I'm Kevin Roose, a tech columnist at the New York Times.
I'm Casey Newton from Platformer, and this is Hard Fork. This week, Anxious Generation author Jonathan Haidt returns to the show with new research on how social media is harming teens. Then, we asked you for your Claude Code experiments, and you blew us away. We'll share some of our favorites. And finally, it's a trip to the Forkverse. Thousands of you have now joined our Fediverse experiment, and Search Engine host PJ Vogt is here to talk about what we've learned so far. For example, Vladimir Putin. He's a problem. Not a good guy.
Well, Casey, we got a big show this week. Lots to talk about, but let's start with social media because there's been a lot happening in that world.
Yeah, you know, last week we talked about the nudification scandal over at Grok, and there's been a bunch more news about that this week. The US Senate managed to pass a bill on Tuesday that would allow victims of these sorts of non-consensual sexually explicit images to sue over, you know, the creation of those images.
Yeah. And Grok also rolled back its image generation feature for free users. You now have to be a paid user to Grok to nudyfy people against their consent.
To commit a crime, that's a premium feature now. Okay. That's going to be behind the paywall.
Yes. So they are not rolling that back fully, of course, because it is generating a huge amount of engagement for them. But they do seem to acknowledge that some people were taking it too far.
That's right, Kevin. And it is really this exact sort of phenomenon that is happening on social media that has led our guests this week to really lead a charge and say when it comes to younger people, we actually just need to keep them away from this stuff altogether.
Yes. The name of our guest today, Jonathan Haidt, will be familiar to longtime Hard Fork listeners. He came on the show back in March 2024 when his book The Anxious Generation came out. Since that time, that book has become a huge mega bestseller. Jonathan has become a leading advocate for things like phone bans in schools and these social media restrictions like the one we talked about in Australia that just went into effect. And there's been a lot of discussion about Haidt's thesis that social media and widespread smartphone use is harming mental health for kids and teens. And a lot of arguments about the science in that book, whether the studies that Haidt and others have cited as proof that social media is harming children are actually holding up under scrutiny.
Mhm. And as a matter of fact, Kevin, Jonathan Haidt and his fellow researcher Zachary Rausch have new research out this week which gave us an excuse to bring him back on the show.
Yes. And this new research by Jonathan and Zachary is designed to address one of the primary criticisms that his book got, which was about the difference between correlation and causation. So, we're going to talk about that research and we're also going to talk about what's been happening since his book came out.
Yeah. As he has become the world's hater-in-chief of social media. So, let's bring in Jonathan Haidt.
Jonathan Haidt, welcome back to Hard Fork. Thank you guys. Great to be back. You've been a busy man. You joined us originally just as your Anxious Generation book was going out into the world. The book has now spent 88 weeks on the New York Times bestseller list. That's a lot of weeks. Sold a bunch of copies, led to discussions all around the world and follow-on projects like a book you and my friend Katherine Price just wrote called The Amazing Generation, which is aimed at kids and teens. And now you are back with a new paper or papers about this issue of social media and mental health. You have a new post on your Substack called Mountains of Evidence and you're here to talk about that today. So just give us the TLDR here, Jonathan. What is the most important new finding in your research since your book came out?
So when my book came out, I thought that the main problem was mental health and there's a big debate in the academic world. Well, okay, people who use social media more are more depressed, but is that correlation or is it causation? And that has kind of defined the scope of the debate and that's gone on for a couple of years now. And it's what Mark Zuckerberg says to defend himself. Oh, there's no evidence of causation. Well, guess what? There are tons and tons of evidence of causation and Meta did some of the best studies to show it. There's so many different kinds of evidence including what the kids themselves say, what the parents say, what the teachers say, like everybody sees it. That's evidence. And then there are experiments with random assignment that show that when you get off social media for at least a week, depression gets less. So I was collecting all that evidence for this big review paper. That's one of the two projects. And while we were working on that, there was additional dumps of studies from Meta that come out in the attorney general's lawsuits against them. They post them online and there's all kinds of information in the briefs. So, we cataloged them at a new website we put up called metasinternalresearch.org. And that includes a couple of true experiments. So, the evidence for causality is now overwhelming. People have to stop saying, 'Oh, it's just correlational.'
So when you say the evidence for causality is overwhelming, break that down for us a little bit. Like what is your understanding of the mechanisms by which social media is harmful to children?
Yes. So Arturo Béjar, the whistleblower, did the bad experiences and encounters framework. So Meta itself, they've done tons of research, they collected, you know, what are kids saying, what's happening to them. What they found is, let's see, they get very high rates of sexual harassment, around 15% each week. Each week they have some person approaching you sexually. So they get a lot of that. Let's see what else do they have here. They get bullying. They see violence. They see hardcore porn. Oh, the biggest clearest one I think actually is sextortion. That's the one that just stands out like a sore thumb. You know, kids, if you're on social media you can get sexted. If you're not, you can't really. And the kids who get sexted are deeply shamed. You know, they shared a picture of themselves, these boys, teenage boys usually. And then their lives are ruined and some of them commit suicide. So, anyway, I'm just saying there's so many different mechanisms by which kids are getting harmed. And depression and anxiety, those are just two of the pathways.
So, you're saying that it's not any one thing. You're saying that the overall environment is so dangerous that it is difficult, maybe impossible, to use a social network like Instagram over an extended period of time and not have one or likely multiple of these harms hit you in some way. That answer somewhat surprises me because I feel like the usual answers we hear is like, well, it's negative social comparison, right? Like it seems like my friends are living a more exciting life than I am and now I'm upset or like the algorithm has driven me into a rabbit hole and like I started out a little curious about losing weight and now I have an eating disorder. So like, how has your understanding of the mechanisms of harm changed, if it has at all, since you wrote your book?
Sure. So when I started the book, I thought that the story was going to be girls using social media get depressed in part by social comparison. That's what everyone was talking about, you know, 5 years ago. And that's definitely one of the mechanisms. And I didn't know what the story was for boys when I started. And by the end of the book, and especially since the book came out, what I've realized is it's the whole goddamn environment. It's all the fish hooks dangled in front of boys. You want porn, gambling, vaping, sports betting. You know, even crypto investing is gamified.
In this paper, you've tried to sort of decouple two questions. One of them is what you call the historical population question. Basically, like the sort of restatement of the thesis of The Anxious Generation, which is like this stuff is like, you know, an entire generation raised on social media is showing all these effects on their mental health from this technology. And the other is just the sort of product safety question. Are these products as currently constituted safe for children and teens? What are you trying to do by sort of decoupling these questions? And I'll tell you how I read that. I read that it's like, oh, hey, Haidt is getting a little wobbly on one of the two sort of prongs of his argument there. Basically saying, you know, we may or may not be able to solve this historical population question. But when it comes to the question of product safety, we can answer that one now. So, am I reading that correctly?
Yeah. Well, yes, you are. Except I wouldn't use the word wobbly. So let me explain that my interest in separating the two questions came about because there was this report issued by the National Academies of Sciences, Engineering, and Medicine in late 2023 and the report comes out and it has a chapter showing all these harms, all these ways that it's harming kids, but yet in the summary of the report there was this sentence. It says, our review, quote, did not support the conclusion that social media causes changes in adolescent health at the population level. So I read this, I'm like, what the hell, guys, you have a whole chapter showing how it's hurting kids. What does it mean, the population level? And what I realized by reading further, what they mean is we can't, we're not confident that this is what caused those big increases after 2012. And okay, he's right because how do you prove something happened in history? It's very hard. You can't run experiments. So I will grant that we can't be 100% certain that what I said in The Anxious Generation is right. Can't be 100% certain. I'd love to hear another hypothesis. But the historical question is very hard to settle. But that's not what people want to know. What parents want to know and what legislators want to know is, is this consumer product which all kids are consuming massive quantities of, is this consumer product hurting kids? That's what they want to know. And guess what? We have seven different lines of evidence saying yes, it's hurting kids. The kids say it, the teachers say it, the experiments say it, the correlations, and Meta. I mean, my god, the data they have on harm is astonishing. So that's why I think it's important to separate the two questions. In science and social science, we're very careful about causality. We have to be certain about causality before we say X caused Y. And so I can't say that I'm certain that social media caused the big increase in 2012. But I can say I am 99.9% confident at this point that social media is hurting kids by the millions.
Yeah, you're understandably and I think appropriately very critical of the tech executives here. But I'd also note that parents gave kids these phones and schools allowed them and regulators did nothing for a decade. So if we're assigning blame for this situation, how much falls on everyone else involved?
I would say close to zero for this reason. The whole key to solving this problem, the reason we didn't solve it for so long, is that it's a series of collective action traps. And as a social psychologist, what we do for a living is we look at the ways that we influence each other. And there are certain situations where, you know, yeah, I don't want to give my 10-year-old a phone, but you know, everyone else has one and she's being left out. So the phones and social media, all these things, they put us in a trap. And so we feel we have to give in. And so since that's the situation, I can't blame the people. My rule is, as a social psychologist, if one person does something really bad, that might be a bad person. If everybody in a situation is doing something bad, that's guaranteed to be a bad situation. So, no, I don't blame. I mean, look, of course, parents should stand up and parent. But look, so many of us are trying and it's really, really hard. It's really, really, everybody's fighting all the time with their kids over this tech. We didn't ask for these fights. So I don't blame the parents. I don't blame the teachers. I blame the companies.
John, your life has changed quite a bit since we last talked to you. Your career has shifted from just, you know, being sort of more of a pure research and public intellectual to really being an activist on this issue. And I'm curious what that's been like for you and what you've learned about advocacy in this new role of yours. What actually seems to change people's minds who may be reluctant when they first encounter some of the ideas you're pushing?
No, thank you for that. Yeah, it's true. My life has changed a lot. You know, I've been a scholar and a researcher for a very long time. And I've started multiple organizations that try to translate findings from behavioral sciences to help important institutions work better. That's kind of my mission statement as an academic. So I've done some of that throughout my career. And when The Anxious Generation came out in March of 2024, I thought I was going to promote it for a few months, take the summer off, recover, and then get back to work on the big book I'm supposed to write on democracy. But because this whole thing blew up and because mothers around the world stood up and started saying, 'Where do we sign up? Let's go. Let's change. We got to save our kids.' So many people were coming to me saying, 'How can I help? What can we do?' Legislators were calling me. Governors were calling me. Because here's the thing. If you're a parent, you've seen it. Everybody has seen the problem and nobody really knew what to do about it because it's a collective action trap. And so since the book proposed four norms to get out of the trap — no smartphone before high school, no social media before 16, phone-free schools, and more independence, free play and responsibility in the real world — I proposed these four norms and people are adopting them all around in other countries that I haven't even been to. So I realized I'm 62. God knows how many years I have left with a brain that's still functioning. You know, I can't count on more than 10. And so how am I going to spend those remaining good years? I could write another book on democracy, which will be out of date by the time it comes out because my god, things are changing fast. Or I could devote myself to pushing for phone-free schools and raising the age.
Yeah. I'm curious. I know you probably can't tell us about all of the, you know, private meetings you're having with lawmakers and heads of state and, you know, tech executives, but just give us the flavor of like one meeting with someone important or influential as you've been schlepping all over, you know, hawking your book and talking about this message.
Yeah, sure. So, you know, the surprise for me has been how easy this has been, that everywhere I go, I'm pushing out open doors because everyone who's a parent has seen it. It's very popular among voters. And so I think the best example of an answer to your question is this. I did a quick trip to Paris, Amsterdam and London back in April of last year. And I worked with More in Common, a wonderful group that does all kinds of pro-democracy stuff. And they set up a dinner party for me to talk to members of French society and psychiatrists. It was really interesting. And one of the people there was in the French parliament. And he said, 'Oh, you should talk to Macron.' Oh, yeah, I'd love to, but you know, I'm leaving in 2 days. He said, 'I'll call him.' And the next day, I get a call from Macron's office. Can you come in tomorrow? So, I had a half-hour meeting with Macron in which I showed him the data because I was told in advance, he loves data. You know, you can show him evidence. So, I went through it with him. I said, 'Look, here's what's happening. You know, here's what we know.' And he thought about it and he said, as he was saying goodbye to me, he said, 'We will act.' And he sure did. He's really been pushing. What he said to me then was, 'I'm going to push it. This should be an EU thing, and if we can't get it through the EU, then I'll do it in France.' So, that's just, I mean, that was the most spectacular one. I was really surprised by that. But that's what I'm getting at, that everyone sees it. Everyone cares about this.
John, there's now this group of plaintiffs, lawyers, and law firms, I'm sure you know many of them and have worked with them before, who go around the country sort of looking for evidence of social media harms, especially against children. They bring these big lawsuits, these class action suits. They sue the tech companies. They try to get these huge settlements out of them. I'm curious what your thoughts on that system are. It's largely grown up since you published your book, and a lot of them cite your work as evidence that these companies should be held liable through the courts and should be forced to pay out huge settlements to the families of the victims here. Are you comfortable with that as sort of a makeshift solution until regulators get their act together? Does that feel like an optimal way of addressing these harms?
Well, I think it seems to be the only way that we have. Let's imagine that there was a new consumer product introduced, you know, some new toy or a new kind of candy bar with some new ingredient. And if we knew from internal reports from the companies that they deliberately designed it to be addictive, they put something addictive in it. And then a couple years after introduction, 90% of kids are eating 10 candy bars a day, and diabetes is going up, they're not eating healthy food, and we see rates rising around the world. At some point, there might be someone who says, 'Maybe this is not a safe consumer product. Maybe this company should be held liable for the harm it's inflicting on literally hundreds of millions of kids.' Literally hundreds of millions. Because social media is used by almost all kids. I mean, there are some countries where it's not, but in the medium to high income countries, it's almost universal. We all know someone who has a kid who's been hospitalized, eating disorder, suicide. We all know people. It is everywhere. And these companies have never faced a jury ever. No one has succeeded in holding them responsible for what they did to their kids. So this is an outrage. And so these lawyers who are taking these cases, yeah, they're heroes.
You know, there's this other dimension to it too, Jonathan, which is that, you know, your book understandably focuses on the crisis with children, but adults are very much living their own version of this, right? Like I was having coffee with some friends at the new year. Two of them are product managers in tech and I asked them, hey, you know, what do you want your 2026 to be about? And they said, I need to change my relationship with my phone. And they have implemented so many different tricks. They're putting it in black and white. They're setting screen time limits. They're leaving it in another room an hour before they go to bed. Like the systems that people are having to devise just to try to reclaim their own time and attention are getting ever more sophisticated. And this is among the people who are building the things on the phones, right? So, I'm curious how you think about sort of bringing some of these same arguments out of childhood and into adulthood because it seems to me like this goes sort of well beyond what is happening to kids under 18.
Yeah. No, you're absolutely right. And I think that's why we're winning. That's why it's so easy to get laws to protect kids all over the world because we all see it. We adults, we're all overwhelmed. I'm overwhelmed. I can't read a book anymore. There's too much other stuff coming in. I haven't focused on adults only because I don't want to legislate for adults. Like I don't want to tell adults what to do. But damn it, I don't want the companies, you know, sucking my children into toxic spaces without my knowledge or permission.
To speak a little bit about Australia. This week, Meta shut down about 550,000 accounts belonging to teens under this new Australian law that just went into effect banning social media for kids under 16. Jonathan, I'm so curious if you have a prediction about what is about to happen in Australia. Because it seems like we have this great natural experiment now. What are you looking for? And how confident are you that your own view that if you just simply remove phones and social media from schools that kids' mental health will improve?
Well, removing phones from schools, we'll know soon. Removing the whole deal from schools seems to greatly improve attention, friendship, discipline problems go down. So on the phone-free schools, the data is pouring in. I've never heard of a school that was unhappy about it or that reversed it. So I'm sure we're going to see that learning improves, friendship improves, fun in school improves, which means that truancy will go down and I think that will have a measurable impact on mental health. That's in the schools. Now what Australia did is even bolder, which is they said, we're going to put it on the companies. We're going to say that you have to be 16 to open an account where you sign a contract to give away your data, your rights without your parents' knowledge or permission. You can't do that till you're 16. At 16, you can do whatever you want. But before 16, you can't do that. And so here's the question. What percentage of Australia's kids will actually be off of social media? And we don't know. If the Australia bill is effective at getting social media use down below, say, 20%, then I think we will see over time kids have to sort of remember how to do other things other than scroll. I think we will see benefits. Especially if the Australians do the full program, which is, hey, go out and play, which is what the prime minister keeps saying, go out on the footy field. He says so if we can succeed in restoring a play-based childhood then we're going to see I think big benefits to mental health.
And like how long should we let that natural experiment run? Like if 5 years from now there's been no significant improvement in mental health among young people in Australia, would you conclude that this was a wrong hypothesis of yours, that we just didn't let it run for long enough? Like when should we sort of judge the success or failure of this intervention?
First of all, tell me when we get to say 60 or 70% of kids actually being off, and so the effectiveness matters. And this is a cat-and-mouse game, so the companies are going to have to up their game. So we'll see. But once we get to say 70% of kids are actually off, now we've broken the collective action trap. Now kids can be off. Now parents can say no more easily. And once we get there, I think, you know, within a year, we're going to start seeing lots of reports of different behavior. Will that show up in the national statistics? Not for a couple years. It takes a couple years before it'll sort of show up in these big national surveys we have. But 5 years, if we don't see anything budge in 5 years, I would then have to conclude that I was wrong in thinking that reducing social media use would improve mental health. But here's the thing. This whole debate has been framed around mental health. And so if it doesn't improve mental health, does that mean we should undo it? I would say no. I would say tell me how many kids have been sexted. Tell me how many kids have died from drug overdose deaths. Let's look at all the other ancillary harms and then we can decide whether it was a good policy.
You know, Jonathan, over time, I have to say I have become more sympathetic to your point of view. Like, I started this debate very much coming from the viewpoint of free expression, wanting to let kids communicate themselves online, wanting them to be able to explore their interests, connect with people like them. Over time, I have just become more persuaded by the product safety argument. I have read the internal documents. I have talked to the people who work at these companies. I do not believe that they care or are investing in protecting kids the way they should be. And to me, that argument is just sort of starting to carry the day. But I was having a conversation with my sister-in-law about this stuff over Thanksgiving. We were talking about some of these issues and she let me know that she kind of took issue with a comment I'd made on a previous episode where I said, 'You know, the one argument that you never hear anybody making is that Instagram is really good for kids.' Which I still basically believe.
Yeah. And you've heard it from some. And my sister-in-law made a version of this argument to me which is that, you know, for my nephew, he has an Instagram account. He's 13 and he is able to explore his interests which include gymnastics and cooking. And so when she, you know, looks at his own usage of Instagram she says this like basically seems okay to me. My question is what avenues do you hope that children use to explore their interests in a world where we do just yank that away from them?
So, I think last time I was on, we talked about a related topic and I said you've got to separate the internet from social media. Social media is a part of the internet. It's one of the worst parts. It's the one that's hurting kids the most. But look, all of us on this call, we're all old enough to remember the '90s when we first got to look at the internet, it was amazing. And you know, if you're LGBTQ in rural areas, suddenly like you can find information, you can find people there. I mean, it was amazing the way it brought people together. So I would never take that away. I would never say kids shouldn't be on the internet. Now, do they also need platforms that use algorithms to force-feed them whatever content was most upvoted by people based on the extremity of its emotions or expressions? Does your nephew benefit from having Instagram pick what he sees as opposed to having him type in what he's looking for? If you take away social media from kids, I don't see any loss.
I'm curious because like Casey, I am also pretty hyped at this moment in time. I am pretty convinced of the arguments you're making. But I would say the one criticism I might have is like, to me it feels a little bit like you are fighting the last war because when I go to schools, high schools now and meet high school students, they tell me like we are talking with AI companions now. That is the thing that we are doing. And I have this sort of fear in the back of my mind that like you will succeed at getting social media banned for under-16s all over the world. And then it will be like there's this new threat that we weren't even paying attention to. And we will pine for the days when teens were using social media to at least, they were communicating with other people and not these like AI companions. So do you worry at all that you have sort of not kept pace with the state of the technology?
No, this is a strategic move for this reason. AI is so new and it's morphing so fast and in the research community, it takes us 5, 10, 20 years to figure anything out. So, we've been arguing about social media for a long time. I think we've got it. I think I'm going to win on this issue. Now, it's a normal academic debate. There are researchers who look at the data and they see something different. But here's the way I'm thinking about it. If we can't win on social media, if we can't get consensus that this is bad and that government should do something, if we can't win on that, then I just give up on AI. Just say it's game over. Our kids are gone. We're never going to see them again. The boys are going to have, you know, sexy chatbots their whole lives. They're never going to reproduce. So if we can't win on social media, then we definitely can't win on AI. And since we are winning on social media and since governments around the world are waking up to the fact that they have an obligation to protect kids, a lot of them have never done anything, especially our federal government has never done a single thing, never ever, to protect kids on the internet. So governments are waking up and so the faster we can win on social media and delaying phones and more play and phone-free schools, then we have a chance, at least we can try to make the case, just get it out of elementary school. Just don't let kids be talking with chatbots when they're little.
Do you miss your obscure academic life at all? Does it bother you to be a public intellectual, an advocate, a man of mystery, a man on a mission?
Well, so yes, in that I'm like working all the time now and I really want to just take the summer off. I want to try reading books. I think if I read physical books, maybe I'll be able to finish one. But I have had a feeling of efficacy, a feeling that, you know, I've been working on this for a long time. And I feel like now I've been gifted the chance to actually make a difference in the world. And it's a kind of satisfaction beyond anything I've known in my life. So I do miss a lot, but I'm actually really energized and I'm actually pretty happy these days.
You know, I'll tell you this story. When I was pitching my book early last year, I was going around to different publishers and, you know, you've made it when the only pitch they wanted was like, 'The Anxious Generation but for AI.' It was like every publisher was like, I was like, 'No, it's not really what I'm working on. It's going to be more of a history thing.' And they're like, 'Are you sure? Are you sure you don't want to write The Anxious Generation But for AI?' I was like, 'Yes, I'm sure. I'm pretty sure John's going to do that.' But anyway, you have made a splash and yeah, if you have any tips on selling a book, I'll use them this fall.
Okay. I'd be happy to read your book if you want someone to read a copy because I am moving. That's my next question.
But you just told me you can't read a book anymore. So I, you know, I don't trust you.
Next summer I'll try to read. Send it to me electronically and then I'll feed it into ChatGPT and it'll give me a chapter-to-chapter summary.
Good. Good. That's what I do.
I have to say it's actually very comforting to me that even Jonathan Haidt is having ChatGPT read books for him because obviously I do that as well and I feel bad about it but I would now feel better.
This is the thing, no one has read a book since 2021 and they're just lying about it.
People are reading one kind of book which is romantic. So, if I ever wrote like a social science treatise, I would try to put it in a world where like dark fairies were having sex with each other.
Have you thought about that, John? Something to consider.
Actually, that is actually the way we open up our children's book. Okay. It's not sex, but it actually does open with a story about dark fairies, evil lizards.
By the way, I love the pandering of the adults. It's The Anxious Generation. But when you want to talk directly to the kids, all of a sudden they're amazing.
Yeah. But the kids better hold out some hope.
That's true. That's true. Yeah. All right, John. Thanks so much.
All right. Thanks, Jonathan.
All right, guys. Thanks so much for having me back. Take care.
Well, Casey, it's been another big week for Claude and Claude Code, which I think seems to be having kind of a ChatGPT moment. You know, I texted you earlier this week and I said, 'Is it crazy to say that Claude Code feels like the most important thing that has happened in AI since ChatGPT?' I realize that may sound crazy to some of our listeners, but I feel like we can make the case.
I think so too. I think this is a big moment, especially for people who are sort of code-inclined or code-curious. We've continued to see many, many posts on social media about people's experiments with Claude Code. It seems like it is actually creating something of an infrastructure crisis for Anthropic, which is also something that happened to OpenAI during their original ChatGPT moment. They just like could not keep up with the demand and the site kept going down. But I would say like the larger thing that has been going on is that non-coders are starting to experiment with this tool. Anthropic also released a version of Claude Code for non-coders this week called Cowork, which sort of sits in the Claude desktop app and allows you to sort of attempt various tasks using the same sort of agent-like framework that Claude Code uses, but in a way that's slightly less intimidating than opening up a terminal app.
Yeah. And if you think about what the ChatGPT moment was, it was a time when millions and millions of people realized simultaneously that you can ask questions and get pretty good answers. And you can generate text for many useful purposes. The Claude Code moment I feel like is the next evolution where people are saying I can now build something that is useful to me even though I don't really know what I'm doing. And whereas maybe 6 months ago the results I would have gotten were pretty bad or would have required a lot of technical know-how, I can now come in and I can type in a box and I can get basically exactly what I want. And that is actually huge. It's huge. And I had a version of that experience myself too in the past week because the read-it-later app that I built, Stash, we talked about my vibe coding experiment last week. I got so many responses from people who are saying, hey, can I use this thing? Are you going to like distribute this and put it in the app store? That I just decided, you know what, I'm going to do it. I'm going to release Stash to the world. And so I had Claude Code go in and like prepare a version of this for the public, build me a little landing page. You can now download it and try it for yourself at kevinroose.com/stash and it is working better than I ever expected and it remains sort of incredible to me that I did all of this as me, an idiot who does not know how to code. I was able to in a matter of hours build and release a working software product.
You know, first of all, congratulations for the launch of Stash. People have been really excited about this. My absolute favorite was you posted about this on X and somebody responded, 'Will it parse my X bookmark graveyard?' And you responded, 'I have honestly no idea what it will do.' And I laughed so hard when I read that. I thought, 'Finally, an honest software developer who just admits that we don't know how any of this stuff works.'
I like, I have no idea how any of this works. I have no idea if it works. It works for me. But I am very delighted that I was just able to sort of vibe launch this software product into the world with like honestly not that much work. It was not that hard. I did not use any special prompting tricks. I'm not even that good at using Claude Code and yet I was able to conduct an act of software engineering that now other people are downloading and using for themselves. I did have one incident that I want to tell you about which is that I accidentally had what I believe is my first AI security breach, which was that when I launched the version of this public Stash app that I was open-sourcing, it did inadvertently reveal my credentials on several of the sites that I use. So, I had to go back to Claude and be like, 'Hey, could you scrub my personal information out of there?'
That's a... I didn't know you had any credentials to do what you do. But you know, we want to talk about a few different things with Claude Code this week and we want to begin with listener responses because last week we asked you, hey, what are you building? And we were absolutely delighted with the response. We have gotten so many messages from you all over the past few days to our email of course, but also in the Forkverse, the Fediverse server that we set up that we're going to be talking about a little bit later in the show. But we just wanted to highlight a few of our favorites because I think they speak to the breadth of what people are able to build and also how little experience folks had before they started trying these things.
Yes. And I think this is like more than just a kind of show-and-tell segment because for me like this, there is something different about trying to do this stuff yourself versus just listening to, you know, us talk about it on our podcast. I really think that our listeners who listened to our episode last week and then went out and tried to vibe code things for themselves were a lot of them shocked by how easy it was, by how good the results were. And I think it, you know, for you, I was joking with you over text this week that it seemed like you were feeling the AGI in a way that maybe you hadn't before. So I think for our listeners maybe this past week has been an experience of starting to feel that, feel the progress, feel the capabilities improving, feel that things that were impossible even a few months ago are now becoming possible.
Absolutely. So as we get into what we heard from listeners...
It is important that I make the following disclosure: my boyfriend works at Anthropic. And I would go a step further and say, if you are feeling like this is edging into shill territory, I really do think you could probably do a lot of similar things with Codex, which is OpenAI's sort of Claude Code equivalent. I'm told Google has one as well, but due to the ongoing disaster of Google marketing, I truly cannot remember what it's called. Maybe you could Google it.
Perfect. So maybe try that. The point is these tools are now available and they're really good, and we want to get into it and talk about what you all are building.
And that's a good chance to make my disclosure too, which is that I work for The New York Times, which is suing OpenAI, Microsoft, and Perplexity over alleged copyright violations related to the training of AI models. Okay, so first up, we got some great examples from our listeners of websites that they have been building. We talked on the show last week about how we had each redesigned our own websites using these tools. One of those came from our listener, Gina, who had never vibe coded before, but after last week's episode told us that she had gone out and designed her own website. She says it took her about two and a half to three hours. And I'm going to look at it right now. Looks great, Gina. It's got some nice animations. She even told us that it has an Easter egg on it. If you hit Ctrl-M, it pulls up what looks like a MySpace profile from the '90s. Very cool. Congrats on your new website, Gina.
Yeah, super, super fun. Love the Easter egg there. You know, another website that we got came in from Sarah Haggard. She built a book site for book recommendations. It turns out that over the winter break, she wanted to catalog her books so that she could give her husband a list of books she has already read so that he doesn't accidentally buy her a duplicate. And she was able to do that, but then went a lot further and now has a website where you, the listener, can go on to sarahbooks.com, s-a-r-a-h-s-books.com, and she will make a book recommendation depending on what mood you're in. If you want a beach read, if you want a story about women's empowerment, many other categories in there, you can create your own account. Like, it is a full-featured web app. And I guess after she put it up, she got in touch with the founder of Goodreads, and now they're going to get together and have a fun conversation. So this is someone else who does not have a lot of technical experience, built something to scratch her own itch, and it is super cool.
And I'll just say there's something very romantic about vibe coding for your partner.
Yeah. A lot of alpha there.
Yeah, absolutely. Go to your partner today and say, 'What would you like me to build you, sweetheart?' And then just see if you can whip it up.
All right. But one more website we heard from a listener who turned his family's Christmas letter into an '80s text adventure inspired by the video game Zork. This is from listener David Phillips. He asked us not to read out the name of the website since it's got information about his kids on it. But I'm looking at it right now. It looks very cool. Very good idea for the annual holiday letter. It's got like green text on a black screen. It's like very much conjuring like an '80s like CRT monitor. And a really fun detail that he puts on the website is that his stepfather actually worked on the original Zork game.
And so this is actually kind of like a family tribute as much as it is a vibe coding project, and I just, I love that aspect of it. Yes, we also heard from several small business owners who told us that Claude Code and other agentic coding tools were allowing them to build tools that they otherwise would have had to pay a lot of money to someone else to build for them. My favorite of these was from Joe. He is a welder. Great AI-proof job of the future, by the way. And he said that he has no experience programming, but over the past few months, Joe told us that he has been teaching himself to build tools for his business. He said, quote, 'I built an MCP server that has turned my Claude chat window into a personal assistant.' He said it's able to track jobs, help him create estimates, organize PDFs and contracts. He said he's even connected Claude to the 3D modeling program that he uses, which has helped him build complex programs for CNC, which is sort of the computerized cutting and etching technology that many builders use. He says he's used Claude Code to make agents to search for potential work in his area and write a weekly lead generator for him. So I just thought this was a really interesting example of what might seem like a low-technology business, but that is actually able to go out there and build some software that helps him get some more business.
Yeah. And he ended his email in a way that really affected me. He said, 'I really want people in my demographic to understand that this isn't I asked ChatGPT to write an email or can you generate an image that has the correct spelling. This is to me, I built my own business infrastructure from scratch with an AI pair programmer despite having zero formal training and a high school education.' Totally pretty cool. Pretty cool.
Another example we got was from Fay Bell, a listener who runs her own business selling wallpaper to interior designers. She told us about a special tool that she has been creating called the wallpaper calculator, which we can look at over here. It is at her website. And this is a tool that basically allows a potential client, an interior designer maybe, to calculate exactly how much wallpaper they need to cover a particular wall. This tool, she said, also generates a layout diagram so that clients can sort of visualize what the wallpaper will look like in their space. She said this is something that she's always been doing by hand in tools like Photoshop or InDesign. But now the clients could just do it themselves and save themselves the time.
Yeah. And Fay wanted to let us know that she had no previous coding experience, and she loves that she can do this without having to pay any hosting costs. Like, she doesn't have to buy this software from some third-party provider, those can be really expensive services, and she just did it for her damn self.
Yes. One more that caught our attention came from a dad who made a web app to track a hilariously specific chore he had asked his kids to complete. Simon wrote that it was the week before Christmas, and his wife and he decided that their kitchen needed a tune-up. So they ordered new handles for the doors and drawers in their kitchen, and it was 70 handles in all. And to motivate their children to help them with this project, Kevin, they offered the kids a dollar per handle changed. And he had noticed that in the past they had quickly lost interest, believe it or not, when given the handle replacement task. So this time he spun up a web app to log their efforts. And the first thing it tells you when you open it is that the project is now complete and all 70 handles were replaced. So this was a success. But basically there's a leaderboard for his four children. And you can see who made the most money. They were able to cash out their winnings via Apple Cash to the bank or via chocolates, which I love. And it was Ben who came in number one with 44 handles replaced. Vic came in at zero. So I guess Vic was holding out for a higher wage to get involved in the knob replacement business. But just again, this is so wacky. And was this necessary? Absolutely not. But what I love about this project is it speaks to a theme that I felt in so many of the things that listeners sent us, which is the joy of creation. It is fun to make things. It is fun to have a dream and to be able to quickly realize it. And so, yeah, I just love this one.
What excites me is less the products themselves. It is that people, I think, are starting to understand the pace and trajectory of AI progress through using these tools. A year ago, you could not type into a box, 'Make me a website for my family's handle replacement project,' or 'Make me a tool for my small business that allows people to estimate the wallpaper needs of their project.' And now you can. And I think unless you have really spent time, even a couple minutes, using one of these tools, you really do not understand the state of the art. And so I think a lot of people unfortunately are in the same position that we were in back when ChatGPT came out in 2022, which is like they are opining on a technology that they do not actually understand because they have not experienced it for themselves. And so for the listeners out there who are listening to this and saying, 'Okay, well these projects sound great, but what does this have to do with me or my life?' I would just say there is no better way right now in 2026 to get a handle on what the frontier of AI progress looks like than to come up with a project, no matter how silly or trivial seeming. Give it to one of these AI coding agents and watch what it does.
I agree with that. You know what I would say, Kevin, is to me this stuff is the flip side of slop. This is anti-slop, right? If slop is about a world where every surface online seems like it's being taken over by these digital creations that you didn't ask for, that you find confusing, they blur the line between reality and fiction, and they just sort of make you feel hypnotized and gross. This is the reverse of that. This is real people saying, 'I have a need in my life, and I'm going to go make it with my own hands.' And we've always talked about AI as a dual-use technology. We do spend a lot of time on the show talking about its many downsides, which are huge and real and scary. But there is also upside here, which is at least in this moment, this can be an empowering tool. And so to me, I think one reason why I was so delighted to see all these listener responses was they spoke to the way that technology tools can still empower us.
Yes. And it also gave me an experience that I had not had before, which is that I now understand how annoying it is to build and release software products in the world, only to have users tell you everything that is wrong with them. Like five minutes after I made my Stash app public and put it up on GitHub so that people could download it and use it themselves, I started getting feature requests. It was like, are you not impressed by the app that I have made and released to you for free? Already I understand in a more visceral and embodied way the pain of our listeners who are product managers at big tech companies.
Yes. But you also now just have a perfect thing you can say back to them, which is just fork it and make the version yourself, right? Go vibe code your own feature.
