John Herrman
LightNews LightNews
jwherrman.bsky.social
John Herrman
@jwherrman.bsky.social
8.5K followers 600 following 680 posts
posting about posts at new york magazine. have me on your podcast!
Posts Replies Media Videos
Reposted by John Herrman
jwherrman.bsky.social
once you start looking for it, you see evidence and consequences of online desocialization everywhere nymag.com/intelligence...
n the mundane course of modern life, you might occasionally find yourself glimpsing the dark abyss - that is, catching a few seconds of a stranger's phone screen. A peek over a shoulder, through a car window at a dash-mounted phone, or perhaps, pointedly, at the hands of a distracted person whose eyes you're trying to raise, will likely reveal a version of the same thing: not, in 2026, a wall of text, a feed, or even a slideshow of stories, but an endless scroll of tall videos. Taken together, the videos sometimes tell a simple story in algorithmic silhouette, with one bikini video after another, a cascade of talking sports heads, an unbroken flow of clothes to buy, or influencers talking over the News with a particular political orientation. Just as often, though, a stolen screen glance reveals a dismal anti-story, in which an AI model has guessed, in a whole bunch of different directions, what the user is "most likely to be interested in or engage with." It has done so by following, to borrow Meta's language, a series of steps: "Gather inventory," "leverage signals," "make predictions," and "rank reels by score." To the intended viewer, the resulting mix of videos is incoherent but probably also intuitive, at least on an individual basis. To you, the screen peeper, the spectacle is profoundly uncomfortable, not just because you're eavesdropping the output of an intimate machine-learning profile gleaned from private and often passive "signals" provided by someone you don't know, but because you know, deep down, that your chained-together clips — and the chains of people you know, love, and respect — would be exactly as alien, and alienating, to anyone outside of your narrow algorithmic cone. You're seeing yourself. You're really just sitting there, recreationally A/B testing content for hours in order to help Meta and TikTok find videos — any videos, about anything — that are a little bit harder for us not to watch.
February 3, 2026 at 4:35 PM
jwherrman.bsky.social
oh absolutely
February 3, 2026 at 6:02 PM
Reposted by John Herrman
robhorning.bsky.social
"desocialization" (like "social deskilling") is a good term for what was once talked about in terms of "social graph" vs. "interest graph," or of "algorithmic recommendation"
nymag.com/intelligence...
If internet firms are defined by their fastest-growing monetized products, well, Meta is basically a Reels company, one that successfully chased TikTok into continued relevance, allowing Mark Zuckerberg to throw money at his next big chase (into generative AI). This isn’t just a formal change from the News Feed to Stories to predictively recommended vertical videos, though. It’s a long (and nearly complete) process of platform desocialization. Platforms originally defined by keeping up with people you know, or have at least heard of, become something fundamentally different.
February 3, 2026 at 4:54 PM
jwherrman.bsky.social
I do agree the pool is shallower than it feels, and that you can suss out hints of an algorithmic mainstream. but what use is it (to us) if it’s invisible?
February 3, 2026 at 5:05 PM
Reposted by John Herrman
tcarmody.bsky.social
All media aspires to the condition of television
jwherrman.bsky.social John Herrman @jwherrman.bsky.social · 1d
once you start looking for it, you see evidence and consequences of online desocialization everywhere nymag.com/intelligence...
n the mundane course of modern life, you might occasionally find yourself glimpsing the dark abyss - that is, catching a few seconds of a stranger's phone screen. A peek over a shoulder, through a car window at a dash-mounted phone, or perhaps, pointedly, at the hands of a distracted person whose eyes you're trying to raise, will likely reveal a version of the same thing: not, in 2026, a wall of text, a feed, or even a slideshow of stories, but an endless scroll of tall videos. Taken together, the videos sometimes tell a simple story in algorithmic silhouette, with one bikini video after another, a cascade of talking sports heads, an unbroken flow of clothes to buy, or influencers talking over the News with a particular political orientation. Just as often, though, a stolen screen glance reveals a dismal anti-story, in which an AI model has guessed, in a whole bunch of different directions, what the user is "most likely to be interested in or engage with." It has done so by following, to borrow Meta's language, a series of steps: "Gather inventory," "leverage signals," "make predictions," and "rank reels by score." To the intended viewer, the resulting mix of videos is incoherent but probably also intuitive, at least on an individual basis. To you, the screen peeper, the spectacle is profoundly uncomfortable, not just because you're eavesdropping the output of an intimate machine-learning profile gleaned from private and often passive "signals" provided by someone you don't know, but because you know, deep down, that your chained-together clips — and the chains of people you know, love, and respect — would be exactly as alien, and alienating, to anyone outside of your narrow algorithmic cone. You're seeing yourself. You're really just sitting there, recreationally A/B testing content for hours in order to help Meta and TikTok find videos — any videos, about anything — that are a little bit harder for us not to watch.
February 3, 2026 at 4:51 PM
jwherrman.bsky.social
"social media isn't actually social" was a poor observation but a great prediction
February 3, 2026 at 4:45 PM
jwherrman.bsky.social
put another way, platforms that collected and arranged communities in order to monetize them with targeted ads are now entirely *ordered* by the logic of an ad network. a wild thing to watch happen over the last decade.
February 3, 2026 at 4:44 PM
jwherrman.bsky.social
up until about five years ago, serving up more "unconnected" content was a known engagement hack that nominally social networks felt the need to rationalize. then, rather suddenly, it was the whole model. that has to matter!
In 2022, in a memo outlining its post-TikTok strategy, Meta laid out how thoroughly it was shifting the emphasis of so-called "social" media toward what it had termed "unconnected" content, not just with Reels but across other platforms. In a 2023 post providing some technical background on the "Al behind unconnected content recommendations on Facebook and Instagram," the company described how showing users "highly personalized recommendations from the tens of billions of pieces of content that are outside of a person's network" actually "enhances their experience." Around the same time, Mark Zuckerberg was telling investors that, already, 20 percent of content Facebook and Instagram users were seeing came from "people, groups, or accounts they don't follow." But I do suspect that the shift from massive platforms full of people seeing things mostly on purpose to platforms full of people seeing things mostly because an algorithm thinks they might engage with them remains an underrated factor in just how strange the internet — and downstream entertainment, and media, and politics — can feel in 2026. A lot of intuitions about what might have been wrong with social media, or at least what effects it might have been having on the world around it, feel less applicable to the systems that emerged in its place, where a vision of social and parasocial connection has been supplanted by a program of systematic desocialization. If the recent story of social media could be told as billions of people being thrown together into a new shared and monetized environment and driving each other slightly insane, the next era of the big platforms is shaping up to one of increasing isolation, of passive consumption stripped of any sense of shared culture, and of, to borrow a term, comprehensive "unconnection," not as a condition to be corrected - at least by other people - but as a goal nearly achieved.
February 3, 2026 at 4:35 PM
jwherrman.bsky.social
once you start looking for it, you see evidence and consequences of online desocialization everywhere nymag.com/intelligence...
n the mundane course of modern life, you might occasionally find yourself glimpsing the dark abyss - that is, catching a few seconds of a stranger's phone screen. A peek over a shoulder, through a car window at a dash-mounted phone, or perhaps, pointedly, at the hands of a distracted person whose eyes you're trying to raise, will likely reveal a version of the same thing: not, in 2026, a wall of text, a feed, or even a slideshow of stories, but an endless scroll of tall videos. Taken together, the videos sometimes tell a simple story in algorithmic silhouette, with one bikini video after another, a cascade of talking sports heads, an unbroken flow of clothes to buy, or influencers talking over the News with a particular political orientation. Just as often, though, a stolen screen glance reveals a dismal anti-story, in which an AI model has guessed, in a whole bunch of different directions, what the user is "most likely to be interested in or engage with." It has done so by following, to borrow Meta's language, a series of steps: "Gather inventory," "leverage signals," "make predictions," and "rank reels by score." To the intended viewer, the resulting mix of videos is incoherent but probably also intuitive, at least on an individual basis. To you, the screen peeper, the spectacle is profoundly uncomfortable, not just because you're eavesdropping the output of an intimate machine-learning profile gleaned from private and often passive "signals" provided by someone you don't know, but because you know, deep down, that your chained-together clips — and the chains of people you know, love, and respect — would be exactly as alien, and alienating, to anyone outside of your narrow algorithmic cone. You're seeing yourself. You're really just sitting there, recreationally A/B testing content for hours in order to help Meta and TikTok find videos — any videos, about anything — that are a little bit harder for us not to watch.
February 3, 2026 at 4:35 PM
jwherrman.bsky.social
educational polarization matters but so does having a cool place to go/look
February 3, 2026 at 3:47 PM
jwherrman.bsky.social
[guy who only talks about platforms voice] there's less of a mainstream to theorize about/against, formerly marginal spaces have been scaled up to the point of collapse, the dull rich techies are uneasily becoming mainstream, a swing against them is impeded by this extended (temporary) dislocation
February 3, 2026 at 3:41 PM
jwherrman.bsky.social
rather, I want to avoid paralysis at the AGI-ish stage of his post to try to understand his and his company’s model of the rest of the world
February 2, 2026 at 9:25 PM
jwherrman.bsky.social
I guess I should have narrowed that post a bit: taking the *rest* of it seriously, since most mainstream coverage seems to stop before we even get there
February 2, 2026 at 9:22 PM
jwherrman.bsky.social
yeah, there are more words here but not much actual connective tissue compared with, like, Sama just going “yeah, UBI maybe”
February 2, 2026 at 8:19 PM
Reposted by John Herrman
ed3d.net
this is a really, really good article.
jwherrman.bsky.social John Herrman @jwherrman.bsky.social · 2d
My attempt to take Dario Amodei's new manifesto literally and seriously: as a call for more liberal democracy written by a prime example of the ways it can be overwhelmed nymag.com/intelligence...
A modei's cautionary vision of the future is better described as two. In one, he spells out what companies like his might need to do in response to what companies like his are building. In the other, he outlines what everyone else might need to do as this unstoppable process plays out. Whatever comfort the first story offers — some critics think he's actually too dismissive of straightforward "Al just kills everyone" scenarios, and it's getting hard to tell where the middle of that discourse is these days — is complicated by his repeated argument, presented as an observation, that slowing things down simply isn't, and hasn't been, an option. ("If one company does not build it, others will do so nearly as fast. If all companies in democratic countries stopped or slowed development, by mutual agreement or regulatory decree, then authoritarian countries would simply keep going.") If we're entering the world he's imagining, where incredible new risks can only be addressed by unprecedented coordination between companies, regulators, and states, it sure seems like we're off to the worst possible start. An even bigger problem is the actual political and economic environment into which Amodei is pleading, which looms over nearly every sentence in "Adolescence" that isn't about Anthropic itself. Once we leave the theoretically terrifying but rhetorically sandboxed world of pure technological change and risks — we're building something very powerful, but don't worry, we can make sure it's helpful and moral - Amodeis proposed remedies for dealing with the downstream consequences of "powerful AI" and automation - on a timeline he sets before 2028! — read like a list of domestic and global anti-trends. They include altruistic coordination between big companies; noblesse oblige as policy; an expanded welfare state; citizens "banding together" to reject autocracy; "rapid vaccine development" and universal air purifiers in a world trying to forget COVID; and mass rejection of surveillance in a country where citizens can't buy enough Ring cameras and government-surveillance contracting is the second-hottest start-up category after AI. There's an awful lot of talk about "our adversaries" at a moment when the category appears to be in flux and the "coalition of the US and its democratic allies" meant to "contain" autocracies is struggling to hold together at all. A constitutional amendment about responsible AI deployment? Tech companies, which have lately been hemorrhaging workers just to invest more in AI, keeping employees around just because? In this economy?
February 2, 2026 at 7:06 PM
jwherrman.bsky.social
yeah, more clearly now than ever. which I think contributes to some of the strangeness here! and emphasizes the many conflicts of his whole deal
February 2, 2026 at 6:34 PM
Reposted by John Herrman
segyges.bsky.social
this is the best analysis of anthropic's public posture i have seen anywhere, in mainstream journalism or elsewhere. i had this itch like i should review Dario's latest and i don't think I do any more
jwherrman.bsky.social John Herrman @jwherrman.bsky.social · 2d
My attempt to take Dario Amodei's new manifesto literally and seriously: as a call for more liberal democracy written by a prime example of the ways it can be overwhelmed nymag.com/intelligence...
A modei's cautionary vision of the future is better described as two. In one, he spells out what companies like his might need to do in response to what companies like his are building. In the other, he outlines what everyone else might need to do as this unstoppable process plays out. Whatever comfort the first story offers — some critics think he's actually too dismissive of straightforward "Al just kills everyone" scenarios, and it's getting hard to tell where the middle of that discourse is these days — is complicated by his repeated argument, presented as an observation, that slowing things down simply isn't, and hasn't been, an option. ("If one company does not build it, others will do so nearly as fast. If all companies in democratic countries stopped or slowed development, by mutual agreement or regulatory decree, then authoritarian countries would simply keep going.") If we're entering the world he's imagining, where incredible new risks can only be addressed by unprecedented coordination between companies, regulators, and states, it sure seems like we're off to the worst possible start. An even bigger problem is the actual political and economic environment into which Amodei is pleading, which looms over nearly every sentence in "Adolescence" that isn't about Anthropic itself. Once we leave the theoretically terrifying but rhetorically sandboxed world of pure technological change and risks — we're building something very powerful, but don't worry, we can make sure it's helpful and moral - Amodeis proposed remedies for dealing with the downstream consequences of "powerful AI" and automation - on a timeline he sets before 2028! — read like a list of domestic and global anti-trends. They include altruistic coordination between big companies; noblesse oblige as policy; an expanded welfare state; citizens "banding together" to reject autocracy; "rapid vaccine development" and universal air purifiers in a world trying to forget COVID; and mass rejection of surveillance in a country where citizens can't buy enough Ring cameras and government-surveillance contracting is the second-hottest start-up category after AI. There's an awful lot of talk about "our adversaries" at a moment when the category appears to be in flux and the "coalition of the US and its democratic allies" meant to "contain" autocracies is struggling to hold together at all. A constitutional amendment about responsible AI deployment? Tech companies, which have lately been hemorrhaging workers just to invest more in AI, keeping employees around just because? In this economy?
February 2, 2026 at 6:05 PM
Reposted by John Herrman
jwherrman.bsky.social
My attempt to take Dario Amodei's new manifesto literally and seriously: as a call for more liberal democracy written by a prime example of the ways it can be overwhelmed nymag.com/intelligence...
A modei's cautionary vision of the future is better described as two. In one, he spells out what companies like his might need to do in response to what companies like his are building. In the other, he outlines what everyone else might need to do as this unstoppable process plays out. Whatever comfort the first story offers — some critics think he's actually too dismissive of straightforward "Al just kills everyone" scenarios, and it's getting hard to tell where the middle of that discourse is these days — is complicated by his repeated argument, presented as an observation, that slowing things down simply isn't, and hasn't been, an option. ("If one company does not build it, others will do so nearly as fast. If all companies in democratic countries stopped or slowed development, by mutual agreement or regulatory decree, then authoritarian countries would simply keep going.") If we're entering the world he's imagining, where incredible new risks can only be addressed by unprecedented coordination between companies, regulators, and states, it sure seems like we're off to the worst possible start. An even bigger problem is the actual political and economic environment into which Amodei is pleading, which looms over nearly every sentence in "Adolescence" that isn't about Anthropic itself. Once we leave the theoretically terrifying but rhetorically sandboxed world of pure technological change and risks — we're building something very powerful, but don't worry, we can make sure it's helpful and moral - Amodeis proposed remedies for dealing with the downstream consequences of "powerful AI" and automation - on a timeline he sets before 2028! — read like a list of domestic and global anti-trends. They include altruistic coordination between big companies; noblesse oblige as policy; an expanded welfare state; citizens "banding together" to reject autocracy; "rapid vaccine development" and universal air purifiers in a world trying to forget COVID; and mass rejection of surveillance in a country where citizens can't buy enough Ring cameras and government-surveillance contracting is the second-hottest start-up category after AI. There's an awful lot of talk about "our adversaries" at a moment when the category appears to be in flux and the "coalition of the US and its democratic allies" meant to "contain" autocracies is struggling to hold together at all. A constitutional amendment about responsible AI deployment? Tech companies, which have lately been hemorrhaging workers just to invest more in AI, keeping employees around just because? In this economy?
February 2, 2026 at 2:19 PM
Reposted by John Herrman
sambiddle.com
Trump has finally delivered the Jade Helm presidency the far right warned of in 2015: a techno-authoritarian assault on the White House’s enemies. Only now instead of raising an alarm over a hoax, they’re cheering the real thing.

theintercept.com/2026/02/02/t...
Welcome to the Jade Helm Presidency
Conservatives once panicked about a supposed federal plot to invade their communities and quash dissent. Now they’re cheering it.
theintercept.com
February 2, 2026 at 4:40 PM
jwherrman.bsky.social
from my last look at this, just what you’re talking about: “AGI” as purposeful and paralytic narrative. these narratives are widely read and shared, I’m trying to talk to people who’ve encountered them
February 2, 2026 at 2:38 PM
jwherrman.bsky.social
yeah, that’s why I’m trying to suss out the political character of the projects: to understand them as current and future lobbyists with a lot of influence and money
February 2, 2026 at 2:33 PM
jwherrman.bsky.social
my take is approximately that this rhetorical strategy still leads to an incoherent place, if you read beyond the lesswrongisms
February 2, 2026 at 2:30 PM
jwherrman.bsky.social
strongly agree! nymag.com/intelligence...
How AI Made the Future Unthinkable
Advanced artificial intelligence will change everything, we are told. Do we get a say?
nymag.com
February 2, 2026 at 2:29 PM
jwherrman.bsky.social
One of the claustrophobic aspects of AGI-ish discourses that they ask you to helplessly pick an ultimate preferred winner, and for a progressive reader that's obviously Amodei. But he pleads helplessness, too. His preferred system has already lost control, and it's not clear how it would get it back
February 2, 2026 at 2:21 PM
jwherrman.bsky.social
Amodei is clearly interested in and nervous about politics, but seems constrained in the ways he can talk and maybe even think about them. In the world he's building into, we'll be in trouble well short of "powerful Al." We're basically doomed already!
It's possible that's just where Amodei is coming from as he witnesses his peers in AI push forward with either hyperwealthy aloofness or an increasingly open devotion to the sorts of post-democratic, post-human ideas that were once relegated to theoretical discussion on LessWrong. (Altman was a prepper before OpenAl took off, when he was merely rich; Musk plainly believes he should, and one day could, run the world.) It's certainly easy to pick up on a deep frustration that his timeline, and Anthropics, happens to overlap with Donald J. Trump's. But while making his case for recommitting to liberal democracy as a response to technological development, and against political tendencies that he finds worrying already, Amodei draws attention to its core vulnerability. I mean, look what already happened! A regulated market economy nestled inside a democracy is now monomaniacally focused on building a machine to replace and disempower its subjects, and some of the people marshaling its resources are simultaneously telling us that it might be a disaster but that they have no choice but to keep moving as fast as possible. We're talking about a system that has failed to contain superintelligence before it even exists. Amodei's vision of people "banding together" within political structures as they exist today doesn't sound like much of a match for what he says his company is trying to conjure, even before it's been conjured. (In "Machines," he admits that he has "no strong reason to believe AI will preferentially or structurally advance democracy." Reading "Adolescence" leaves one with doubts that we'll find out either way.)
February 2, 2026 at 2:20 PM