Greetings from Read Max HQ! This week, a collection of thoughts about a new trend in tech criticism masquerading as a lumpy and overstuffed essay.

A reminder: This piece, and all the pieces you read on the Read Max newsletter, is funded almost entirely by paying subscribers of this newsletter. I am able to write these columns and record the podcasts and videos thanks to the support of nearly 4,000 people who value what I do, for whatever strange reason. Unfortunately, because of the reality of subscription businesses, I need to keep growing in order to not shrink, which means every week I need to beg more people to sign up. Do you like Read Max? Do you find it entertaining, educational, distracting, fascinating, or otherwise valuable, such that you would buy me a cheap-ish beer at a bar every month? If so, consider signing up for the low price of $5/month or $50/year

Subscribe now

A new wave of techlash

The new moderate-liberal Substack publication The Argument ran a fascinating piece by civil rights attorney (and Tottenham blogger) Joel Wertheimer last week arguing that policymakers should “Treat Big Tech like Big Tobacco”:

The problem with Big Tobacco was not that it could charge excess prices because of its market power. The problem with Big Tobacco was that cigarettes were too cheap. Cigarettes caused both externalities to society and also internalities between the higher-level self that wanted to quit smoking and the primary self that could not quit an addictive substance. So, we taxed and regulated their use.

The fight regarding social media platforms has centered around antitrust and the sheer size of Big Tech companies. But these platforms are not so much a problem because they are big; they are big because they are a problem. Policy solutions need to actually address the main problem with the brain-cooking internet.

Wertheimer argues that the famous Section 230 of the Communications Decency Act--which protects companies from liability for content posted by users to their websites--needs to be reinterpreted to exclude “platforms that actively promote content using reinforcement learning-based recommendation algorithms.” I’m not exactly qualified to weigh in on the legal questions, but I find the logic of the argument persuasive in its broad strokes: The idea is that while message boards and blog comment sections--which host third-person speech but do nothing active to promote it--deserve Section 230 protection, platforms that use algorithmic recommendations (i.e. Facebook, Instagram, TikTok, and X.com) are not simply “passively hosting content but actively recommending” it, an act that should be considered “first-person speech” and therefore subject to liability claims.

But what really strikes me about Wertheimer’s piece is the public-health metaphor he uses to explain the particular harms of social-media platforms (and that, in turn, justify his remedy). The contemporary web is bad for us, the argument goes, in the way cigarettes are bad for us: Cheap, readily available, highly addictive, and making us incredibly sick at unbelievably high cost.

In this, Wertheimer is following a line of argument increasingly prominent among both pundits and politicians. In April, David Grimes made a less policy-focused version of the same argument in Scientific American; just last month, speaking with Ezra Klein on his podcast, Utah Governor Spencer Cox drew on both Big Tobacco and the opioid industry:

The social graphs that they use, which know us better than we know ourselves, that allow us, as you so eloquently stated and better than I could, to understand what makes us emotional and what keeps our eyeballs on there — so that when a kid is somehow, even if they don’t want to be, on TikTok at 3 a.m., just going from video to video, and they’ve given up their free will — that is unbelievably dangerous.

When tobacco companies addicted us, we figured out a way out of that. When opioid companies did that to us — we’re figuring our way out of that. And I’m just here to say that I believe these tech companies, with trillion-dollar market caps combined, are doing the same thing — the same thing that tobacco companies did, the same thing that the opioid companies did. And I think we have a moral responsibility to stand up, to hold them accountable and to take back our free will.

A few days after Wertheimer’s piece, Abundance author Derek Thompson posted a podcast interview of Massachusetts Representative Jake Auchincloss, who has proposed a digital value-added tax designed, like Wertheimer’s proposal around Section 230, to internalize the costs of social media. In his introduction, Thompson directly compared the digital V.A.T. to “sugar taxes and cigarette taxes”:

Massachusetts Congressman Jake Auchincloss has a proposal that he calls a digital sin tax, a way to push back on the business model of social media platforms that profit from hijacking our attention, especially our kids’ attention.

You’ve heard of sugar taxes and cigarette taxes. Well, this would be an attempt to price the harms of the attention economy and route the proceeds to public goods. I think it’s an interesting idea.

Comparing Big Social to Big Tobacco (or Big Opioid or Big Sugar) is in some sense a no-brainer, and certainly such analogies have been drawn many times over the last few decades. But the increasing popularity of this conceit is less a coincidence, I’d argue, than a function of the gathering power of a new wave of the now-decade old “techlash.”

Subscribe now

This burgeoning movement seeks to root criticism of (and response to) Big Tech in ideas of health (public, social, intellectual, and spiritual) and morality rather than size and power, positioning the rise of social media and the platform giants as something between a public-health scare and a spiritual threat, rather than (solely) a problem of political economy or market design. I see versions of this school of thought not just in speeches and op-eds from Auchincloss and Cox or blog posts from Thompson, but in Chris Hayes’ book The Siren’s Call and in the inescapable work of Jonathan Haidt. (You might broadly think of Hayes and Haidt as representing “left” and “right” tendencies of the broader movement.) Notably, all of the above-mentioned have found platforms on Klein’s podcast. Back in a January interview with Hayes, Klein offered up a kind of political vision or prediction rooted in this tendency:

I think that the next really successful Democrat, although it could be a Republican, is going to be oppositional to [the tech industry]. In the way that when Barack Obama ran in ’08 — and I really think people forget this part of his appeal — he ran against cable news, against 24-hour news cycles, against political consultants.

People didn’t like the structure and feeling of political attention then. And I don’t think there was anywhere near the level of disgust and concern and feeling that we were being corroded in our souls as there is now.

And I think that, at some point, you are going to see a candidate come up who is going to weaponize this feeling. They are going to run not against Facebook or Meta as a big company that needs to be broken up. They’re going to run against all of it — that society and modernity and politics shouldn’t feel like this.

And some of that will be banning phones in schools. It’ll have a dimension that is policy. But some of it is going to be absolutely radiating a disgust for what it is doing to us and to ourselves. I mean, your book has a lot of this in it. I think that political space is weirdly open, but it seems very clear to me somebody is going to grab it.

Thompson dubs this loose movement, or at least the version touted by Auchincloss, “touch-grass populism,” but I think this is wrong: The framework in question is distinctly not “populist” (unlike, say, the neo-Brandeisian “new antitrust” movement that has been a major focus of the “techlash” to date) so much as progressive in the original sense, a reform ideology rooted in middle-class concerns for general social welfare in the wake of sweeping technological change. At its broadest you could maybe call this budding program of restriction, restraint, and regulation “Platform Temperance,” and regard the scattered campaign to ban smartphones in schools as its first stirrings as a movement.

Why Platform Temperance now?

One way of thinking about the past half-decade or so of life on the internet is that we’ve all become test subjects in a grand experiment to see just how bad “good enough” can be. Since Elon Musk’s purchase of Twitter in 2022, and the subsequent industry-wide cutbacks to “trust and safety” teams meant to moderate content, most of the major social platforms have been flooded with fraud, bait, and spam--a process exponentially accelerated by the arrival of ChatGPT and its generative-A.I. peers. Take, e.g., these YouTube ads discovered by BlueSky user Ken Plume:

As I tweeted at the time, I’ve been covering tech companies for years and I still find myself taken aback at how completely they’ve abdicated any kind of oversight or moderation. Setting aside the increasingly toxic and directly corrosive “politics” now inescapable on social platforms, Facebook and Instagram and YouTube are utterly awash in depressing low-rent non-political slop, and no one who owns, runs, or even works at these platforms seems even to be embarrassed, let alone appalled.

But why would they be? People working at tech giants are watching the metrics and seeing that the depressing low-rent slop is getting engagement--probably even to a greater extent than whatever expensive, substantive, wholesome content it’s being placed next to on the feed. Their sense, backed up by unprecedentedly large data sets, is that slop of various kinds is what people want, because it’s what they click on, watch, and engage with. (I would even go so far as to suggest that some portion of Silicon Valley’s broad reactionary turn since 2020 can be chalked up to what I think of as “black-pilling via metrics”: The industry’s longtime condescension toward its users finally curdling into outright contempt.)

For much of the past decade, this revealed preference for fake news, engagement bait, sexualized content, and other types of feedslop has been blamed on “platform manipulation”: Bad, possibly foreign actors were “manipulating” the platforms, or, worse, the platforms themselves were “manipulating” their users, deploying “dopamine feedback loops” and “exploiting a vulnerability in human psychology,” as Sean Parker said back in 2017.

But these accounts have never been wholly satisfying: Too technical, too determinist, too reliant on the idea that there is “authentic” or “innocent” desire being “manipulated.” In some sense they don’t blame we, the users, enough, or assign us the kind of agency we know we have. Most of us are aware from everyday experience that even absent “manipulation” we desire all kinds things that are bad for us, and that we give in or restrain from temptation based on any number of factors.

Subscribe now

This, I think, is the basic dynamic from which Platform Temperance evolves: a general, non-partisan, somewhat moralistic disgust at even the non-political outcomes of unregulated and unmoderated platforms; a dissatisfaction with both the “revealed preference” framework and the more rigidly behavioralist explanations of platform activity; and a sense of an accelerating downward spiral with the advent of generative A.I., which both further debases the platforms and provides a possibly even more tractable and dangerous user experience itself. In response, Platform Temperance offers a focus on health, social welfare, and the idea of discipline and restraint in the face of unmoderated consumption--that is, temperance.

The politics of platform temperance

There’s another aspect to mention here. Platform Temperance as it has evolved recently seems to be largely a school of thought from the (broad) center--adjacent, in its membership and institutional affiliations, to the “Abundance” faction of elite politics.

There’s an obvious cynical reading here: As Klein says, this “political space is weirdly open,” and “somebody is going to grab it.” Thompson’s podcast interview with Auchincloss is framed around the idea of Platform Temperance (or “Touch-Grass Populism”) as a “big idea” around which moderates can rally:

Winning elections has to be more about articulating a mission than arriving at the midpoint of every possible debate… The center has not been nearly as successful [as the Trumpian right] at pouring itself into truly bold visions of a better future. So today we’re going to talk about one very big idea from a politician who doesn’t maybe neatly fit into any particular Twitter tribe. […]

I think [digital V.A.T.] a somewhat problematic idea. But more than interesting and problematic, I think it’s a big idea. And I’m excited that someone in the political center is offering it.

It’s easy, given this kind of positioning, to read Platform Temperance as a new front in an ongoing factional war within the Democratic party--and, indeed, the tendency is often pointedly positioned against the more populist and anti-establishment New Antitrust movement that has been among the most prominent strains of the Techlash thus far.

Subscribe now

But while the political valence is important for context, I don’t think Platform Temperance is wholly (or even mostly) a cynical Trojan Horse for intra-Democrat political battles. Versions of the ideas, remedies, and rhetoric emerging from the big tent that I’m calling Platform Temperance have been circulating in the Techlash for many years, usually from left-wing critics and academics. I’m thinking here of, e.g., James Bridle’s A New Dark Age, Jenny Odell’s How to Do Nothing, and Richard Seymour’s The Twittering Machine, three excellent books that take psychosocial and psychoanalytic approaches to the problems posed by platforms. In the early days of this newsletter I myself made a kind of proto-Platform Temperance argument under the headline “Maybe we need a moral panic about Facebook”:

I’ve written a lot about Facebook and its peers over the last half decade or so, and I can say pretty conclusively that people are, in general, much less interested in structural accounts, no matter how rigorous and explanatory, than they are in writing about the affective dimension of tech power. “Structure” is abstract and difficult to touch; feelings are immediate and intuitive. Readers want help articulating how life on the platforms makes them feel, and why.

I think the change-vs.-continuity debate, focused as it is on Facebook’s role in the political and media spheres, can miss entirely that the affective critique of Facebook — Facebook is bad because it makes people feel bad — is the most powerful. That’s not to say it’s empirically rigorous, or that it’s wholly new, or that it’s not a way to blame the fundamental dynamic of capitalist alienation on a new technology. But if the point is to transform Facebook from something that works on us to something that works for us — and, barring that, to shut it down — it’s useful to remember what people hate about it.

I still believe--as Klein does--that there is a lot of political power in harnessing people’s deep ambivalence about (or outright disgust with) a platform-mediated social, cultural, and political life. And I often find myself eager to reach for public-health metaphors when discussing the experience of life under the thumb of the software industry, if not outright spiritual ones. (I don’t believe in the “soul,” but I am hard-pressed to think of a better way to succinctly describe the effects of, say, TikTok than to say it’s bad for your soul.)

But I also want to be conscious of how easy it is for this kind of rhetoric to slip into reactionary moral panics. As David Sessions recently wrote:

My working hypothesis is that neo-atomization discourses have formed a knot of moral panic about technology that is being used as a Trojan horse for social conservatism. In some quarters, like the drafters of Project 2025 or Michigan’s outrageous proposed ban on porn and VPNs—that revived social conservatism is simply the usual Christian-right suspects doing their thing. But even among Christian social conservatives, explicit appeals to God and natural order have been replaced by a pseudo-public health language of “epidemics,” “addiction,” and social breakdown. What’s perhaps even worse is the extent to which essentially the same ideas have conquered liberal discourse in places like the New York Times op-ed page; a liminal figure like Christine Emba stands as something of a bridge between the two.

Emba’s book Rethinking Sex is a perfect example of this: you don’t even have to say that porn is immoral and casual sex is damaging (though she does that, too); you can say technology is alienating; apps are disrupting our natural, healthy forms of relation and association, harming women and children; our sex and relationships are dehumanizing and socially corrosive because we basically treat them like ordering DoorDash. You can even throw in some superficial anti-capitalism to give it a progressive spin. I’m not sure people even realize how dominant these views have become among liberals and how, in the absence of true ethnographic curiosity about the forms of life the social web has created, authoritarian policy responses are starting to sound like bipartisan common sense. […]

Here are just a few reasons contemporary tech panic needs to be pressed further than it often is: social scientific research about the “loneliness epidemic” is in fact highly contested, as are linkages of depression and poor mental health to phones and social media. “Porn addiction” is a made-up, faux-medical rebrand of classic right-wing evangelical ideology, and the evidence that porn is all somehow “increasingly” violent and misogynistic is flimsy enough to basically constitute a folk myth, no matter how many times it’s repeated in the New York Times.

I think Platform Temperance as an affective and political framework has a lot to offer tech skeptics, critics, and “the left” more broadly. But like the progressive movements that emerged in the late 19th century, it can produce both grounded, persuasive, important liberal-technocratic visions, and paternalistic, pseudoscientific moral panics. We should be careful about which we’re pursuing.

Chimpanzees revise their beliefs if they encounter new information, a hallmark of rationality that was once assumed to be unique to humans, according to a study published on Thursday in Science.

Researchers working with chimpanzees at the Ngamba Island Chimpanzee Sanctuary in Uganda probed how the primates judged evidence using treats inside boxes, such as a “weak” clue—for example, the sound of a treat inside a shaken box—and a "strong" clue, such as a direct line of sight to the treat. 

The chimpanzees were able to rationally evaluate forms of evidence and to change their existing beliefs if presented with more compelling clues. The results reveal that non-human animals can exhibit key aspects of rationality, some of which had never been directly tested before, which shed new light on the evolution of rational thought and critical thinking in humans and other intelligent animals.      

“Rationality has been linked to this ability to think about evidence and revise your beliefs in light of evidence,” said co-author Jan Engelmann, associate professor at the department of psychology at the University of California, Berkeley, in a call with 404 Media. “That’s the real big picture perspective of this study.”

While it’s impossible to directly experience the perspective of a chimpanzee, Engelmann and his colleagues designed five controlled experiments for groups of anywhere from 15 to 23 chimpanzee participants. 

In the first and second experiments, the chimps received a weak clue and a strong clue for a food reward in a box. The chimpanzees consistently made their choices based on the stronger evidence, regardless of the sequence in which the clues were presented. In the third experiment, the chimps were shown an empty box in addition to the strong and weak clues. After this presentation, the box with the strong evidence was removed. In this experiment, the chimpanzees still largely chose the weak clue over the empty box. 

In the fourth experiment, chimpanzees were given a second “redundant” weak clue—for instance, the experimenter would shake a box twice. Then, they were given a new type of clue, like a second piece of food being dropped into a box in front of them. They were significantly more likely to change their beliefs if the clue provided fresh information, demonstrating an ability to distinguish between redundant and genuinely new evidence.

Finally, in the fifth experiment, the chimpanzees were presented with a so-called “defeater” that undermined the strong clue, such as a direct line of sight to a picture of food inside the box, or a shaken box containing a stone, not a real treat. The chimps were significantly more likely to revise their choice about the location of the food in the defeater experiments than in experiments with no defeater. This experiment showcased an ability to judge that evidence that initially seems strong can be weakened with new information.

“The most surprising result was, for sure, experiment five,” Engelmann said. “No one really believed that they would do it, for many different reasons.”

For one thing, he said, the methodology of the fifth experiment demanded a lot of attention and cognitive work from the chimpanzees, which they successfully performed. The result also challenges the assumption that complex language is required to update beliefs with new information. Despite lacking this linguistic ability, chimpanzees are somehow able to flexibly assign strength to different pieces of evidence.

Speaking from the perspective of the chimps, Engelmann outlined the responses to experiment five as: “I used to believe food was in there because I heard it in there, but now you showed me that there was a stone in there, so this defeats my evidence. Now I have to give up that belief.” 

“Even using language, it takes me ten seconds to explain it,” he continued. “The question is, how do they do it? It’s one of the trickiest questions, but also one of the most interesting ones. To put it succinctly, how to think without words?”

To hone in on that mystery, Engelmann and his colleagues are currently repeating the experiment with different primates, including capuchins, baboons, rhesus macaques, and human toddlers and children. Eventually, similar experiments could be applied to other intelligent species, such as corvids or octopuses, which may yield new insights about the abundance and variability of rationality in non-human species.

“I think the really interesting ramification for human rationality is that so many people often think that only humans can reflect on evidence,” Engelmann said. “But our results obviously show that this is not necessarily the case. So the question is, what's special about human rationality then?”

Engelmann and his colleagues hypothesize that humans differ in the social dimensions of our rational thought; we are able to collectively evaluate evidence not only with our contemporaries, but by consulting the work of thinkers who may have lived thousands of years ago. Of course, humans also often refuse to update beliefs in light of new evidence, which is known as “belief entrenchment” or “belief perseveration” (many such cases). These complicated nuances add to the challenge of unraveling the evolutionary underpinnings of rationality.

That said, one thing is clear: many non-human animals exist somewhere on the gradient of rational thought. In light of the recent passing of Jane Goodall, the famed primatologist who popularized the incredible capacities of chimpanzees, the new study carries on a tradition of showing that these primates, our closest living relatives, share some degree of our ability to think and act in rational ways.

Goodall “was the first Western scientist to observe tool use in chimpanzees and really change our beliefs about what makes humans unique,” Engelmann said. “We're definitely adding to this puzzle by showing that rationality, which has so long been considered unique to humans, is at least in some forms present in non-human animals.”

What if your coworkers were AI? What if AI agents, not humans, monitored security camera feeds? What if you had an AI tech recruiter to find motivated candidates, or an entirely autonomous recruiting firm? What if UPS, but AI, or finance, but DeepMind?

Does that sound like absolute hell on Earth? Well, too bad, because the giant Silicon Valley investment firm Andreessen Horowitz (a16z) is giving companies up to $1 million each to develop every single one of these ideas as part of its Speedrun program. 

Speedrun is an accelerator program startups can apply to in order to receive funding from a16z as well as a “fast‐paced, 12-week startup program that guides founders through every critical stage of their growth,” according to Speedrun’s site. “It kicks off with an orientation to introduce the cohort, then dives into rapid product development—helping founders think through MVP while addressing key topics like customer acquisition and design partnerships.”

The program covers brand building, customer acquisition and launch, fundraising, team building, and more. The selected startups and founders meet each other, and receive the curriculum via workshops and keynote sections from “luminary speakers” such as Zynga founder Mark Pincus, Figma co-founder Dylan Field, a16z’s namesakes Marc Andreessen and Ben Horowitz, and others. 

Silicon Valley incubators and accelerators are common, but I’ve rarely seen such an unappetizing buffet of bad ideas as Speedrun’s AI-centric 2025 cohort. 

Last week, I wrote about Doublespeed, essentially a click farm that sells “synthetic influencers” to astroturf whatever product or service you want across social media, despite it being a clear violation of every social media platform's policy on inauthentic behavior. But that was just the tip of the iceberg. 

a16z’s Speedrun is also backing:

• Creed: An AI company “rooted in Christian Values” which produces Lenny, a “Bible-based AI buddy who's always got your back with wise words, scripture-inspired guidance, and a listening ear whenever you need it.”
• Zingroll: The “world’s largest Netflix-quality AI streaming platform,” which is another way of saying it’s a Netflix populated exclusively with AI Slop.
• Vega: which is building “AI-powered social orbits.” What does that mean? Not entirely clear, but the company has produced one of the most beautiful Mad Libs paragraphs I’ve ever seen: “We’re building the largest textual data moat on human relationships by gamifying the way people leave notes for each other. For the first time, LLMs can analyze millions of raw, human-written notes at scale and turn them into structured meaning, powering the most annotated social graph ever created.”
• Moona Health: an AI-powered Sleep care app the company says is covered by insurance. “Our AI-powered platform automates insurance claims and scheduling and analyzes sleep data – providing personalized session guidelines to therapists,” Moona says.
• Jooba: “The world’s first autonomous recruiting firm.”
• Margin: “The World’s first AI powered credit card.” Margin says “Customers earn points, with dynamic rewards that adapt to their preferences in real time.”
• First Voyage: A wellness app that gives you AI “mythological pets that turn wellness into play.”
• Axon Capital: billed as “DeepMind for Finance,” Axon says it has “pioneered brain-inspired, low-latency AI for financial markets.”

Part of the strategy for these types of accelerators and Silicon Valley venture capital firms more broadly is to place a lot of bets on a lot of startups with the knowledge that most of them are not going to make it. A million dollars is not a lot of money to a16z, especially when it only needs one of these companies to 100x its investment in order to make the whole endeavor profitable. What makes this Speedrun and the current moment we’re in with generative AI different is that a lot of AI implementations are going to be shoved down our throats before investors realize what AI is and isn’t good for. 

Are Doublespeed’s AI-generated social media accounts actually going to convince people to use whatever products they’re promoting? The accounts I’ve seen lead me to believe that the answer is no, but until then, they will continue to flood social media with garbage. Is Jooba going to entirely replace HR professionals and recruiters? I don’t know, but a whole bunch of people who are trying to get a job to pay rent are going to get caught up in a dehumanizing process until we find out. 

So the next time you find yourself asking why you're being inundated with AI wherever you go, remember that the answer is that someone with millions of dollars to spare paid for it on the off chance that it will yield a nice return.

Since last year’s disastrous rollout of Google’s AI Overviews, the world at large has been aware of how AI-powered search results can differ wildly from the traditional list of links search engines have generated for decades. Now, new research helps quantify that difference, showing that AI search engines tend to cite less popular websites and ones that wouldn’t even appear in the Top 100 links listed in an “organic” Google search.

In the pre-print paper “Characterizing Web Search in The Age of Generative AI,” researchers from Ruhr University in Bochum, Germany, and the Max Planck Institute for Software Systems compared traditional link results from Google’s search engine to its AI Overviews and Gemini-2.5-Flash. The researchers also looked at GPT-4o’s web search mode and the separate “GPT-4o with Search Tool,” which resorts to searching the web only when the LLM decides it needs information found outside its own pre-trained data.

The researchers drew test queries from a number of sources, including specific questions submitted to ChatGPT in the WildChat dataset, general political topics listed on AllSides, and products included in the 100 most-searched Amazon products list.

Read full article

Comments

Click here to go see the bonus panel!

Hovertext:
This is my gift to you, Dads of Internet.