declining referral traffic for news publications. Also a solid topic choice, but the output was clearly LLM-generated as well, even though I gave ChatGPT the same newsletters as reference. Again, if curious, here is an excerpt.

The Summary Is Not the Story

Last week, several independent writers began noticing something unusual.

Their traffic had not dipped gradually. It had fallen off a cliff.

Not because they wrote less. Not because they lost subscribers. Not because public interest waned.

But because their work was summarized.

In the end, I was spending more time trying to prompt these drafts into something passable than it would take me to just write this myself (though, I do realize this is a bit of a fevered rant, so the jury is out on whether the AI-written newsletters would have been a better read). But I think we are all in tune enough with AI-style writing to recognize when something is not written by a human. With so much AI-generated content these days, most people would likely skim past anything that doesn’t pass the sniff test.

And for good reason. Why should I expect you to spend your time reading a newsletter I prompted into existence when I didn’t even take the time to put thought into writing it?

Note, I say all this as someone who uses AI extensively in my daily workflow. I have had a ChatGPT Plus subscription since it was first offered in 2023. I’ve gone through stints with the research and max plans offered by OpenAI and Anthropic, and now maintain subscriptions to both providers. I have an OpenClaw agent running on a Raspberry Pi on my desk. I use Claude Code and Codex daily, and I even competed in an Anthropic-hosted coding event last weekend. I have built AI-native apps and spent countless hours studying the underlying tech.

I am an expert by no means, but all of this is to say that I have more experience with LLMs than your average bear, and I have witnessed how these tools have evolved over the past few years. While the hype cycle might make it feel as if we are sprinting towards the singularity and all of our jobs will be replaced by AI in the next few years, my actual experience does not align with this narrative.

The fundamental capabilities of LLMs have not progressed much since I started using them heavily in 2023. Yes, they are more efficient and our usage limits have increased significantly. Yes, there is better tooling so it is much easier to integrate an LLM into existing workflows. Yes, chain-of-reasoning allows us to abstract our prompts and rely more on the LLM for planning long tasks and executing incremental steps itself.

But these are all quality-of-life improvements. 85% of the value-add from LLMs was there 3 years ago. My OpenClaw agent rarely knows what day it is. I can’t even trust it to manage my to-do list. Claude Code agents get stuck in tasks that end up burning a full day’s usage limit without making meaningful progress. Opus 4.6 was compute-constrained last weekend, causing it to not even respond to requests.

And the flagship models couldn’t even write this newsletter.

They could write a newsletter, yes, but not one that could replace my actual work. And this represents my beliefs on the current state of AI. These models are great for putting together demos and producing B- grade work, but they are not yet at the point where we can confidently rely on them without a human in the loop. The progress over the last 3 years has not materially changed this experience.

I am not alone in this belief, and I will appeal to authority to support these claims. Yann LeCun, considered one of the “Godfathers of AI”, believes that LLMs are a dead end on the path to superintelligence. Similarly, Richard Sutton, the father of reinforcement learning, shares the belief of LLMs being a dead end.

In terms of outcomes for the current iteration of AI, I believe it looks something more like what the Jevons Paradox people are pushing. Our ability to leverage AI tools to conduct our work more efficiently will have a positive impact on economic growth. This will more likely result in people doing more work in their current jobs. Some jobs and industries might rotate their scope of work, but they likely won’t be replaced entirely. The winners of this AI cycle will more likely be the ones that are applying LLM efficiency to proprietary data streams and tightly defined use-cases (such as assistance for finance, law, specific fields of scientific research, etc.).

Though, there is mounting evidence that supports the contrary. For one, Block laid off 40% of their staff today because they believe they can be just as efficient with less people. Some might say that they were overstaffed in the first place, and I would point out that Jack Dorsey was likely avoiding capital letters in his X post to announce the layoffs so that the post didn’t appear to be written by an LLM (much like how the above excerpt on Anthropic’s attempt at this newsletter didn’t use capital letters when I prompted it to avoid using LLM-style writing). But this does fit in nicely with the beginning of the story told in the Citrini piece.

And a researcher at an AI lab studying LLM use in writing scientific papers told me that they are witnessing LLMs come up with novel discoveries (though, he admits that these discoveries are mainly a result of being able to cross-reference the entire history of research papers in a given field, often finding solutions to problems in overlooked footnotes in decades-old papers). The case of LLMs as a dead end is that they are just aggregators of information, and not capable of coming up with novel discoveries on their own. This piece of anecdotal evidence seems to gray that line a bit, so it’s worth seeing how it progresses.

I will point out that I have used fairly soft language in making the above claims. Saying things are “likely” or not, so I can avoid taking a hard stance and being proven wrong. And again, perhaps a brief AI-generated piece on news avoidance would have been a more enlightening read than this loosely constructed rant on my frustrations with AI not being able to do my work for me today.

However, what I can say definitively, is that LLMs will not replace the journalistic process. To understand the world around us, LLMs rely on a constant stream of human-generated data, encompassing the full diversity of beliefs and biases that make up the collective human perspective in real-time. This is the fundamental process of truth-seeking, where we come up with novel ideas to share and discuss as we attempt to uncover the truth.

With LLMs, this is “the Galileo test”. Though, I would call this the Copernicus test. If we trained an LLM on all of the available data prior to Copernicus, would it be able to prove the heliocentric model of our solar system? Could the LLM go against the consensus narrative and uncover the truth from the data, proposing methods to fill the necessary observational gaps? Or do we need to rely on humans to have these paradigm-shifting ideas?

We watch as AI research labs commit billions of dollars to license the reporting coming out of professional newsrooms. High-quality, real-time data is a constraint in training models and allowing them to provide accurate answers on questions about current events. The only source of this information is through the human-centric journalistic process. If I were an AI research lab expecting to achieve AGI in the next few years, I certainly wouldn’t spend $3b on 5-year licensing contracts with The Wall Street Journal and The New York Times. Which is why I am confident in saying that journalism isn’t going to be replaced any time soon, and in fact, the demand for human-generated news is increasing drastically.

This is why with Aemula, we are building an ecosystem for real people to share real news. With verified humans independently reporting, we can ensure that readers have access to high-quality information in a focused environment free of AI-generated slop. Additionally, we can enable independent journalists to actually get paid for their work when an LLM cites it as a source, or allow them to protect their work from being used in LLM responses altogether.

Where we leverage AI, we are doing so to increase the efficiency of independent reporting, giving journalists tools to assist in source discovery and managing their drafting worflows. And for our readers, we can utilize LLMs to ease the process of discovering new perspectives and articles from the journalists they support, rather than simply summarizing work and losing important context. AI is a tool that should be leveraged effectively, but in its current iteration, we are committed to supporting the human-centric journalistic process.

This week, we highlight writers discussing the topics mentioned above. We encourage you to explore their work and consider subscribing directly.

Citrini Research

Written by Citrini Research, a thematic investment publication focused on identifying early-stage megatrends through macro analysis, second-order thinking, and carefully constructed equity baskets designed to capture asymmetric, long-term returns.

Citrini Research

THE 2028 GLOBAL INTELLIGENCE CRISIS

Preface…

Read more

3 months ago · 6242 likes · 83 comments · Citrini and Alap Shah

“What follows is a scenario, not a prediction. This isn’t bear porn or AI doomer fan-fiction. The sole intent of this piece is modeling a scenario that’s been relatively underexplored. Our friend Alap Shah posed the question, and together we brainstormed the answer. We wrote this part, and he’s written two others you can find here.

Hopefully, reading this leaves you more prepared for potential left tail risks as AI makes the economy increasingly weird.”

Fakepixels

Written by Tina He, a writer, designer, and entrepreneur, now leading developer tools at Base, and the former co-founder and CEO of Station Labs, which Coinbase acquired in June, 2024, previously featured in our spotlight, “Life Imitates Art”.

Fakepixels

Jevons Paradox: A personal perspective

“I'm dying. Of overwork…

Read more

a year ago · 366 likes · Tina He

“This is the paradox of our time: the very tools designed to free us from labor are trapping us in an endless cycle of escalating work. As our productivity increases, our standards and expectations rise even faster, creating a psychological Jevons Paradox that threatens to consume our humanity in the pursuit of ever-greater output. We become victims of our own efficiency.”

Astral Codex Ten

Written by Scott Alexander, whose writing first came to prominence through the blog Slate Star Codex, where Meditations on Moloch became an often referenced piece in Ethereum subculture, featured in last week’s spotlight, “We Can Predict the Future”.

Astral Codex Ten

Next-Token Predictor Is An AI's Job, Not Its Species

Read more

3 months ago · 279 likes · 261 comments · Scott Alexander

“In The Argument, Kelsey Piper gives a good description of the ways that AIs are more than just “next-token predictors” or “stochastic parrots” - for example, they also use fine-tuning and RLHF. But commenters, while appreciating the subtleties she introduces, object that they’re still just extra layers on top of a machine that basically runs on next-token prediction.

I want to approach this from a different direction. I think overemphasizing next-token prediction is a confusion of levels. On the levels where AI is a next-token predictor, you are also a next-token (technically: next-sense-datum) predictor. On the levels where you’re not a next-token predictor, AI isn’t one either.”

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

The timing of this announcement is fortuitous, as we have also just released new prediction market features on Aemula, aligning with our roadmap to outline how prediction markets will improve our information environments. This week, we discuss prediction markets, their benefits, their downsides, and how Aemula addresses common concerns in our implementation of the technology.

Prediction Markets

Prediction markets, in their common form, are binary (yes/no) option markets for real-world events. Participants are able to place yes or no bets on whether or not they think a specific future event will occur, with the price of placing those bets reflecting the market-implied probability of the outcome. After the event, the market resolves, with the correct side receiving a payout.

To use specific numbers (and highlight the new Substack feature), see the below prediction market on whether or not Prince Andrew will be sentenced to prison after his arrest on suspicion of misconduct in public office in relation to his connections with Jeffrey Epstein.

At the time of this writing, the market indicates a 10% chance of Prince Andrew being sentenced to prison. Using traditional market terminology, this means you can purchase a “yes” bet for $0.10/share. If you believe the odds of a prison sentence are greater than 10%, then a “yes” bet would be underpriced at $0.10/share, creating an opportunity for you to profit by purchasing a “yes” share.

You could either sell your “yes” share once the price exceeds a probability you think is accurate, or you could hold onto the “yes” until the market resolves. In the event Prince Andrew is sentenced to prison, the value of your bet would resolve to $1/share (100% probability since it already occurred).

For example, if you bought 100 “yes” shares for $10 and correctly held them until the prison sentencing, you would receive back $100. However, if he were not sentenced, your shares would resolve to $0/share (0% probability) and you would lose your $10.

So the inverse of the market is also true, where if you think 10% odds of a prison sentence are too high, you can purchase a “no” bet for $0.90/share (90% chance of no, 10% chance of yes). With a $10 bet, you could purchase 11.11 shares, which would receive $11.11 if you held onto the bet and Prince Andrew was not sentenced to prison.

When you believe the probability of an outcome is misstated, then you can profit by placing a bet on the side of the market you think is underpriced. People will continuously place bets until the market price moves to a point where they believe it accurately represents the real probability of the event occurring, meaning there is no perceived opportunity to profit. Of course, there are additional frictions to this process in practice, but this is the basic premise.

Fundamentally, the purpose of prediction markets is to use price signals and market dynamics to determine the general consensus probability of a specific outcome. Everyone has their own information and their own opinions on the probability of the outcome, which they record by placing bets until all of the relevant information is reflected in the market price.

In practice, prediction markets have proven their ability to accurately predict the outcome of important events. Notably, in the 2024 US Presidential Election, Polymarket and other prediction markets were more accurate in forecasting the election results than traditional polling.

For this reason, Polymarket included in their announcement of the Substack partnership that "journalism is better when it’s backed by live markets.”

Admittedly, this relationship might not be immediately clear, but in ideal conditions, as journalists work diligently to investigate sources, provide credible information, and uncover the truth, prediction markets can be a powerful complement to their process, informing them on public sentiment and consensus expectations. Importantly, prediction markets should be viewed as a tool for discovery rather than a replacement for the journalistic process. They can signal what stories are capturing public attention and which uncertain events would benefit from additional reporting, but market participants are only able to make accurate predictions if they have access to high-quality information.

Journalism and prediction markets create an iterative feedback loop that improves overall information quality, so it comes as no surprise that newsrooms are working closely with prediction markets to provide these tools to their journalists. Kalshi, the primary competitor to Polymarket in the prediction market arms race, recently announced their partnership with CNN. Similarly, Polymarket struck an exclusive partnership with Dow Jones, a division of News Corp and parent to The Wall Street Journal, Barron’s, and MarketWatch.

However, as prediction markets continue to become integrated as a tool for information discovery, people are voicing their valid concerns over the potential for these tool to introduce misaligned incentives.

History of Prediction Markets

To address these concerns, it is worth understanding how prediction markets rose to prominence in their modern form. The concept of collective wisdom through independent decision-making has existed for hundreds of years (the history of which is unfortunately outside of the scope of this newsletter), but modern prediction markets first started to take form when three University of Iowa professors ideated the Iowa Political Stock Market (later referred to as the Iowa Electronic Markets) as a financial market for betting on the outcomes of political elections based on their belief that market dynamics could outperform traditional polling. The accuracy of this early implementation began to gain traction, leading to the advent of corporate prediction markets, first implemented as part of Project Xanadu, the first hypertext protocol and competitor to the World Wide Web.

Robin Hanson, creator of the prediction market project at Xanadu Inc and now a professor of economics at George Mason University, began to formalize the concepts of prediction markets into the idea for Futarchy, a form of democratic governance where market speculators would bet on the predicted effectiveness of potential public policies, informing elected officials’ decisions on the best options to implement. He first proposed the idea in 2000 and continued to refine it through official publication of the proposed governance method in the Journal of Political Philosophy in 2013.

Around the same time, the release of new decentralized technologies were beginning to show promise in their capability of providing the infrastructure necessary to implement these decentralized governance ideas in our modern digital landscape. Notably, Vitalik Buterin, the founder of Ethereum, wrote how Futarchy and prediction markets could be used to facilitate collective decision making in decentralized organizations. With the launch of Ethereum in 2015, Augur began development to become the first truly decentralized prediction market, allowing us to test how these tools would work in practice.

After Augur’s launch in 2018, it did not take long for market speculators to begin creating death pools and assassination markets, highlighting how financial incentives for determining the outcome of real-world events can introduce perverse incentives. This raises the question of whether prediction markets are a “thermometer or a thermostat”. Do they allow us to gain insight into the world around us, or do they empower us to incentivize actions and outcomes that would otherwise not have occurred?

The Dark Side of Prediction Markets

For instance, take the prediction market bets around this year’s Super Bowl (to use a more lighthearted example compared to assassination markets). Ahead of the game, there were social media posts circulating about how prediction markets were offering bets on the probability of a streaker running onto the field during the event. After doing some quick math, fans realized they could place a bet that a streaker would halt play, then attend the game and streak the field themselves, earning a profit after accounting for tickets and travel. A streaker did inevitably crash Super Bowl LX, and while it was rumored that he placed a prediction market bet or was paid an endorsement to promote a company, it appears that he was doing it for his own thrills and viral fame, indicating that prediction markets might not be necessary to incentivize such behavior.

More recently though, prediction markets made headlines when social media users called attention to the existence of a Polymarket bet on the odds that the Artemis II crewed mission to the moon would explode. The market has since been pulled offline, but it opened the door to the possibility of a saboteur being incentivized to disrupt the launch with fatal consequences.

However, perverse incentive structures always exist even without prediction markets. In the case of a Super Bowl streaker, companies could avoid shelling out for an expensive broadcast advertising slot by simply paying for someone to attend the game and capture more attention by streaking the field waving a banner. Or, as seen with the 2021 streaker, someone could have acquaintances place prop bets with offshore sports book, Bovada, to earn a payout from the act (though the bets were voided after the streaker announced this is what they were doing). It also doesn’t take much of a conspiracy to imagine how a foreign adversary would attempt to incentivize the sabotage of a US space mission. Fortunately, we have a rule of law in place to punish illegal behavior, which typically serves as the deterrent against these negative incentives.

After the moral false starts of decentralized prediction markets, regulators were quick to limit their use until they had a better understanding of their effects and mechanics. As the regulatory landscape began to incorporate the new technologies, and decentralized platforms improved their user experience and addressed the technical limitations stunting their adoption, prediction markets began to make their return. Notably, Polymarket and Kalshi, now regulated by the CFTC, were allowed to grow their US-based audience in recent months.

Yet, these prediction market platforms still face an existential threat since the public still views them as “gambling on everything”. And for the most part, this is not wrong. As with any nascent technology, the volatility of initial adoption and lack of oversight introduces the ability for speculation and manipulation by early risk-takers. In the case of decentralized technology, it is important that we look past this phase in order to discern the computer from the casino.

Still, for now, the vast majority of recent trading volume on both platforms is sports betting, and the remaining minority of bets still raise the question of whether people should be allowed to profit from gambling on overthrowing foreign regimes, exploding rockets, and the Second Coming of Jesus Christ. However, it is specifically in these scenarios, the ones involving life or death decisions, that we should leverage the most accurate information possible. Of the tools at our disposal, markets do a phenomenal job (not a perfect job) at predicting the future, drawing our attention where it matters most.

For example, after the explosion of the Space Shuttle Challenger, financial markets determined the contractor responsible for the faulty O-rings within minutes, signaling this information five months before the government investigation concluded the cause of the disaster. Within journalism specifically, financial markets predicted the declining profitability of newsrooms due to competition with internet platforms for more than five years before revenues actually started to decline. If we have the ability to access accurate forecasting for the outcomes of critical events, we should leverage it to the best of our abilities.

Prediction Markets on Aemula

Fortunately, there are better ways to benefit from the insights offered by prediction markets while avoiding many of the negative edge cases. In November of 2024, Vitalik Buterin outlined the vision for “info finance”, which utilizes prediction markets to improve informational quality and “create better implementations of social media, science, news, governance, and other fields.”

We largely agree with Vitalik’s vision of improving our information environments. However, we differentiate in our belief that “finance” is not explicitly necessary to achieve the benefits of info finance. Instead, we align more with the implementation of the Good Judgment Open, a reputational-based forecasting competition as part of the Good Judgment Project, the IARPA-funded superforecasting research initiative (we recommend the book Superforecasting, which discusses this in more detail). With Good Judgment, forecasters stake their reputations rather than their finances when casting their predictions. While the forecasters participating in the Good Judgment Open challenges do it mostly out of their love for the game, the concept of betting reputations in prediction markets solves some core issues in their implementation.

Importantly, financial markets use prices as their signal because there is a correlation between the amount of money someone can wager and their ability to make a judgement on an outcome. People who have a financial stake in a company’s financial performance are most incentivized to ensure the price of their shares are accurately reflected. People who are best at discerning the credibility of share prices in financial markets typically earn more money, and are able to place larger bets, weighting the decision making power that drives price signals into the hands of the experts. The key mechanic is that the more conviction someone has that they are right, the bigger a bet they are willing to place. Additionally, the size of the bet they consider to be big is relative to the amount of financial wealth they control, which should theoretically be proportional to their historical financial success.

However, in information markets, the amount of capital you have access to is not necessarily correlated with your ability to make a judgement on a specific outcome. The type of participant we want to incentivize into the market is someone who has unique insight into a specific event. Prediction markets, when adopted at scale, should be wide and shallow, meaning there will be a large number of events to wager on, with each event having relatively low trading volumes. You do not want deep-pocketed generalists coming in to manipulate low-volume markets, but you want individuals holding niche expertise providing high-signal bets on potential outcomes. Currently, in this type of prediction market, financial capital is an external resource that is not correlated to one’s success in the prediction market as a whole.

If we wish to reconcile this misalignment, we must rely on internal resources to inform the weights we assign to people’s influence in the market. With Aemula, we rely on reputation. Journalists and readers earn their reputation as they build a track-record of providing high-signal insight into the quality of information circulating on the platform.

If you consistently publish credible reporting that is supported by diverse user groups, you will have a high reputation as a journalist. As a reader, if you consistently support high-quality articles, disagree with poorly reported articles, and report harmful content, you will build a strong reputation as a reader. These interactions do not require anyone to make additional financial decisions or place bets, but they continuously refine and update user reputations, which the community can then use to determine what information is worth paying attention to on any given topic.

While wagering reputations is similar to the Good Judgment Open, Aemula does provide financial incentive to increase your reputation. Having a higher reputation as a journalist means that your articles are more likely to be widely circulated and supported, earning you bigger payouts from our community subscription pool for your work. Similarly, high-reputation readers earn a higher weighting in moderation decisions on the platform, providing the opportunity to earn larger payouts from moderation bounties. The key is that the weight of the financial incentives is not tied to your preexisting access to financial capital, but is instead weighted based on the reputation you have built from consistently upholding the values the Aemula community is aiming to promote on the platform.

We are intensely focused on preserving freedom of speech and individual liberties. We believe all journalism should be conducted independently. And we believe that independent journalists should have access to the best tools possible to conduct high-quality reporting. Decentralized prediction markets are a powerful tool to support this mission, as they provide accurate insights derived from individual decision making.

To begin implementing prediction market features into Aemula, we have released a trending topic monitor in the Aemula Newsroom, integrating live market data from Kalshi to show which topics are currently capturing the most attention and which topics are experiencing the most volatility as they develop. These metrics signal a market’s need for more information through better reporting. Note that we are simply using public data and have no partnership with Kalshi, but we aim to make this information easily accessible to journalists within their workflow on Aemula so they can best determine where to focus their efforts.

We are committed to aligning incentives across our media environment, and we will leverage powerful tools to support high-quality independent journalism to the full extent we are able to do so without compromising that alignment. We are realists in our understanding that prediction markets are speculative betting platforms that are still working out the issues of their early form. However, with the correct implementation, we believe they are capable of powering our collective mission to rebuild trust and reverse polarization in media by providing accurate insights into the state of our consensus shared reality.

This week, we highlight writers discussing prediction markets and their impact on social platforms and world events. We encourage you to explore their work and consider subscribing directly.

Feed Me

Written by Emily Sundberg, a New York–based writer and director who draws on her experience across media, brand strategy, and cultural reporting for outlets including New York Magazine, GQ, ELLE, and The New York Times, previously featured in our spotlight, “Forecasting”.

Feed Me

Polymarket wants writers to get into prediction markets.

Good morning everyone. Free letter today…

Read more

3 months ago · 62 likes · 30 comments · Emily Sundberg

“Chris Best, the CEO of Substack, is ‘hyped’ about the platform’s new partnership with Polymarket. It includes updates to Polymarket’s embedding feature, which has been available since 2024. Polymarket is also part of Substack’s sponsorships pilot, so you’ll see ads from them in more newsletters this year. Substack’s announcement letter demoed the feature via shit you should care about and a Polymarket wager about Harry Styles, which tells you this isn’t just about new revenue streams for Substack; it’s about Polymarket positioning itself alongside cultural conversations. I’m guessing we’ll see more female-focused marketing from Polymarket in 2026. I get concerned when people speak about Substack as if it’s a sorority or a community — it is a venture-backed tech company. Here’s how some writers have responded to this news”

Pantsuit Politics

Written by Sarah Stewart Holland and Beth Silvers, cohosts of the independent podcast Pantsuit Politics, whose decade-long conversations and bestselling books focus on making political news more manageable, thoughtful, and grounded in healthier civic dialogue.

Pantsuit Politics

Everything is just a bet away

In late January, I released an episode of More to Say about prediction markets…

Read more

4 months ago · 41 likes · 7 comments · Beth Silvers

“‘The prediction market is partnering with an AI engine that tracks social media data to create markets about cultural relevance’ is a sentence that makes me want to join a nice off-the-grid commune.

As much as I don’t like any of this, and I very much do not, it’s important to understand it. So, here’s a short overview of prediction markets and what you need to know as they gobble up ever more money and attention.”

Astral Codex Ten

Written by Scott Alexander, whose writing first came to prominence through the blog Slate Star Codex, where Meditations on Moloch became an often referenced piece in Ethereum subculture, previously featured in our spotlight, “Information Diets”.

Astral Codex Ten

Prediction Market FAQ

Read more

3 years ago · 163 likes · 316 comments · Scott Alexander

“Prediction markets have two good qualities: in ideal situations, they are accurate and canonical.

By accurate, I mean that that over the long run, they will be at least as accurate as any other source of information.

By canonical, I mean that they short-circuit discussion of ‘which expert should we trust?’ or ‘how do we know which sources are biased?’ All prediction markets speak with a single unified voice, that voice will always be at least as trustworthy as any individual expert, and it cannot be biased. If you’re not sure which of many competing experts (or supposed experts) to trust, you should always trust a prediction market instead of any of them. And the same is true of people on the opposite side of the political spectrum who doubt all the sources you trust and vice versa.”

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

I was weighing two options for this week’s spotlight. The first centered on the updated National Speech Index survey results and what they reveal about perceptions of free speech in the US. The second was more anecdotal, reflecting on how it feels like a full-time job to sift through the overwhelming amount of news generated each day.

Both topics have been the subject of past spotlights, and while worth reiterating, a more intriguing idea lies at their intersection.

The National Speech Index survey aligns with a pattern we have discussed before, where members of the political party in control are more confident in their ability to express themselves freely. When power changes party hands, anxiety inverts across partisan lines. Whether this is driven by systemic suppression or perceived vulnerability is beside the point. This should never be a concern in the US.

Yet, nearly 75% of Americans believe the country is headed in the wrong direction. The vast majority of US adults feel that their freedom of speech is at risk.

Historically, we placed our trust in institutions to protect these rights. Publications, courts, and public forums are expected to operate independently of partisan control to safeguard open discourse. While our First Amendment remains intact, our faith in these institutions has eroded. We now question the durability of this right.

The collapse of institutional trust also explains the chaos of tracking the modern news cycle.

More information is generated each day than we have the capacity to consume. To filter high-signal content from the unruly sea of daily media, we rely on curators. From editorial boards to algorithmic feeds, we place our trust in curators and their ability to selectively promote content to paint an accurate picture of reality. Without trusted curators, we become lost, unable to make sense of the world around us.

The rise of independent media is a direct response to this collapse in trust. We have taken it upon ourselves to discover reputable individuals to subscribe to independently. We spend countless hours scrolling to find new voices, passing quick judgements to discern credibility, in an attempt to curate our own information. Yet, the result is unsustainable. Every day we receive dozens of notifications across disjointed platforms for more newsletters than we have time to read. The reward for our diligence is to pay for an unreasonable number of premium subscriptions.

We have rediscovered the need for curation from first principles. To find stability in this fragmented environment, we gravitate toward the voices with which we already agree. Rather than discover new perspectives, our existing beliefs are only reinforced.

The solution to both of these problems is to reestablish trust. We need to be able to trust that our First Amendment rights will not be infringed, and we need to be able to trust that our curators of information are providing us with an accurate representation of reality. And we are past the point of relying on faith in institutions.

Fortunately, we now have the ability to build trust through proof rather than faith.

Decentralization allows us to replace promises with verifiable protections. As a neutral information platform, Aemula does not ask you to trust our intentions. Instead, we make everything transparent and democratically governed. You don’t need faith in how we operate. You can directly observe and participate in it. If you disagree with our process, you can propose a change and enact it.

You no longer need to heuristically infer the credibility of sources through branding and subscriber counts when the reputations and track records of individual reporters are public and readily available for review.

All of these protections are written into the architecture itself. You do not need to believe that our platform will remain neutral. You can independently prove it.

And it was at this point in drafting this newsletter that I ran into an issue. Substack’s servers went down.

I couldn’t access this draft, and was unable to read our past spotlights to make the above references. While the outage only lasted 16 minutes, and Substack’s reliability has been strong over the past two years, the moment was indicative of a fundamental issue. Centralized platforms are a single point of failure. Whether it’s an accidental outage or a decision made at their whim, when the platform is unavailable, you do not have access to your own work.

Your ownership of your work depends entirely on access to a server that you do not own or control. You own nothing but a corporate promise for your continued access.

Decentralized architecture places ownership in your hands rather than relying on a centralized platform. With Aemula, your articles are stored on a peer-to-peer network and remain accessible even if our servers go offline. We simply provide a gateway to the network, but everyone is capable of establishing their own gateway to articles published on Aemula. Journalists have the ability to store and serve their own content (if you are writing on Substack and want to store and serve your own work, reach out writers@aemula.com and we will set you up with a gateway you control).

This is what we mean when we say that no single entity controls the flow of information on Aemula.

Taking it a step further, the entire Aemula Protocol runs on the Ethereum network. Even if we cease operations or attempt to wrest control from the community, the protocol persists in an open environment. You retain your data, your interactions, and your article recommendations. Journalists retain their reputations and ownership of their work. Everything remains open source and controlled by the community. Notably, Ethereum has had 100% uptime since launching in July of 2015. With no single point of failure, no one can restrict your access to information.

These same principles apply to your personal data, too.

Initially, we were not going to address the recent security breach at Substack. Nihilistically, email addresses and phone numbers are already out there for anyone who wants to access them, so the consequences were negligible. But if our intent is to expand individual ownership online, data security only becomes more important.

If you do not own your data, someone else does, and you are at the mercy of their security practices.

Decentralization does not eliminate the need for security, but it allows you to hold the keys. Rather than entrusting your information security to corporate databases that are keen to sell your data to the highest bidder, you can rest assured that your sensitive information is under your control.

In fact, there are many birds in which we can apply this one stone of decentralization.

To rebuild trust, reverse polarization, grant affordable access to high-quality information, and efficiently provide the resources necessary to produce professional reporting, we no longer need to rely on institutional gatekeepers. We can leverage new technology to shift to true independence and individual liberties.

The problems are complex, but the solution is simple.

This week, we highlight writers exploring the downstream effects of these foundational issues. We encourage you to explore their work and consider subscribing directly.

Expression

Written by the Foundation for Individual Rights and Expression (FIRE), a nonpartisan nonprofit that defends free speech rights for all Americans, previously featured in our spotlight, “Threat of Fractured Reality”.

Expression

Poll: Americans continue to say free speech is in decline

FIRE’s National Speech Index has been taking America’s temperature on expressive freedom every three months since January 2024, asking many of the same questions each run. That repetition is the point. When you keep the wording constant, shifts are harder to dismiss as noise. Patterns more readily stand out. And looking back, two notable trends stand ou…

Read more

4 months ago · 10 likes · 3 comments · Sean Stevens

“This is what you’d expect if “wrong direction” is partly a referendum on who’s running the executive branch. But the story can’t be reduced to just that. The liberal side’s volatility over the past few years indicates something beyond simple party-in-power logic and suggests that recent events have driven their perception of the country’s expressive environment. So while the color of the president’s necktie sets a baseline, it doesn’t lock perceptions in place. Events and institutions beyond the presidency also plausibly shape how safe people feel speaking freely.”

Studio Self

Written by JA Westenberg, founder of Sydney-based creative agency Studio Self, distilling years of experience shaping how technology leaders communicate to offer strategic insight on narrative, authority, and building brands that stand out in fast-moving markets.

Studio Self

The Gutenberg Parenthesis is Closing

For roughly 570 years, humanity ran an experiment in information transmission: literacy. The written word and the document, the careful and structured transfer of ideas from one mind to another through the precise arrangement of symbols on a surface…

Read more

4 months ago · 152 likes · 14 comments · JA Westenberg

“Text is a compression format. To convert an idea from its native form (the messy, associative, emotionally-tinged, contextually-embedded pattern of neural activity that constitutes human thought) into writing, you have to squeeze it through a narrow channel. You have to linearize what isn’t linear, make discrete what isn’t discrete, strip out tone and gesture and facial expression and shared context, and hope the person on the other end can decompress it accurately.

Most of the time, they can’t. Most writing fails, and most readers misunderstand most texts most of the time. The utter inability of people on Threads to parse satire illustrates this well. The elaborate apparatus of scholarly commentary that interprets great works of literature shows how poorly even the finest writing conveys meaning across time and context. We’ve been trained to see this compression as a feature (it forces clarity! it enables permanence! it scales infinitely!) rather than a bug, but it’s possible we’ve been rationalizing for half a millennium.”

Thomas Peermohamed Lambert

Eponymously written by Thomas Peermohamed Lambert, a London-born writer and Oxford Clarendon Scholar whose debut novel Shibboleth was published by Europa Editions in 2025, dividing time between fiction, journalism, and academic work in literature and culture.

Thomas Peermohamed Lambert

Against Literacy

You wake into the weekday darkness, and reach for the portal you keep beside your bed. A few minutes can’t hurt. There are so many beautiful essays out there, and they all look so appealing, all those orange blobs waiting to be punctured. With the chewiest part of your thumb, you click on one…

Read more

4 months ago · 51 likes · 4 comments · Thomas Peermohamed Lambert

“If you want to write well in 2026, you need to figure out what the prevailing incentives of the form or medium you write in are, and violently reject them. The best writing, after all, is like a kind of parasite: it is hosted by the technology of writing with all its unfortunate tendencies, but it is not of that technology. It does not conform to what writing wants of it. One of the most effective ways of achieving this effect is to do exactly what Plato did: to invent a kind of hybrid form, where the abstract rigours of writing are tempered by the rhythms of the spoken word. A philosophical dialogue, for instance; or a nice big realist novel. But there are also countless other, stranger ways. If I had to put the general procedure into words, I might say something like: ‘the antidote to literacy is literature – that rare kind of writing that cleaves apart the language-object from its referent, and thereby destabilises the thetic relation between signifier and signified in a way that is productive of new meanings.’ But see? It doesn’t work. I’ve lapsed into jargon. Horrible, Derridean, overliterate stodge. You can hardly blame me, though: all I have to work with are the tools of the enemy.”

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

“This ranks among the darkest days in the history of one of the world’s greatest news organizations. The Washington Post’s ambitions will be sharply diminished, its talented and brave staff will be further depleted, and the public will be denied the ground-level, fact-based reporting in our communities and around the world that is needed more than ever.”

— Marty Baron, former editor of The Washington Post

Yesterday, The Washington Post announced layoffs of 300 journalists, roughly a third of their newsroom, gutting and reshaping their sports, metro reporting, books section, and foreign coverage desks. Lizzie Johnson, the Post’s Ukraine correspondent covering the war, discovered she was laid off while stationed in an active war zone (The Washington Post Guild is running a fundraiser to support the journalists affected by these layoffs, and we encourage you to consider a donation). While surely a shocking announcement to those involved, the news of these layoffs was dulled by the tired narrative of shrinking newsrooms.

These layoffs are the latest in a wave of consolidation in the industry. As legacy publications struggle to compete for attention in our modern digital media environment, they fight for survival through consolidation. As they cut costs, we are left with fewer voices reporting the news we read.

Among the fallout, Substack and Beehiiv have made commendable efforts to support the journalists affected by the layoffs. As these platforms have proven, they provide critical tools for journalists to build independent brands, freeing them from staking a career on the shaky foundations of legacy media institutions.

The success of Substack and Beehiiv has demonstrated the viability of the industry-wide shift towards independence, as they now support millions of readers and tens of thousands of writers. However, they are unable to truly fill the void of professional-grade newsroom support resources traditionally provided by institutional publications.

As Tyler Denk, co-founder and CEO of Beehiiv, puts it:

“unfortunately the types of important journalism that requires months-long investigative work, or being stationed in dangerous conflict zones isn't super conducive to this new independent media model (at least not yet). this is where the larger media organizations really shine.

I'm self aware enough to acknowledge that beehiiv, as a nascent technology platform focused on building great product and a platform for independent voices to flourish, cannot possibly offer the security clearances or assurances of a NYT/WaPo to properly support those journalists in active war zones.

I also think it's naive and slightly dishonest of substack to think otherwise”

And we agree with Tyler’s sentiment, with the exception of one key statement — “at least not yet”.

We currently have the ability to solve these problems. Journalists no longer need to accept the tradeoff between independence and institutional-grade support resources. This is exactly the role we aim to fill with Aemula, which we outline in our full response to Tyler’s post:

The choice between individual ownership and access to support resources is a false dichotomy. It was merely a technical limitation that has since been solved. The remaining challenge is one of adoption and scale, and we have never been more confident in the ability to solve this challenge.

Marc Andreessen, in his interview with Packy McCormick on the a16z podcast, discusses the “enormous latent demand” for high-quality media. “The issue is not lack of demand, it is lack of supply”, he acknowledges in addition to the belief that there is potential to increase the size of the existing media news industry by 1000x in terms of collective value.

And it is clear that the belief in this opportunity is no secret. Substack raised $100m last year to grow to “internet scale”, Beehiiv is hoping to double annual revenue to $50m in 2026 in their challenge to Substack’s throne, and X is offering a $1m prize to the top article published on the platform in their effort to promote long-form content.

Where we diverge is in our belief that decentralized, independent journalism is the only way to achieve the scale and efficiency necessary to incentivize the creation of enough high-quality news to meet this latent demand in the industry. Specifically, the type of reporting that requires the months-long investigative work and access to professional resources that Tyler notes Beehiiv and Substack are unable to provide.

Journalism is often viewed as a dying industry only because we have lost our faith in legacy media’s ability to succeed. Their failed attempts to force an antiquated business model onto the modern world only support this theory further. Yet, it is only the institutions that are crumbling. Tens of thousands of journalists work tirelessly to report the news we read, and they deserve to be rewarded in line with the immense value they create, while having access to the resources necessary to produce high-quality reporting. The opportunity for the industry to restructure into a modern model has never been greater.

We have a new model for media. It is time we leverage it to its full potential.

This week, we highlight writers discussing the recent layoffs at The Washington Post and their impact on the industry as a whole. We encourage you to explore their work and consider subscribing directly.

Glenn Kessler

Eponymously written by Glenn Kessler, a longtime Washington Post reporter and former Fact Checker whose Pulitzer Prize–winning career and bestselling books inform his clear-eyed analysis of truth, politics, and public discourse.

BY GLENN KESSLER

I worked at The Washington Post for 28 years. Jeff Bezos just destroyed it.

I wrote an opinion article for MSNOW about the devastating cuts — one third of the journalists — announced at The Washington Post yesterday. Please follow this link to read my analysis of a sad day in American journalism. This Substack is reader-supported. To receive new posts and support my work, consider bec…

Read more

4 months ago · 24 likes · 7 comments · BY GLENN KESSLER

“I spent almost a decade as The Washington Post’s diplomatic correspondent — one of the key national security beats at any news organization. But many of my biggest scoops couldn’t have been written without the assistance of Post foreign correspondents on the ground, journalists with their own sources and expertise. When I jaunted around the globe with the U.S. secretary of state, trips often included a stop in a capital where a Post correspondent was based.

That combination of resources and ambition to report news far and wide, without fear or favor, helped make The Washington Post a great news organization, until Wednesday, when Post leadership announced cutbacks so severe and strategy changes so shortsighted that not just the Post but American journalism — and democracy — will be left diminished.”

Silver Bulletin

Written by Nate Silver, a statistician and renowned election forecaster, professional poker player, and author of The Signal and the Noise, previously featured in our spotlight “Divergence”.

Silver Bulletin

The sad and self-inflicted decline of the Washington Post, in one chart

UPDATE Feb. 4: Today, the Washington Post announced that it was laying off more than a third of its staff. This is a truly devastating round of cuts. The entire sports staff has been let go or repurp…

Read more

4 months ago · 348 likes · 96 comments · Nate Silver

“The fracturing of the media ecosystem has produced winners (including Substack, certainly) as well as losers. And outlets like the Wall Street Journal frequently provide excellent political coverage, in addition to their coverage of financial markets. But local and regional news coverage has been hollowed out. It would be good to have at least a handful of truly muscular national news outlets to complement all the niche publishers and opinion sites. We may be increasingly heading to the point where there are just one or two truly comprehensive national news brands, and a long tail of everything else.”

SmallBites

A publication offering concise, no-frills analysis designed to deliver sharp, immediately useful insights for readers seeking clarity, urgency, and actionable understanding without wasted words.

Small Bites

Washington Post: A Newspaper Gutted, a Trust Broken

Small Bites is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber…

Read more

4 months ago · 2 likes · 4 comments · SmallBites

“The silence of leadership weighed heavily. The Post’s publisher, Will Lewis, did not speak during the meeting. One laid-off editor, speaking anonymously, said Lewis’s legacy would be defined by “having enabled Bezos to tank an American institution.”

For years after Bezos purchased the Post in 2013, the newsroom expanded and ambitions grew. That era has clearly ended. Buyouts in late 2023 and 2024 thinned the ranks, and previous layoffs hit technology, advertising, and print operations. But Wednesday’s cuts struck at the journalistic core.

Over the past week, staffers pleaded directly with Bezos, signing letters and posting public appeals that emphasized the civic importance of the Post’s reporting. Those messages went unanswered. Bezos did not respond to letters from the foreign, local, or White House teams.”

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

Ten years after Andy George, the creator behind How to Make Everything, went viral for spending $1,500 and 6 months to make a sandwich from scratch, he returns to his sandwich-making career. This time, Andy takes his experiment a step further, deciding to make all of the tools necessary for making a sandwich from scratch as well. In this second installment of the sandwich series, Andy collects ores to forge knives, blows glass to make jars for pickling cucumbers, carves stones to mill grain, and even builds a homemade flintlock rifle to hunt a feral hog for bacon.

Andy’s 10-year journey to make a sandwich all on his own highlights the extensive interconnectedness of our global economy, providing us with a modern-day manifestation of Leonard Read’s famous I, Pencil essay. As both Andy George and Leonard Read show, the seemingly simple products we take for granted in our daily lives, such as sandwiches and pencils, are the result of incomprehensibly complex supply chain networks. Billions of individuals seamlessly self-coordinate to sustain our modern way of life.

In New York City today, you can exchange 49 minutes of your time, after taxes while working at minimum wage, to be able to afford a sandwich similar to the one Andy George makes in his video. Comparatively, Andy spent $7,560 for materials and ingredients and devoted 460 hours of his time.

While this is not a scientifically perfect comparison, the anecdotal evidence is representative of just how much we rely on one another. As a global economy, we are operating near the limits of efficiency in producing goods and services. Despite its complexity, this system is impressively adaptive and resilient. However, any disruption to this global machine could prove detrimental to our modern way of life.

Yet, we are witnessing a worrisome trend as we revert from globalization towards nationalism, and experience increasing political polarization domestically. In our age of information, we have more widespread access to knowledge and social connection than ever before in human history, but we are finding it increasingly difficult to communicate across ideological divides. We risk losing our ability to collaborate effectively at societal scale.

This may be getting too abstract, but hear us out…

We all work together to achieve things that no single person could possibly achieve. Humanity is a prime example of emergence, the concept that many simple things can work together to create something that exceeds the sum of its parts.

Our global economy is incredibly complex, but this complexity generates a growing excess of resources that increases the quality of life for people around the world. As we move forward, we are sure to experience increasingly complex problems. Innovation is not inevitable. We must constantly work together to solve new problems as they arise.

Our ability to problem solve is a function of our collective intelligence. People receive, process, and share information to collectively work on large-scale projects. We operate as one large neural network. The power of this network is only limited by the number of people and the ease at which they can communicate.

Our focus with Aemula is on communication.

To date, nearly all of [society’s] advancements in communication have been focused on increasing the speed at which two people can communicate. We can now transfer information at the speed of light. We are nearing the limits of efficiency in this regard.

However, we have neglected to devote significant resources to determining how to route information effectively. Yes, two people can communicate instantaneously, but how are we ensuring that the information is flowing to the correct people along the shortest path? Currently, that flow of information is restricted by our centralized distribution systems and controlled by conflicting incentives.

We believe information is our greatest resource. Currently, it is not being allocated correctly.

This is the problem that we must solve.

Excerpt from the original Aemula letter dated 3.2.24

This week, we highlight writers discussing the cracks that are beginning to form within our channels of communication and how they are affecting our ability to work together at scale. We encourage you to explore their work and consider subscribing directly.

Conspicuous Cognition

Written by Dan Williams, an academic philosopher with a PhD from the University of Cambridge, a lecturer at the University of Sussex, and an associate fellow at the Leverhulme Centre for the Future of Intelligence at the University of Cambridge, previously featured in our Spotlight, “The Truth”.

Conspicuous Cognition

The world outside and the pictures in our heads

This corresponds to the first topic and lecture in my seven-week ‘Politics, truth, and ideology’ course…

Read more

a year ago · 256 likes · 28 comments · Dan Williams

“However, the ultimate reason for Lippmann’s pessimism is less the undeniable fact of our epistemic interdependence—our reliance on testimony and an unimaginably complex and fragile division of labour—than the second filter mediating between the real environment and the pseudo-environments of journalists and citizens: interpretation.

The environment is not only ‘too big, too complex, and too fleeting for direct acquaintance’. Its vastness and complexity also mean that we must ‘reconstruct it on a simpler model before we can manage it.’”

The Power of Us

Written by Dr. Jay Van Bavel, a Professor of Psychology and Neural Science and Director of the Center for Conflict and Cooperation at NYU, and Dr. Dominic Packer, a Professor of Psychology and Associate Vice Provost for Research at Lehigh University, with Yvonne Phan, a researcher and science communicator serving as managing editor of The Power of Us newsletter.

The below post was drafted by Daniel Thilo Schroeder and Jonas Kunst, with edits from Jay Van Bavel, based on their recent paper.

The Power of Us

When AI Can Fake Majorities, Democracy Dies Quietly

Imagine you’re doomscrolling through your social media feed. A political controversy breaks—and within minutes, it feels like a tidal wave of commentary. Thousands of “ordinary people” pile on, repeating a theme, sharing links, and “liking” each other’s posts while drowning out dissent…

Read more

4 months ago · 18 likes · Dominic Packer & Jay Van Bavel, Daniel Thilo, and Jonas R. Kunst

“Democracy doesn’t require perfect truth—but it does require something more fragile: independent voices. The ‘wisdom of crowds’ depends on independence between judgments. If a single actor can speak through thousands of inauthentic accounts, the apparent consensus of the crowd stops being informative.

One pathway of harm is synthetic consensus: creating the illusion of a majority opinion. Swarms can seed narratives across niches and amplify them via coordinated liking, replying, and cross-posting until it looks like broad support. People update beliefs partly through social evidence—what seems normal, common, or widely endorsed. Synthetic consensus exploits that cognitive shortcut.

A second pathway is segmented realities. Because swarms can mimic local language, emotion, and identity cues, they can tailor narratives community-by-community, reinforcing polarization and making cross-group cooperation and consensus harder.”

Into the Machine

Written by Tobias Rose-Stockwell, a designer, technologist, and media researcher exploring the effects of social media on our morals, society, and democracy, with work previously featured in The Atlantic, Wired, BBC, and NPR.

Into The Machine

Why We Can See Everything and Agree on Nothing

There's a pothole on my street. I know this because I ride over it every day, muttering curses at the city and making mental notes to finally submit that complaint. I also know, with equal certainty, that a neighbor's oak tree dropped a branch during last week's storm…

Read more

a year ago · 18 likes · Tobias Rose-Stockwell

“Democracy is a big coordination game. We need to know what’s happening in order to vote people into office that can solve our problems. In this way the democratic process is a sort of problem-solving machine — if you’re angry about something you see in your community, you should be able to vote for change. If that thing is fixed by that representative, great! Vote them in again. If not, vote them out and try again with a different guy.

For local issues this is pretty easy, because there are fewer mediated steps between us and the issue. Like the pothole on my street, the observable distance is still very low. School boards and city governments function pretty well because people can see what is wrong and which policies are failing.

But for national issues, most of our problems are not directly observable. We rely on mediators — journalists, scientists, pundits, influencers, politicians, and our smart friends — to help us figure out which issues we should pay attention to and how we should vote. The internet was supposed to help us here”

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.

While this holds true for our trivial daily decisions, it also dictates how people choose to participate in public discourse. We only share our thoughts if we expect them to have an impact. Otherwise we would just be shouting into the void (though sometimes void shouting is cathartic).

The work it would take us to share our thoughts also can’t be too difficult. Even on social media, where the bar for sharing our perspectives is on the floor, most of us choose to lurk, for sound reason. Even if anyone can create an account and start posting for free, financial rewards are only available for the select few who achieve influencer status. Coupled with the ever-present social risk of etching your bad or embarrassing takes in the annals of digital history, continuously scrutinized by internet sleuths and the judgement of your loose acquaintances, the rewards rarely outweigh the risks.

To plagiarize ourselves, “only those who are passionate enough about an uncommon belief are willing to commit the time to posting on social media without pay. The result is the elevation of fringe voices, further radicalized by algorithms that are fundamentally geared towards generating controversy.”

If we want to see more high-quality perspectives reflected in our media environment, we need to increase the incentives for contribution while lowering the barriers to entry. We need to cultivate an ecosystem that encourages original, in-depth reporting. We need to know that pressing publish is actually worth our time and effort.

But accurately reporting the news is hard work. Discovery, research, verification, drafting, and revision are cognitively expensive tasks. Doing it well requires sustained attention and a deep willingness to invest time without any guarantee of a payout. Journalists are currently trapped in a persistent tradeoff between working within institutions that provide support resources in exchange for limited autonomy, or going independent to gain full creative control while accepting significant personal risk.

To successfully and sustainably go independent relies on building credibility, attracting attention, converting paid support, and assembling reporting infrastructure, all while also doing the hard work of the reporting. Most people, even highly capable ones, rationally decide that these costs outweigh the potential reward, especially when the success stories are so sparse. Additionally, one’s ability to report on their topic of expertise is rarely correlated with their ability to build a brand and business from scratch.

Our goal with Aemula is to eliminate the tradeoff that journalists face. We want to de-risk the decision of going independent by providing publication infrastructure, distribution to an existing network of paid subscribers, and professional resources for research and collaboration, all while empowering writers to retain full ownership and creative control of their work.

With strong incentives and low friction, more people with real expertise will choose to contribute their perspectives to the public discourse. With more diverse, independent perspectives, we all will be able to build a more accurate understanding of the world around us.

Journalists no longer have to choose between support and freedom. When we say that Aemula is a platform for independent journalists, we see a future where every journalist is independent.

Writing? Publish on Aemula

This week, we highlight writers discussing the mechanics at play behind the scenes that affect the quality and types of content we consume. We encourage you to explore their work and consider subscribing directly.

The Prism

Written by Gurwinder, a British-Indian author whose work has been featured in The Free Press, Areo, Quillette, The Humanist, and The Sunday Express, and previously featured in our spotlight, “Time Theft”.

The Prism

Why You Are Probably An NPC

“The enemy is the gramophone mind, whether or not one agrees with the record that is being played…

Read more

3 years ago · 2923 likes · 285 comments · Gurwinder

“the brain is commonly regarded as a thinking machine, but it’s more often the opposite: a machine that tries to circumvent thinking. This is because cognition costs time and calories, which in our evolutionary history were scant resources.

As such, the brain evolved to be a “cognitive miser” that operates according to the principle of least effort, taking shortcuts in thinking and perceiving that build a workable but hugely simplified (and cost effective) model of the world.”

Experimental History

Written by Adam Mastroianni, an experimental psychologist with a PhD from Harvard, with publications in The New York Times and The Atlantic, and covered by many top blogs and publications, previously featured in our spotlight, “Complexity”.

Experimental History

Text is king

The hot new theory online is that reading is kaput, and therefore civilization is too. The rise of hyper-addictive digital technologies has shattered our attention spans and extinguished our taste for text. Books are disappearing from our culture, and so are our capacities for complex and rational thought. We are careening toward a pos…

Read more

4 months ago · 487 likes · 54 comments · Adam Mastroianni

“The hot new theory online is that reading is kaput, and therefore civilization is too. The rise of hyper-addictive digital technologies has shattered our attention spans and extinguished our taste for text. Books are disappearing from our culture, and so are our capacities for complex and rational thought. We are careening toward a post-literate society, where myth, intuition, and emotion replace logic, evidence, and science. Nobody needs to bomb us back to the Stone Age; we have decided to walk there ourselves.”

New Science

Written by New Science, a research nonprofit dedicated to advancing basic science, examining how scientific research works in practice by profiling the people, tools, and institutions shaping the future of life science, with the below piece written by Roger’s Bacon, co-founder and head editor of Seeds of Science, a journal dedicated to nurturing independent, exploratory thought at the frontiers and fringes of science.

New Science

Research Papers Used to Have Style. What Happened?

The advocates of the new science in the seventeenth century so reacted against the excesses of stylistic artistry that a reluctance to use any artistry at all seems to have prevailed ever since…

Read more

4 years ago · 89 likes · 12 comments · Roger’s Bacon

“Selection improves the average fitness of a population, but drains it of diversity and limits its potential for future adaptation. Mutation increases diversity, but lowers the average fitness of the population.

Like mutation, aesthetics are diversifying and generative, but generally harmful to clarity and concision. And like selection, restrictions on style and format may improve the average quality of our writing, but at the cost of creative potential.”

Are you writing on Substack? You can easily set up automatic cross-posting with Aemula to instantly:

• Increase your earnings

• Expand your audience

• Verifiably own your work

Plus, you will have opportunities to access community resources and grants to support the content you want to create!

Link your Substack to your Aemula account using this link or reach out to writers@aemula.com to get started!

No cost, no obligations, and you can stop at any time.

If you want to support any of the writers we spotlight in our Substack, we highly encourage you to subscribe to their individual publications.

If you want to support independent journalism more broadly, create an account on the Aemula platform.

aemula.com

To stay up to speed on platform updates and the writers we are adding to our community, Follow us on X or subscribe to our Substack!

Any writers you want to see featured here? Send them our way! We are always searching for great new publications.