Pod Insights Hard Fork Superconductor Superconfusion, KOSA’s Hidden Costs and HatGPT About this episode Researchers in Korea claim they’ve identified a material that could unlock a technological revolution: the room temperature superconductor. Material scientists are skeptical, but enthusiasts on Twitter are enthusiastic. Why is the internet so excited about superconductors?
Then, the Kids Online Safety Act is headed to the Senate floor. Would it actually keep children safe? And how would it change the internet?
Plus: Kevin and Casey play HatGPT.
Transcript 13,358 words · 5 speakers
Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.
0:00 Speaker 00 Support for this podcast comes from Dell Technologies. It's time to put AI to work. Is your business ready? The challenge isn't getting started with AI. It's scaling beyond pilots while maintaining control of your data, governance, infrastructure, and costs. Dell AI Factory provides the integrated foundation you need to scale AI securely and efficiently with one trusted partner, Dell, AI on your terms. From idea to impact. Learn more at dell.com/yourwaytoAI. That's dell.com/yourwaytoAI.
0:35 Kevin Roose? I'm sweaty. Is my sweat showing up on cam?
0:37 Casey Newton? Yeah, the live chat is going wild about it. They're saying, somebody cool this guy down.
0:43 Kevin Roose? Gang, gang. Mmm, ice cream's so good.
0:49 Casey Newton? Did you see? Somebody sent me a video yesterday. It's just like an old German man saying that. Anyways. I really hope that we become NPC streamers.
0:59 Kevin Roose? Oh, let's. How funny would that be? Hard Fork's so good.
1:03 Casey Newton? Hard Fork's so good.
1:11 Kevin Roose? I'm Kevin Rus, a tech columnist at The New York Times.
1:13 Casey Newton? I'm Casey New from Platformer.
1:14 Kevin Roose? And you're listening to Hard Fork.
1:16 Casey Newton? This week on the show, the hunt for the mythical room temperature superconductor. Then, the kids' online safety act. It might make teens safer, but will it break the internet in the process? And finally, our favorite game returns. It's time for Hatch GPT.
1:47 Kevin Roose? So Casey?
1:48 Casey Newton? Mm-hmm.
1:48 Kevin Roose? It's sort of a slow news week out there, not gonna lie.
1:50 Casey Newton? We can't start the opening block of the show by saying it's been a slow news week.
1:55 Kevin Roose? I like to be honest with our listeners.
1:57 Casey Newton? I hope you're just trying to like lower people's expectations so that you can over deliver.
2:00 Kevin Roose? Yes. Well, so what I was gonna say is that it has been a pretty slow news week in tech, except for. The possible discovery of room temperature superconductors. Let's get into it. So Casey, I don't know if you've noticed, but tons of tech people, basically the same people who have been very into crypto and very into AI and tweeting about both of those topics for years, have been obsessing over the past week with this possible, maybe, discovery of something called a room temperature superconductor. What do you know about this story?
2:34 Casey Newton? OK, what I know about it is very little. I would not describe myself as a science person, you know, when it comes to matters of material science. Usually I just think, well, that's none of my business. And if it becomes important to me, someone will tell me about it. But it seems like maybe we've arrived at that point.
2:52 Kevin Roose? Well, we might have arrived at that.
2:54 Casey Newton? OK.
2:54 Kevin Roose? So so I want to bracket all of this discussion that we're about to have with sort of a disclaimer that there is still a chance that all of this is garbage, is not real. But, by the way, big theme on Hard Fork, big chance that all of this is garbage, but go on. But I think it's really fascinating that this has attracted so much interest in the tech world, you know, like people like Sam Altman have been tweeting about it, all these sort of science hobbyists and venture capitalists. A surprising chunk of the internet seems to be suddenly obsessed with superconductors. So I feel like we should talk about what is going on and why these people are so fired up. So, OK, let's start with a little science lesson, Casey. So what do you know about electrical transmission?
3:39 Casey Newton? What I know about electrical transmission is that when you transmit electricity, you lose some of it as it gets from place to place.
3:46 Kevin Roose? Correct. So if you think about electricity moving through a wire, there's a current flowing that experiences friction because electrons bump into vibrating atoms.
3:56 Casey Newton? Yeah, you know, when you're making wine and like some of the wine evaporates and they call that the angel's share?
4:02 Kevin Roose? They do?
4:02 Casey Newton? They do. That's what they call it.
4:03 Kevin Roose? That's very poetic.
4:04 Casey Newton? Isn't that very poetic? Well, so what you're saying is that there's like an angel's share of electricity transmission.
4:09 Kevin Roose? Exactly. So some amount of the electricity that you put through a wire is lost. When your phone gets hot when you use it a lot, that is because there is Resistance, there is like an imperfect transmission of electricity happening, and some of that electricity is being wasted as heat.
4:24 Casey Newton? You know what else it's like? It's like, you ever like having a conversation with a romantic partner? And they're like, sort of not getting it. It's because 5 to 10% of what you're trying to say has actually been lost in transit. So there's a lot of parallels for this in our world.
4:35 Kevin Roose? Exactly. So for more than a century, scientists have been very interested in what we call superconductors, which are materials that conduct electricity perfectly with no loss or waste whatsoever. And we have actually discovered somebody. I say we, because I play a huge role in this discovery, based on reading several thousand tweets about it and some Wikipedia articles. So basically, there are materials that if you take them down to like minus 320 degrees Fahrenheit, or even colder than that sometimes, they can become superconductors. They transmit electricity with no loss. But that's not really that useful, because how many situations do you have where like you have the ability to cool something down to like these extremely cold temperatures?
5:20 Casey Newton? The only thing that cold in 2023 is the crypto industry. So very limited use cases for that.
5:26 Kevin Roose? So for the past several decades, scientists have been chasing this dream of a superconductor that works at room temperature, where you could just sort of make a wire out of something and have it be in a device or on a power line, and it would just work, and you would have no waste, no loss in the system.
5:44 Casey Newton? So it's like if Silly Putty turned out to be a superconductor, that'd be great, because, you know, it's shelf-stable, you don't have to do anything to protect it, you just sprinkle a little bit of it on whatever you want.
5:54 Kevin Roose? Out of all the materials you could have chosen as your example, I'm fascinated that Silly Putty made it to the top of your list.
6:00 Casey Newton? Well, because that's what I think of as like a vaguely like scientific space-age material, you know? It's like, that was sort of the last great advance in material science was Silly Putty.
6:09 Kevin Roose? It's true. Okay, so here is what is happening in the world of material science. Last week, these two academic articles showed up on Archive, which is a platform for what they call preprints, which are basically things that have not been peer-reviewed but that people can go and look at. And one of them had the sort of off-putting and academic title, "Superconductor PB10XCUX P0460."
6:36 Casey Newton? Wait, Kevin, that's my Hulu password.
6:41 Kevin Roose? Showing levitation at room temperature and atmospheric pressure and mechanism. The other one had a much better, some social media editor got their hands on this one, and it had an attention grabbing title called "The First Room-Temperature Ambient Pressure Superconductor." Now, there we go.
6:56 Casey Newton? Now, that's a pre-print title.
6:58 Kevin Roose? Now, we're cooking. So, both of these pre-prints came from a group of researchers at the Quantum Energy Research Center in Seoul, who claimed to have identified what they call LK99, which is a room temperature ambient pressure superconductor. Basically, a material that can work as a superconductor not only at room temperature, but in sort of normal pressure situations. This is this holy grail that researchers have been chasing for decades. And what's also really unusual about this is that their sort of process seems strangely simple. Like, scientists have sort of taken for granted for many years that if we ever did discover
7:40 Kevin Roose? one of these room temperature superconductors, it would involve some like crazy rare earth metal, or some process that made it very hard to produce.
7:48 Casey Newton? Yeah, it's like whenever they discover a new element of the periodic table, it's like, you know, they're bouncing 45 lasers off of these rare metals inside of a supercollider, right? And like, it only exists for 0.1 seconds. Like, that's how we make new things in this world.
8:00 Kevin Roose? Totally. And actually, funny story. So, do you remember the first Avatar movie?
8:04 Casey Newton? I'll never forget it.
8:05 Kevin Roose? There was this thing called unobtanium. Unobtanium was a room temperature superconductor. And it cost trillions of dollars. It was the most valuable material on the earth, which is why they had to go to Pandora to get it.
8:17 Casey Newton? And wipe out the Na'vi.
8:18 Kevin Roose? Yes.
8:19 Casey Newton? Yeah. So, wait, I mean, how far ahead is James Cameron thinking?
8:22 Kevin Roose? That man is living in the year 3000.
8:25 Casey Newton? Seriously? What a year he's having, between the the superconductor and the submersible stuff.
8:31 Kevin Roose? Yeah. So, we assumed that this would be very expensive to produce if we ever found it.
8:34 Casey Newton? Yeah.
8:35 Kevin Roose? These Korean researchers, they appear to have done this with like very conventional processes. It seems almost too simple.
8:43 Casey Newton? Now, can you say anything about what these materials are or what the process? Like, I'm trying to imagine this LK99. Can I hold it in my hand? Is it in a vial of some sort? Is there a Bunsen burner involved? What's happening here?
8:57 Kevin Roose? Well, I'm glad you asked, Casey. So it turns out that this recipe that they basically put in this paper includes fairly common chemical elements, things like lead, sulfate, copper, and phosphorus. They sort of go through the step-by-step process. It involves mixing things in a ceramic crucible, heating things in a furnace at 725 degrees Celsius.
9:20 Casey Newton? Now, that is hotter than your oven will typically go at your home.
9:23 Kevin Roose? It is, yes. Yeah, you're gonna need one of those fancy pizza ovens to replicate this in your home. You've gotta like, you know, mix some lanarkite with some Cu3P, things that I definitely know what they are. And eventually, you sort of mix all this stuff together, bake it at a really high temperature, do a few more steps like this, and then it turns into a superconductor.
9:43 Casey Newton? Okay.
9:44 Kevin Roose? So the other fun thing about this is that these researchers, in addition to putting up this paper with this sort of recipe in it, they also show off a video.
9:53 Casey Newton? Okay.
9:53 Kevin Roose? Now, can I show you this video?
9:54 Casey Newton? Well, let's see it.
9:55 Kevin Roose? Can I see this video? Okay. I'll play this video for you. And I want you to describe what you're seeing in this video.
10:01 Casey Newton? There is a sort of metallic disk, and some sort of material has been placed onto it. Looks like kind of like a little, like a chip, almost.
10:11 Kevin Roose? And it's floating.
10:12 Casey Newton? Oh, it is floating.
10:13 Kevin Roose? It's float-- a little bit of it is touching, but it's floating.
10:16 Casey Newton? It looks, it looks almost like a magnet. And so what we're looking at, that is LK99.
10:21 Kevin Roose? That is LK99.
10:22 Casey Newton? So, okay, so it is actually something. You could hold it in your hand.
10:25 Kevin Roose? Yes. So it's a little complicated. But basically, if something is a superconductor, it produces something called the Meissner effect. Now, we're getting into--
10:34 Casey Newton? Can I just say, every week, you're asking me about, have I ever heard of this effect or this conundrum or this postulate? And it's become one of my favorite things on the show. So anyway, yeah, please tell me about the Meissner effect.
10:45 Kevin Roose? So when something is a superconductor, basically, it levitates. It, it, when it, when exposed to a magnet, when exposed to a magnet, it, it hovers above the magnet.
10:56 Casey Newton? Okay.
10:56 Kevin Roose? And this is the principle behind sort of maglev trains and things like that. So this video that these South Korean researchers put out shows a little pellet of LK99 kind of expelling a magnetic field, which they say is sort of a telltale sign of superconductivity. So that is the claim, that these Korean researchers have figured out this holy grail in material science. And these papers, this claim by this group of Korean researchers, has totally lit this particular corner of the internet that I am in on fire. There's a subreddit called rSingularity, which is usually talking about like AI and scientific progress and stuff, which has just been totally taken over by posts about LK99
11:41 Kevin Roose? and videos of people claiming to replicate these findings and discussions of what the implications could be if this is real. There's actually been like a prediction market about whether this experiment will be replicated.
11:54 Casey Newton? And, and what are people saying?
11:55 Kevin Roose? Well, right now, on manifold markets, so the, the question that people are betting on is, will the LK99 room temp, ambient pressure, superconductivity preprint replicate before 2025? And right now, the market says that there is a 36% chance that this will happen.
12:16 Casey Newton? Okay, so a significant chance, but the, the betting money is still against it.
12:21 Kevin Roose? Well, yes, researchers are sort of, I'm saying researchers, but most of the people betting on this stuff.
12:27 Casey Newton? Degenerate gamblers.
12:27 Kevin Roose? Degenerate gamblers believe that there is a 36% chance that this will replicate in the next two years. In addition to the degenerate gamblers, there are also just a ton of Discord channels and Telegram channels where this is being discussed. And, you know, on Twitter, these people with like sort of previously very small presences have just been gaining tons and tons of followers because they're doing all these threads about like, how LK99 will change everything, and here's the investment strategy. Energy that you need.
12:56 Casey Newton? Here's how you can use LK99 to advance in your career.
12:58 Kevin Roose? Right, exactly. Right, the LinkedIn thread boys have arrived. It's sort of just this vigilante science movement that has cropped up online, not only to follow along with what's going on, but just to speculate about what it all means. And I'm sure you've seen these experiments where people are trying to recreate this thing, to replicate the results of this experiment.
13:20 Casey Newton? Wait, is there a viral TikTok challenge or something?
13:24 Kevin Roose? Sort of, almost. There is a guy who has been livestreaming his attempt to replicate LK99 on Twitch.
13:32 Casey Newton? That's amazing.
13:33 Kevin Roose? There's another, this one I'm really obsessed with. There is an account on Twitter, IrisIGB, who is a Russian scientist with an anime girl avatar, who claims that she is making her own samples of LK99 in her kitchen.
13:47 Casey Newton? Wow, so she does have an oven hot enough.
13:49 Kevin Roose? Well, it may be that she has actually modified her oven in some way that is extremely unsafe to be able to get it to these high temperatures. And she's not following along with the exact instructions laid out in this Korean preprint.
14:00 Casey Newton? Well, look, if you're making LK99 at home, part of the fun is putting your own spin on it.
14:04 Kevin Roose? Exactly, a little cumin, a little garlic powder. You know, who needs to measure? Basically, this has become a folk science project that the internet scientists and sleuths are throwing themselves into. Because in their minds, this would be a game-changing innovation if it were actually real. And it's become a moment where the buttoned-up world of white lab-coat material science merges with the weird fandoms of the internet and makes this weird and crazy thing.
14:37 Casey Newton? I mean, material scientists must be so excited that the spotlight is shining on their industry. And there's now a wellspring of people rising up to learn about what they do.
14:47 Kevin Roose? I would say yes and no. I mean, the material scientists that I have seen reacting to this are sort of like, we have a process for replicating things like this. There is a scientific method.
14:58 Casey Newton? There's a woman in Russia, she's got an oven. We gotta get some time with her.
15:04 Kevin Roose? So they're basically like, let's just, before we get too excited, before we start proclaiming that our new technological utopia of superconductors has arrived, let's actually check out the science and make sure it works. Now, you may be asking yourself, why does this matter?
15:20 Casey Newton? That was exactly what I was asking myself.
15:22 Kevin Roose? So if this is true, it is a potentially transformative innovation. Because what you could do, if you actually had a room temperature, ambient pressure superconductor, is you could replace all the power lines in the world with this material. And all of a sudden, your whole energy grid gets more efficient. You could also have vastly better things like charging. Your devices could charge almost instantaneously. They wouldn't get hot when you used them too much. It would also apparently be a very big deal for things like quantum computers, which I've been on the record on this show saying I think quantum computers are fake.
15:58 Casey Newton? But maybe this will make them real.
15:59 Kevin Roose? But maybe this can make them real. And basically, this would be a huge breakthrough that would mean that essentially all of our electrical infrastructure could be rebuilt in a much more efficient way.
16:09 Casey Newton? Oh, so that would be, obviously, hugely consequential. I'm sure we would be talking about that a lot more on the show. But my assumption is because, well, everything that you described, Kevin, it frankly, sounds a little bit too good to be true.
16:23 Kevin Roose? It does. And so one of the interesting things is that even though these preprints are going viral on the internet, and actually, there are some scientists who do think that we should be pursuing trying to confirm these findings, there are scientists in some of the relevant fields who are not as excited. They've got misgivings about the data, the way it was presented. I've seen some going as far as to say that these preprints are not serious and basically we shouldn't be paying attention to them. So in general, I would say that the scientific community is approaching these findings with kind of curiosity, but also a lot of skepticism, which is understandable because there hasn't been a successful replication of these findings from a reputable lab.
17:05 Kevin Roose? And there have been several other instances in recent years, including one just this year, of scientists claiming to have identified a room temperature superconductor and then later having to walk their claims back.
17:17 Casey Newton? Now, here's a question. Have we asked ChatGPT whether this is a real superconductor?
17:21 Kevin Roose? I did ask ChatGPT and Claude to analyze the LK99 research and tell me whether they thought it was legitimate or not. They have no idea. They're just making stuff up. So let's just stipulate what we've stipulated, which is that this may be bogus. This may not be real. It is fascinating to me how the internet has latched on to this.
17:43 Casey Newton? And look, I mean, I like this part of the internet, right? Like, people coming together around something that could be positive. You love to see it. There was a time when average people did not have access to pre-print papers on the internet, right? You had to have an academic library access, something like that. Now with Archive, there is so much research available to people. And yes, it is not peer reviewed, which means there's a lot more stuff out there that isn't going to hold up under scrutiny, but it gives access to average people who might have an interest or a background in it to say, you know what? Let's tear this apart. Let's see what we can make of this.
18:13 Kevin Roose? Yeah, let's crank up the old oven and start making some LK99 in the kitchen.
18:17 Casey Newton? Now, is it dangerous to make LK99? Do we need to put like a content advisory on this?
18:22 Kevin Roose? Yeah, I mean, I would say I am not going to be trying to fabricate my own LK99. I can't even bake sourdough, so I'm not going to be attempting this in my own kitchen. I would also recommend that our listeners, unless you are a material scientist or Russian anime girls, not try to do this at home. But I also think this is exciting. Like, I hope this is real. I want to believe. I think it would be very exciting. I think we do need scientific breakthroughs, and I actually think it is good for people to be sort of interested in this stuff at like a citizen level, right? I want kids growing up thinking like, I want to discover a superconductor, because that
19:07 Kevin Roose? is the kind of thing that actually could make our world better. At the same time... The other side of this that makes me slightly nervous is that when you do have all of these sort of citizen scientists on the internet, looking at pre-prints and, you know, doing experiments in their kitchens, like, I also think that undermines trust in the scientific establishment. And, you know, I'm old enough to remember the, the long ago year of 2021, when a bunch of people on the internet got really excited about these bogus miracle cures for COVID, right? They were looking at the pre-prints, showing that maybe, you know, maybe ivermectin was the cure for COVID. You know, maybe hydroxychloroquine
19:49 Kevin Roose? was the magic bullet that was finally going to prevent this awful pandemic. And the same kind of phenomenon emerged, where like, on very little evidence, with very little support from the scientific establishment, these sort of renegade Twitter scientists just started getting all kinds of clout for making these outlandish, and it turns out, completely false, claims about these COVID cures. And it just really did set us back as a country and as a society. So I'm wondering, like, do you think that the positives of this kind of citizen engagement with science outweigh the negatives?
20:22 Casey Newton? I, I think so. I mean, you know, maybe we need to develop better filtering mechanisms. I don't think the issue was necessarily that people wanted to study whether ivermectin could be helpful. It was more that, like, after there started to be evidence that it wasn't, Joe Rogan still had on a controversial doctor saying, "Actually, you know, it cured my COVID," or whatever. So I think it's much more about how we direct attention to people after we have learned that what they were saying was probably not true. You know, the scientific process is always going to be full of people getting things wrong, right? You know, the whole scientific journey is this very iterative path, and there's just going to be a lot of stuff that isn't true that we learn along the way. So what I would say is, like, basically, no, I'm not worried about a bunch of people getting really interested
21:04 Casey Newton? in science and doing experiments and doing analysis, you know, because the alternative is, you know, watching Bravo or, you know, playing Fortnite. You know, if you have some spare time you want to do science, do some damn science!
21:18 Kevin Roose? I don't know. See, I, I, I want to agree with you, because I, I want to open up science to the masses. I want people to be interested in it. I even want people to be trying, like, weird stuff, you know, as long as it's safe in their own homes.
21:31 Kevin Roose? But I would just say, like, ask the immunologists and the, like, public health officials during COVID if they would rather have done their response in a world with no social media. And the answer that you may have gotten from them is yes, that they actually do feel like social media and the sort of citizen interest and the pre-prints and the sort of mass availability of not only good information about the virus and the responses to the virus, but a lot of bad information actually did make their jobs and their lives harder. So-
22:01 Casey Newton? That's true, but there was also a point where that same establishment was saying, "Hey, don't bother wearing a mask. It's not going to help," right? And it was people on social media saying, "This doesn't make any sense to us."
22:09 Kevin Roose? Right. So, Casey, we've been very careful and sort of responsible in talking about this.
22:14 Casey Newton? Incredibly responsible.
22:16 Kevin Roose? I just want to ask you the question.
22:18 Casey Newton? Yeah.
22:19 Kevin Roose? Do you think this is real? Do you think the rocks float? Do you think LK99 is real? And do you think that we are all going to be riding around in maglev trains and charging our phones in two minutes?
22:32 Casey Newton? Kevin, I have no idea. Absolutely not. You know, people probably wonder, "Why do people start a podcast?" It's so they can sound smart talking about things. I cannot sound smart talking about this. Like, everything that you've just said, I would have probably gotten a C minus in if I had to take an exam in it in high school physics.
22:51 Kevin Roose? Yeah. I mean, I feel the same way. I am by no means an expert on this topic, but I did spend the vast majority of the past couple of days learning about superconductors, which is not something-
23:04 Casey Newton? After your two days of research into material science-
23:06 Kevin Roose? I'm ready to be a thread boy.
23:09 Casey Newton? I like that you're like, "It's so dangerous when the public embraces these new scientific trends and spouts off at the mouth." But that said, after 48 hours of Wikipedia, I'm prepared to make my statement.
23:20 Kevin Roose? Except for me.
23:20 Casey Newton? Except for you.
23:22 Kevin Roose? My favorite tweet about this, by the way, was from someone called India Venom, who said, "If LK99 proves to be a replicable superconductor, then it may become possible to have two or more tabs open in Chrome without plugging your laptop in."
23:36 Casey Newton? And my favorite tweet about it was the person who said that the only superconductor they knew was Lydia Tarr.
23:48 Casey Newton? When we come back, the controversy around a new bill that backers say will protect kids, critics say could break the internet.
24:13 Speaker 00 Support for this podcast comes from Dell Technologies. It's time to put AI to work. Is your business ready? The challenge isn't getting started with AI. It's scaling beyond pilots while maintaining control of your data, governance, infrastructure, and costs. Dell AI Factory provides the integrated foundation you need to scale AI securely and efficiently with one trusted partner. Dell, AI on your terms. From idea to impact. Learn more at dell.com/yourwaytoAI. That's dell.com/yourwaytoAI.
24:47 Speaker 03 Companies are pouring money into AI, but few leaders can explain how those AI investments work with the systems you already have. You know, the ones that actually run your business? Without Systems, you don't have to start over with AI. You can choose the projects where AI will have the most impact. And OutSystems gives you the platform to make it agentic on your terms with complete control every step of the way. So you can trust AI with the mission-critical systems that run your business. Own your agentic future. Learn more at outsystems.com/NYT. That's outsystems.com/NYT.
25:22 Speaker 04 AI spend is growing fast, but most businesses still can't see exactly where that money is going. Ramp gives businesses visibility and control across AI providers with real-time guardrails by team and model, helping ensure spend goes to the highest ROI areas. Stop guessing where your AI spend is going. Get started with Ramp. And if you're new to Ramp, you can get $250 off at ramp.com/hardfork. Go to ramp.com/hardfork. Ramp.com/hardfork.
25:55 Kevin Roose? Kevin.
25:56 Casey Newton? It's time to talk about a bill sitting up on Capitol Hill. Have you heard of the Kids Online Safety Act?
26:02 Kevin Roose? Well, I had not. And then you told me that you wanted to talk about it on the show this week. So I studied up. So I now know just enough to be dangerous about the Kids Online Safety Act. And I'm excited to talk about it because I think it actually is quite important.
26:13 Casey Newton? Yeah, well, one of the things we've talked about over the past few months is how nervous we ought to be about the effects of the internet on children and mental health. There is a mental health crisis among young people in this country. And this year, the Surgeon General issued an advisory saying that social media use can lead to bad mental health outcomes for some kids. So when you have a finding like that, the general thing you want to see is Congress to do something about it, right? You want the lawmakers to come together and figure out, OK, well, if there's some sort of harm here, how can we mitigate it? And one of the things that Congress is trying is this Kids Online Safety Act, K O S A, COSA, is I think how people are referring to it online. You may see that if you're browsing the internet.
26:56 Casey Newton? And what makes it interesting for our purposes is, one, last week lawmakers voted to send it to the floor. But, two, unlike almost all of the other attempts to regulate big tech over the past half a decade or so, this one has a lot of cosponsors. I believe 43 cosponsors in the Senate.
27:12 Kevin Roose? So this could actually happen. This could actually become law.
27:14 Casey Newton? It is a bipartisan group that is rooting for it. And so I think it's just kind of time to take a look at what it does. Because in the end, and we'll get into this, I think this bill could stop us from being able to use the internet in the ways we want to.
27:28 Kevin Roose? Right. And I will also say that COSA, this bill, appears to have crossed some threshold where like people actually know what it is and care about what it is. Like, I've been seeing TikToks where people are talking about their takes on this bill and how it would affect the internet. And it sort of feels like, do you remember when like FOSTA and SESTA and before that, like SOPA and PIPA were these other bills that were sort of like, lots of people started caring about them all of a sudden and writing angry letters to their congresspeople? And it sort of became like a mainstream-ish concern. And that feels like it's happening with COSA.
28:02 Casey Newton? Also, don't FOSTA, SESTA, SOPA, and PIPA sound like a Norwegian family? Just like a beautiful, beautiful family. But anyway, so what would this bill do? One of the big pillars is that the bill says that platforms will now have what they call a duty of care to prevent and mitigate various harms. And those harms include mental health disorders, eating disorders, bullying, harassment, sexual exploitation, drugs, deceptive marketing practices. All things that we can agree are bad, right? Like, we don't want minors to be exposed to these things.
28:37 Kevin Roose? Correct.
28:38 Casey Newton? Yeah.
28:38 Kevin Roose? But what does duty of care actually do? Like, it sounds good, and you want the platforms to feel a duty to care about this stuff. But what is the, is that like a legal phrase?
28:48 Casey Newton? It is, but you've hit on what is the real challenge here, which is that what this law would ask platforms to do is to guess in advance which speech on their platforms might harm people in the future. And that's just extraordinarily difficult to do, right? Some kids might see eating disorders content and develop an eating disorder. Other kids might see that content and think, oh, gosh, like, that seems a really, a really bad thing. Now I've been educated about this thing, and I am going to avoid it. The platform is not going to be able to predict that in advance. And one of the real issues here is the way that this law takes this one-size-fits-all approach to trying to protect teens when, you know, kids are individuals, and they have all sorts of circumstances that may inform how they react to content they see online.
29:31 Kevin Roose? So it basically opens up legal liability for platforms. They can be sued if minors encounter certain types of harmful content on their service that leads them to do harmful things.
29:45 Casey Newton? Yeah, that's right. It empowers the Federal Trade Commission to take action against these platforms by forcing them to change in some way. And then it also lets state attorneys general sue these platforms if they feel like any of their residents have been harmed.
30:01 Kevin Roose? So let's put a pin in that because I think that is really important. And that's something that I imagine we'll come back to because that is one of the pieces about this that worries me. But there's also other stuff in this COSA bill. So what else does this bill do?
30:12 Casey Newton? So the other big thing is that, well, if you're going to try to hold platforms accountable for their effects on young people, you need to know who on the platform is a young person. And you need to mandate somehow that platforms identify young users on their platform. If you want to protect them, you have to know who these users are. When this bill was first being discussed, it required the platforms to verify the ages of their users. And this has a lot of privacy harms for average people, right? Because teenagers and adults use the same websites in many cases. And so that doesn't just mean that if you're a kid, you're going to have to prove that you're 17 in order to access some website. It means that all adults are going to have to prove that they are adults to access the websites that they are already using.
30:56 Casey Newton? And how are they going to do that? Well, we only have so many easy ways to verify our identity. Like, we have driver's licenses, but then how many private companies do you want to have to upload a full copy of your driver's license to? So Congress gets a lot of criticism for this measure, and they come back and essentially they just make it more vague. And they say, essentially, "Well, to the best of your ability, platforms, you need to guess a user's age." But the very strong implication is that we're going to put pressure on you to know the ages of your users. And if you're serving up content that we think is inappropriate for younger users to see, then you are legally liable. So they moved from this very explicit thing, which was bad, to this kind of vague wink, wink, nudge, nudge.
31:39 Kevin Roose? Right. Another piece of the COSA bill that I found interesting is this idea of parental controls. From what I understand, and correct me if I'm wrong, but it basically, the bill basically says that if you are a platform that is used by minors, you have to build in parental controls so that parents can say, "My child can only interact with this app or this service in these ways." You can turn off things like messaging. You can limit their time in the app. Basically, the controls that you would get- Over something like Netflix for your kids account, like, all services with minors on them have to have those built-in protections. And to me, that like, that actually sounds like a reasonable idea, but there are people who are very upset about this. So why is that?
32:24 Casey Newton? Well, I think the main reason is that there are going to be some kids, particularly in minority groups, who are going to be seeking content that they don't necessarily want their parents to know about. And, look, I agree with you. I think parents obviously have a lot of control over what their kids are seeing, particularly at younger ages. But once you're 14, 15, or 16, I do think you have some right to autonomy. And if you are a young LGBT person, you want to connect with other folks like you online, and you have parents who are, you know, very anti-LGBT, that could potentially put you at risk. And, you know, one of the points in that Surgeon General's advisory
33:04 Casey Newton? was that a lot of these same kids can benefit greatly from finding that community online. It can actually improve their mental health. And so the danger is that we're going to over-police what is on the phones of teenagers in a way that will make some of them less safe.
33:20 Kevin Roose? I can see that. That makes sense. So let's go back to this piece about the FTC and state attorneys general being able to go after platforms and content providers on the internet over this sort of harmful content. Because this is an area where I actually do think the criticisms of this bill make a lot of sense to me. So just explain what that means and sort of how this could be weaponized.
33:47 Casey Newton? Sure. So let me tell you a scenario that I think is extremely likely to happen if this law were to be passed. The Heritage Foundation is this right-wing group. They will essentially lobby for someone to sue the platforms for exposing minors to content involving transgender people. And in the view of the Heritage Foundation, that is a harm for a young person to know that trans people exist. And so they will go out and they will find a Republican Attorney General, and that Attorney General will sue the platforms, saying, "You have exposed the children of this state." ...to this harmful content and then that will become a big legal case, right? And, well, you might be saying that will probably be thrown out.
34:30 Casey Newton? This clearly seems like an unconstitutional restriction on speech rights. But you have to remember just the chilling effect that this sort of thing has on these platforms. These are corporations. They are incentivized to avoid litigation, to avoid negative headlines. They want to avoid being in these kinds of fights. So it might not take that many of these attorneys general to come forward and to file these lawsuits before these platforms just really start cracking down on the kinds of things that you can say online, right? So the overall effect of this, even though I do think that there are some good intentions here to protect children, to address this mental health crisis, I really do think that the end result of it is it's going to be harder for you and me to discuss things online. It's going to be a restriction on our speech rights.
35:14 Casey Newton? And there are just a lot more of us adults than there are of children.
35:17 Kevin Roose? Right, and I also think there's a big difference between a 13-year-old and a 16-year-old. I remember when I was 16, yeah, I was interested in looking at stuff on the internet, some of which my parents probably wouldn't have approved of, right? It was not all like puppies and rainbows and homework assignments. And I think that's okay. And I think that if I had lived in a world where this bill was in effect, like it still would have been possible for, I would have found a way around it because I was a little hacker kid and I loved like finding ways around the restrictions on my school's internet, for example. But I think it would have just been a deterrent and I think ultimately that would have been bad for me.
36:00 Kevin Roose? I do think there is a difference in emotional maturity between a 13-year-old and a 16-year-old. At the same time, like you and I have talked on this show many times about how much of a minefield social media especially is for young people and how harmful it can be. And I do want Congress to do something about that. The problem is, when you allow the government to police speech on internet platforms, or to go after tech companies for making certain types of content available, you also enable the kind of speech restriction and chilling effects on other types of content. Like, it's hard to come up with a bill that allows for education about eating disorders
36:43 Kevin Roose? on social media, that does not also allow attorneys general to sort of try to take down content about what they called, quote, "gender ideology," or "trans people," essentially.
36:55 Casey Newton? Yeah, I agree with you. You know, one place I would like to see the government act is just to invest in more research about this. I think that while there is a growing body of research around the questions of teens and social media, I don't think that we have much that is particularly conclusive. And it just makes me worried that we don't fully understand the problem here. It also definitely seems the case that some kinds of content are harmful for some people and just totally innocuous to others. And I just think it's really hard to design laws around that, you know? And I don't want to be the person that says, like, "Well, because this bill has, like, one flaw in it, we have to all throw up our hands and do nothing." But, I don't know, take a step back. There is a real war on speech in this country. Look at what is happening in the libraries of
37:38 Casey Newton? this country. Look at what is happening in the school curriculums in this country, right? Republicans are trying to just eliminate vast amounts of speech from the public square. And so, when Democrats team up with those same Republicans to try to protect kids, and I'm sort of putting that in air quotes, but the main effect of that is it's just going to be harder to talk politics online. I think you want to be really, really careful about that, and ask whether there's any other path to get to protecting those kids.
38:07 Kevin Roose? I mean, I think what the designers of this bill did, right, is call it the Kids Online Safety Act. I mean, if you just say that something is protecting the children, like, it's very hard to argue with that. And actually, you know, Joe Biden himself came out recently and urged Congress to pass this bill, because, you know, something called the Kids Online Safety Act, if you're a congressperson or a lawmaker who doesn't study this- Stuff for a living, you just say like, yeah, who's against protecting kids on the internet? I will say, you know, after reading the full text of the proposed bill, there are a couple parts of it that I actually thought were good ideas. And so I think we should call that out in the interest of fairness. I don't think this is all bad.
38:47 Kevin Roose? There was a part about sort of transparency and allowing researchers access to data from the platforms about how minors are using their products.
38:58 Casey Newton? I'm very into that.
38:59 Kevin Roose? I think that's a good idea. I think limiting data collection and ad targeting aimed at minors is a good idea. And I do like this parental controls thing. I mean, one of the things, you know, my kid is not of internet using age yet. But when he is, like, I will want to know what he is doing on the internet for some amount of time. When he's 16, maybe I'll care less. But when he's, you know, 11, 12, 13, 14, like, that is going to be information that I want. And I will want easy and sort of intuitive ways that are built into the systems themselves to sort of be able to see and control what he's able to do. So I actually think that is a good idea. I think that should be limited to platforms above a certain size. I mean, one thing that you typically see in internet regulation bills
39:44 Kevin Roose? is that there are sort of rules that kick in only after you have, you know, 50 million users or 100 million users. To my knowledge, this COSA bill does not do that. So a platform with 50,000 users would have to follow the same rules as Facebook and Instagram, which just seems burdensome for the smaller companies.
40:03 Casey Newton? Mike Masnick has this great blog called Tech Dirt, actually profiled in the past week in the New York Times by Cash Hill. Really great story. And he's been writing a ton about COSA, has very much sort of shaped my own thinking. And one of the things he's written about is I don't want to have to like, collect age verification information from everyone who wants to come read my blog, you know? But like, that is the risk here, that just people, honest entrepreneurs like Mike and myself, just trying to run little websites. It's like, well, you know, you got to prove that you're 14 to read my little blog about tech policy. It just feels silly.
40:34 Kevin Roose? But when Platformer does hit 100 million subscribers, I think that you should be subject to. Just excruciating amounts of regulation and compliance.
40:43 Casey Newton? And look, I already have to deal with the California Franchise Tax Board. And trust me, that's more than enough torture for any one person.
40:51 Kevin Roose? Okay, so in conclusion, I think we're both agreed that COSA, as it is currently written, is not a great example of internet regulation and that it actually poses some real risks if it were to be passed.
41:03 Casey Newton? There is, and you know, I want to say one more thing, which is there is a kind of magical thinking. And it goes something like this. If we could just pull the plug out of the back of Facebook and TikTok and YouTube and all these other social platforms, the world would sort of be restored to a pristine state. Our politics would be healed, mental health issues would disappear, and we'd just sort of be living in utopia. And I just think that the evidence shows that's probably not the case. You know, in the past week, we saw four big studies released. And this was a bunch of researchers who'd gotten access to data from Facebook during the 2020 election. And they tweaked people's news feeds in various ways to show them more politics or less politics, or we're going to show you reshares,
41:47 Casey Newton? or we're not going to show you reshares. And the big question that they took into that was, well, can you affect people's political beliefs based on what they're seeing in the news feed? Right? And that's the spirit of so many of these regulations is that what is in your timeline is destiny. What do they find? Very small effects for the most part. Now, these studies are limited in some ways. They have their flaws. We need to see more of them, and that's kind of a separate conversation. But when I see that sort of thing, I apply the lesson to something like COSA, which is do not believe for a second that if you just change the content on TikTok, that the teen mental health crisis is going to disappear. These are situations that are, there are just so many connected systems here.
42:31 Casey Newton? And while I hate that argument, because it feels like what I'm really telling people to do is to throw up their hands and do nothing. I also don't want people to get away with getting rid of a bunch of speech on the internet in this like phony pursuit of a utopia that is just never going to arrive.
42:47 Kevin Roose? See, I disagree. I don't think it is a phony pursuit of utopia. I think it is people who are freaking out about something that is. Really, really affecting kids and their mental health. Like, I think that the problem is real. I think that there are probably solutions that are more narrow and specific and less sweeping and broad than the ones proposed in COSA. But I don't think this is a fake problem. I think that, you know, when I see the tech companies lobbying to kill stuff like this, it actually makes me think, well, maybe there's something there because they wouldn't be worried if it was just, you know, a total garbage bill.
43:23 Casey Newton? Well, they're worried about the massive, like, legal bills that are gonna stem from it.
43:26 Kevin Roose? Of course, but like, I want these companies worrying about that. Companies that are building products that are aimed at kids, and let's be real, a lot of these companies have made explicit attempts to recruit teens and even pre-teens onto their services. They have been treating those teens historically as just a goldmine of engagement, of, you know, money to be made, of habits to be formed. It's been almost a predatory system. That is not how I want companies to approach their teenage users, as just a source of growth, a source of engagement, and a potential future customer base. Like, I want them to be a little scared when they design a platform that they know is
44:07 Kevin Roose? gonna be, you know, inviting to teens, and I want them to be more careful than they would be if they were just doing, you know, normal engagement hacking, growth hacking for their adult users.
44:17 Casey Newton? So I totally agree with that. I just also believe that 16-year-olds have speech rights, and I believe they have the right to consume a really wide array of content, and I don't always think that they should have to provide an ID because they wanna read or watch something. And that is the tension between those two things, is it's very hard to square that circle.
44:37 Kevin Roose? Yeah. Well, in conclusion, put us in control of Congress, we will write better bills than this.
44:43 Casey Newton? In conclusion, it's easier to build LK99 at home than it is to write a good internet regulation.
44:47 Kevin Roose? So true, so true.
44:54 Casey Newton? When we come back, we're bringing back the hats. It's time for HatGPT.
45:22 Speaker 00 Support for this podcast comes from Dell Technologies. It's time to put AI to work. Is your business ready? The challenge isn't getting started with AI. It's scaling beyond pilots while maintaining control of your data, governance, infrastructure, and costs. Dell AI Factory provides the integrated foundation you need to scale AI securely and efficiently with one trusted partner. Dell, AI on your terms. From idea to impact. Learn more at dell.com/yourwaytoAI. That's dell.com/yourwaytoAI.
45:56 Speaker 03 It's time to innovate at the speed of AI without compromising control. That's Outsystems. Take a look at what happened with one global retailer during a period of massive expansion. They used Outsystems to build AI agents to automate invoicing and translation processes, shifting 95% of their daily store operations to a new digital app, saving thousands of hours of manual work. Outsystems gives you the flexibility to deploy how you want and the governance to do it safely. Own your agentic future. Learn more at outsystems.com/NYT. That's outsystems.com/NYT.
46:31 Speaker 04 AI spend is growing fast, but most businesses still can't see exactly where that money is going. Ramp gives businesses visibility and control across AI providers with real-time guardrails by team and model, helping ensure spend goes to the highest ROI areas. Stop guessing where your AI spend is going. Get started with Ramp. And if you're new to Ramp, you can get $250 off at ramp.com/hardfork. Go to ramp.com/hardfork. Ramp.com/hardfork.
47:04 Kevin Roose? Casey, it is time again to play HatGPT. Oh, finally, I've been waiting for this.
47:15 Kevin Roose? As a reminder, this is our segment where we put a bunch of tech news stories into a hat, mix them up, pick them out one by one, and generate some plausible sounding language about them. And as a reminder to you, if one of us gets bored by the other person's talking, we can just say, "Stop generating."
47:30 Casey Newton? Beautiful.
47:31 Kevin Roose? All right. I have an actual hat today. I did remember to bring one. And we have our little slips of paper here with our stories.
47:38 Casey Newton? Here, I'll show you, Kevin. All right. So you do this. So we reach in. And then, let's see. First story, the BBC launches an experimental Mastodon server. So this is from The Verge, which lets us know that the BBC has launched its own experimental Mastodon server, marking one of the first major news outlets to establish an instance on the Twitter alternative. So Kevin, are you optimistic about the prospects of the BBC Mastodon server?
48:05 Kevin Roose? No. And no offense to the BBC. Love the BBC. Mastodon has zero juice. I'm sorry. I know you are a Mastodon head. You're a tutor, whatever the hell you guys are calling yourselves over there. I spent one day trying out Mastodon. I was like, this thing has zero juice. I think Mastodon, you know, I appreciate that they have blazed a path for decentralizing social media.
48:32 Casey Newton? They really did. They really did.
48:33 Kevin Roose? I appreciate that they came out with some ludicrous terms for what it is called to post on their platform, which is, I believe, "toot."
48:39 Casey Newton? It's a toot.
48:40 Kevin Roose? It's a toot. But Mastodon has zero juice, never has, and in my opinion, never will. I think right now, every media company in the world is trying to figure out what replaces Twitter. What replaces the role that Twitter used to have in our newsroom as a place where we could distribute our stories, as a place where our journalists could like talk about their stories and build their own audiences and direct people back to our website? I do not think Mastodon is that place. I don't think Threads or Blue Sky are either. And I just think media companies are going to be spending the next couple of years trying to sort of figure out what comes after Twitter, and the answer may be nothing.
49:14 Casey Newton? Okay. Well, you could not be more wrong about any of this. And here's why. So Mastodon uses the Activity Pub standards, and Activity Pub- Within a couple of years is going to be powering a huge part of digital media. So Tumblr, WordPress already are working on integrations with this. So in the future, media organizations are going to be able to have their own owned and operated little piece of the web. And it all starts with building your own little slice of the social media sphere. So is Mastodon full of very tedious and pedantic people? Yes, it is. Is it fun to use? No. But is the idea of a digital media organization trying to build its own part of the social
49:59 Casey Newton? web a good idea? Yes, it's a good idea.
50:01 Kevin Roose? It's a theoretical yes and an actual no, is what you're saying.
50:05 Casey Newton? We're going to have to see how this all shakes out. But I can tell you, the BBC is not the only media organization that is thinking about how can we use Activity Pub to take advantage of all the best parts of the social web, which are essentially promoting your stuff to a big audience in real time while getting rid of the walled garden nonsense that has infuriated so many people for so long.
50:24 Kevin Roose? I just love imagining like the BBC journalist who has spent their career like climbing the rungs of the UK's media world and you finally make it to the Beeb, as they call it. And you're a prestigious and respected outlet and the first day, your job, you get told that you have to toot. And it's just like, well, why did I climb that ladder? If tooting is at the top of the ladder, that's not a ladder I want to be on anymore.
50:51 Casey Newton? Yeah, I'm going back to one of those Rupert Murdoch newspapers where you just spy on celebrities. All right, pass the hat.
51:00 Kevin Roose? Mr. Beast sues his food delivery partner over inedible burgers.
51:05 Casey Newton? Oh, my God, I love this story.
51:06 Kevin Roose? Story was so good. So, OK, James Donaldson, a YouTube star, better known as Mr. Beast, who we've talked about on the Hard Fork Podcast very recently, has sued his partner in a food delivery business saying the company sacrificed quality in its bid for rapid expansion. So this is a lawsuit that was recently filed by Jimmy Donaldson, aka Mr. Beast.
51:26 Casey Newton? James to the court.
51:27 Kevin Roose? Against something called virtual dining concepts, which is sort of one of these ghost kitchen businesses. What they do is they essentially like, if you run, let's say, a Chinese restaurant, but you're looking for a little extra revenue and you have all the equipment that you would need to make burgers and sort of open up a virtual storefront, you could go to This Mr. Beast Burgers virtual restaurant and say, hey, I would like to start selling Mr. Beast Burgers on DoorDash and Uber Eats. And they would send you all the like ingredients and the packaging. And you, as the owner of the Chinese restaurant, could essentially open up like a stealth, secret storefront to sell Mr. Beast Burgers to your customers.
52:12 Casey Newton? And now, if you were gonna ask me, what is the worst way to make a hamburger? I think that's actually what I would tell you.
52:17 Kevin Roose? And there have been some amazing just online roasts of these Mr. Beast Burgers. In some of them were actually in the complaint, which is one of the funniest legal complaints I've ever read, because it's just like, the headlines on the complaint are all like, revolting, worst burger I've ever eaten. So there have been these sort of attempts to kind of track down where these Mr. Beast Burgers are actually being made. Because like, to be clear, these Mr. Beast Burgers are not being made at restaurants that Mr. Beast owns or has any real connection to.
52:47 Casey Newton? No, Mr. Beast has absolutely no, no idea what is going on with this empire of restaurants that he built under his own name.
52:54 Kevin Roose? Right, and so that would be fine if these burgers were actually good, but it turns out that a lot of them are actually quite horrible. There's very little quality control at these restaurants. And some of them are actually, one was someone who actually Googled the address that his Mr. Beast Burger came from, and it was a 7-Eleven. So 7-Eleven is making Mr. Beast Burgers in some places.
53:14 Casey Newton? They were making them up like that's the same rotisserie where they just have that hot dog that's been spiraling for like 16 years.
53:18 Kevin Roose? Right, so Mr. Beast is now trying to get out of this contract with virtual dining concepts, so he no longer has to slap his name and likeness on these terrible burgers.
53:29 Casey Newton? Yeah, let me say something. Yeah, we have a lot of celebrities and influencers that listen to this podcast, Kevin. And one thing that happens when you get famous is that people come to you with these money making schemes, and they say, you're gonna about to make the easiest money you've ever made in your entire life. All you have to do is sign away your name and likeness, and the money will start rolling in. And I hope all of you fellow influencers and creators and celebrities out there, take the lesson of the Beast Burger, which is, you have to have the quality control standards in place, or otherwise, people are gonna have diarrhea.
54:03 Kevin Roose? Wise words. And I will say that if you are interested in opening up a Hard Fork Burger, I mean, the fork is right there in the name. We're interested in partnerships and sponsorship opportunities. Please get in touch. But but also have like have a certificate from the health department, please, and know how to make food. Stop generating. OK.
54:22 Casey Newton? Now it's my turn.
54:25 Casey Newton? OK. Meta prepares chatbots with personas to try to retain users. This is from the Financial Times. Facebook owner Meta is preparing to launch a range of artificial intelligence powered chatbots that exhibit different personalities as soon as next month in an attempt to boost engagement with its social media platforms. These people said some of the chatbots, which staffers have dubbed personas, take the form of different characters. The company has explored launching one that emulates Abraham Lincoln and another that advises on travel options in the style of a surfer. So I guess my question, Kevin, is who are you more excited to talk about on Facebook? Abraham Lincoln or someone who advises on travel options in the style of a surfer?
55:08 Kevin Roose? I wish that I could combine them and get travel advice from Abraham Lincoln.
55:13 Casey Newton? It's like, well, I'm gonna tell you one theater in the Washington area to avoid.
55:16 Kevin Roose? Do not go to the fourth theater. Avoid zero stars.
55:27 Kevin Roose? Is it too soon? It's too soon.
55:29 Casey Newton? It's too soon.
55:29 Kevin Roose? Oh, God. So I read about this story in a little newsletter called Platformer this week. Yeah, it's a good one. You had a good post about this. And it is really interesting to me that Meta is sort of plowing ahead with this plan that, from what I can tell, zero actual people have expressed an interest is. Like, I understand that they are, you know, trying to sort of jam generative AI into as many of their products as they can. They're also doing this like open source strategy with their language model, Llama, which is very interesting, and we've talked about. But I do not get this personas thing at all. So can you just explain to me what they think the appeal of this is going to be and why they think this is going to sort of boost engagement on their products?
56:11 Casey Newton? Yeah, so the big theory here is that you are going to interact with many different kinds of generative AI personas in the future. So you might have an AI Spanish tutor, you might have an AI fitness coach, you may have an AI therapist, and Meta wants to get to work building some of those now. But those things aren't really ready for primetime. And so what we have instead are these chatbots that can mimic Abraham Lincoln or a surfer with travel tips. And so that's what they're rolling out. There is some evidence that people want things like this. So you know, the company Character AI, this is essentially what they do. You can go to character.ai right now, login with your account, and you can start
56:52 Casey Newton? chatting with any number of fictional or nonfictional people. And there is a pretty vibrant community on Reddit of people sharing their conversations that they're having there. So it's clear to me that there is some kind of appetite for this, but I don't know. When I've used the Character AI bots, it just feels like a total passing fad to me. There is not a lot of long-term benefit to me there. You know, I was writing in the newsletter this week, like, what is like, the long-term value of an Abraham Lincoln bot? Like, I get it if you're writing a school report or something, and you want to ask him, like, hey, like, who killed you? Or whatever. But like, beyond that, I don't really get it.
57:31 Kevin Roose? I would like an Abraham Lincoln bot that could give me the weather, but like, using Abraham, like, it's gonna be fourscore and two degrees up today. Might want to put on sunblock.
57:42 Casey Newton? So, look, I think Meta has a lot of technology and not a lot of idea about what to do with it at the moment. I think one reason why they're giving all this stuff away is that it's not clear to them, like, what exactly are you gonna put in?
57:55 Kevin Roose? Right. They don't make office productivity software that can, you know, be augmented with AI to like, make workers more productive, or save businesses a lot of money. Like, they make social media products. I guess my curiosity around this is, for many years, social media companies have been trying to get bots off their services, right? It's not good for them if some percentage of their engagement is coming from these like, artificial sources. So why is this different than that?
58:21 Casey Newton? Well, because those bots are created by others.
58:24 Kevin Roose? They're like, we want to own the bots.
58:25 Casey Newton? Exactly. Meta wants to own the bot. I mean, what Meta wants is eyeballs to be on Meta products. And if bots bring eyeballs to Meta products, then it will build them.
58:32 Kevin Roose? Okay. Next one. Stop generating.
58:37 Casey Newton? Greg Ruckauski was removed from Stable Diffusion, but AI artists brought him back. This is from Decrypt. So there's this digital artist that we are both familiar with his work, Kevin, Greg Ruckauski.
58:48 Kevin Roose? Yeah, we've talked about him on the show before.
58:50 Casey Newton? Yeah, he is one of the most popular artists in the AI world because he makes these sort of sweeping fantastical creations. And so when people are using these text to image models, they love to invoke his name because they will get such cool images back.
59:03 Kevin Roose? This was sort of like a hack or like a cheat code almost that people figured out when these sort of mid-journey, stable diffusion, Dolly image generators were in their infancy is like, if you wanted to create like a very cool, you know, image of a dragon, you, all you had to do was like, insert Greg Ruckauski's name into the prompt because like, he was the best person out there at making these like fantasy drawings. And so he would just make it look like one of his things, which is great if you're trying to make cool art. It's not so great if you're Greg Ruckauski.
59:34 Casey Newton? Yeah, and so Greg, which I'll just call him Greg for ease of pronunciation, doesn't want anything to do with any of this. And so he says, hey, like, take me out of your image generator. And so Stable Diffusion, one of the companies that makes these text image generators, responds by removing his work from their data set. But the community has now created a tool to mimic Ruckauski's style against his wishes. And since Stable Diffusion is open source, there's nothing he or Stability AI can do about it, says Decrypt. So Kevin, what do we make of the fact that even when you're an artist and you say, I don't want to be part of this, and the company that makes the AI image generator takes you out of it, eh, it doesn't matter, you're still going to be in it.
1:00:11 Kevin Roose? It sucks, man. I mean, I'm thinking about this from Greg Ruckauski's perspective. He was just a normal, like, working artist, doing his job, making great art that people loved, had like a good, solid career as an artist, which is like a hard thing to pull off, you know, in any era, but especially this one. And now, you know, his reward for that is that when these new AI image generators get invented, like, he sort of becomes this like ghost in the machine that people are just sort of using to conjure up these things that look exactly like the stuff that he drew on his own. And like, on one hand, a lot more people know about Greg Ratajkowski now than did a few years ago. We're talking about him. So it clearly hasn't been all bad for
1:00:52 Kevin Roose? his career. But I do sympathize with his frustration, which is like, I don't want to be like the test case guinea pig for all these AI image generators. And I worry about his ability to like, make a living when now anyone can, you know, just put his name into an image generator and pull up something that looks vaguely like he might have done it. So I do think that concern is real. At the same time, I think this sort of exposes the futility of these artists and content creators trying to sort of get removed from the training sets of these things.
1:01:26 Casey Newton? Yeah, I agree with that. I would also say we're early enough in this journey that it would be great if companies could build tools that effectively do remove artists or creators or journalists who do not want their work to be part of the training set. I do think that people should have the right to remove themselves from the training data. All that said, my curiosity is, if we talk to Greg Ratajkowski five years from now, is he making more money than he did before he became one of the big stars of AI text image generation?
1:01:55 Kevin Roose? I bet he will be, and I bet, but I bet he will still be pissed off that all this is happening.
1:01:58 Casey Newton? Wow, that sounds like you know him personally.
1:02:00 Kevin Roose? Yeah, Greg, call a guy.
1:02:03 Casey Newton? All right, stop saying you're hating.
1:02:04 Kevin Roose? Okay, next one. Twitter removes X on headquarters. This is from the New York Times. Workers on Monday dismantled a giant X sign that was only briefly displayed on the roof of Twitter's headquarters in San Francisco after residents complained and the city issued a violation notice for lacking proper permits, according to local officials. The sign was installed on Friday to reflect the company's new branding and spurred immediate concern. Over the weekend, the city received 24 complaints, which included concerns about its structural safety and flashing lights. One complaint described extremely intense white stroboscopic light that was causing distress and nausea. Casey, what do you make of the giant X being taken down from Twitter's roof?
1:02:45 Casey Newton? I mean, I love this story because it goes to show you what the actual most powerful force in all of San Francisco is, which is NIMBYs. If you want to build something in this town and you don't have the proper permitting, you are about to experience the full force of the state, okay? You're gonna have more people beating a path down your door than you thought was possible. And so Elon Musk, who has been able to wield his authority over so many people and get everything that he wanted, ever since he marched into Twitter headquarters with that sink, finally found the one force that he could not overcome, and it was the San Francisco NIMBYs.
1:03:20 Kevin Roose? Yes, Elon Musk is the most powerful man in the world, except when it comes to building things in San Francisco, at which point his power ceased to become meaningful.
1:03:26 Casey Newton? But you know, if I could say one more serious thing, it's that last week I talked about my theory that Elon Musk is a cultural vandal. And to me, this is more cultural vandalism. It's like, I'm literally just gonna put a giant, flashing, disgusting X on the top of this historic building as a kind of symbolic middle finger to the entire city. Ha ha, lol. Like, it is just so in keeping with his character.
1:03:52 Kevin Roose? Did you actually see it when it was, during the brief period?
1:03:54 Casey Newton? No, and here's why. This thing was so tall, and it was held down by sandbags, or at least they appeared to be sandbags from the overhead photography. And I thought, I'm willing to sacrifice a lot to cover the story of Elon Musk, but I am not gonna die when a stiff wind blows that thing off Market Street and onto my head.
1:04:13 Kevin Roose? It did look not that stable. It was also like, it was a choice to make it flash, you know, stroboscopic lights. It was, it was a choice to not have it just be a lit up X, but to have it be the most annoying version of an X possible.
1:04:26 Casey Newton? The only thing it didn't do was like, sound an air horn every time the lights flashed, but they probably just didn't have the time to install that yet.
1:04:34 Kevin Roose? Don't give many ideas.
1:04:35 Casey Newton? Stop generating.
1:04:37 Kevin Roose? All right, your turn, last one.
1:04:40 Casey Newton? Bop bop.
1:04:42 Casey Newton? Uber turns its first ever operating profit. This is from the Wall Street Journal. The results for the three months through June were driven by solid growth in both of Uber's core businesses as the number of rides in the US and Canada surpassed pre-pandemic levels for the first time. And demand for delivery stayed strong despite restaurant reopenings. This was the first quarter that Uber was profitable since it was founded in 2009. So Kevin, in the very early days of this show, we did a segment where we wondered whether Uber would ever turn a profit. And I don't know at the time that either of us thought that it would have been like in 2023.
1:05:19 Kevin Roose? Yeah, I mean, this is sort of a fascinating story to me because for so long, we have heard so much about these venture-funded startups that just sold products or services at a loss, essentially. Used their venture funding to subsidize prices to sort of flood the market and drown out competition. And did that very successfully, but also lost billions and billions of dollars in the process. And for many years, Uber was one of those companies. And there was a huge question about whether it actually would ever be able to turn a profit. And now it has and may keep turning profits into the future. So do you think this is sort of like a vindication of this strategy that all these companies had of like
1:06:01 Kevin Roose? getting your prices as low as possible by pouring venture capital money into the market? And then sort of jacking up your prices later so that you can actually turn a profit?
1:06:12 Casey Newton? Basically, I do because even when it was at its most unprofitable, no part of me ever thought that Uber or something like it was going to go away. Because it is just unimaginable to me now that you won't be able to be in a relatively big city and tap a button on your phone and a car appears. People desperately want that. And as we've learned over the past year, they will pay higher prices in order to get that service. Maybe than they were used to paying when there was all that venture capital money in it. So I think it's a pretty big achievement for the Uber folks that they were able. To get there, we'll see if they can keep it up. I would say that surprising to me, Lee, large part of their business is Uber Eats. And I think to the extent that there are actual profits there,
1:06:54 Casey Newton? it is much more about being a delivery business than a rideshare business. But, you know, one thing that they always said at Uber was if they could get you a car at the touch of a button, they could get you a lot of other things. And so I'm just curious to see whether there is more money for them in delivery and related businesses.
1:07:10 Kevin Roose? It's a little depressing to me, though, honestly, because part of what the strategy was all along, I believe. And I've talked to a lot of the venture capitalists who funded a lot of these businesses in the 2010s that were sort of using this model of like subsidize the prices of the goods and services. A part of that strategy was always to kill the competition, right? If you're Uber, you want to get your prices so low that you drive every taxi company out of business. If you are Airbnb, you want to make your prices low enough that you kill the hotels so that your only choice is to go with Uber or Airbnb, even if your prices go up. And that is what is happening with Uber. We now have Ubers that are so, I don't know if you've looked at your Uber receipts recently. They're so expensive. It's so expensive to get an Uber.
1:07:54 Kevin Roose? It's more expensive than taxis were. But it's not like we're going back to taxis. Like, are you actually going to call up a taxi company and say, please meet me on this corner? No, absolutely. You're not going to do that.
1:08:04 Casey Newton? No, also taxis were horrible. Like, the entire system was rigged and it was bad. You know, I'm sorry for a lot of the individual cab drivers, but my God, I'm not sorry to see the taxi industry be disrupted.
1:08:15 Kevin Roose? Right. So now it seems like the lesson of what is happening at Uber is that this sort of subsidy strategy, where you lower your prices, you flood the market, you drive your competitors out of business. And then when you have a monopoly, you jack up your prices again to turn a profit. It actually does work if you're willing to wait like 14 years and also burn hundreds of billions of dollars in venture capital in the meantime.
1:08:36 Casey Newton? Yeah. So don't be surprised when in 14 years, Hard Fork starts charging you a monthly subscription fee. Why do you think it's free now? It's all part of the plan. You know, I want to say one more thing about Uber, which is I think that there's another secret to their profitability, which is that no company has ever sent more push notifications in the history of the world, okay? You have to leave the notifications on to know when the driver is at your house.
1:08:58 Kevin Roose? They text me more than my wife.
1:08:59 Casey Newton? Seriously, every time I look down at my phone, it's like, hey, want to rent a car this weekend?
1:09:04 Kevin Roose? Would you like a Mr. Beast burger?
1:09:08 Kevin Roose? All right, stop generating. Wow. And that was HATGPT.
1:09:23 Speaker 00 Support for this podcast comes from Dell Technologies. It's time to put AI to work. Is your business ready? The challenge isn't getting started with AI. It's scaling beyond pilots while maintaining control of your data, governance, infrastructure, and costs. Dell AI Factory provides the integrated foundation you need to scale AI securely and efficiently with one trusted partner. Dell, AI on your terms. From idea to impact. Learn more at dell.com/yourwaytoAI. That's dell.com/yourwaytoAI.
1:09:57 Speaker 03 Companies are pouring money into AI, but few leaders can explain how those AI investments work with the systems you already have. You know, the ones that actually run your business? With OutSystems, you don't have to start over with AI. You can choose the projects where AI will have the most impact. And OutSystems gives you the platform to make it agentic on your terms with complete control every step of the way. So you can trust AI with the mission-critical systems that run your business. Own your agentic future. Learn more at outsystems.com/NYT. That's outsystems.com/NYT.
1:10:32 Speaker 04 AI spend is growing fast, but most businesses still can't see exactly where that money is going. Ramp gives businesses visibility and control across AI providers with real-time guardrails by team and model, helping ensure spend goes to the highest ROI areas. Stop guessing where your AI spend is going. Get started with Ramp. And if you're new to Ramp, you can get $250 off at ramp.com/hardfork. Go to ramp.com/hardfork. Ramp.com/hardfork.
1:11:05 Casey Newton? Hard Fork is produced by Rachel Cohn and Davis Land. We're edited by Jen Poyant. This episode was fact-checked by Will Peichel. Today's show was engineered by Alyssa Moxley. Original music by Dan Powell, Alicia Bietup, and Marian Lozano. Special thanks to Paula Schuman, Hui Wing Tam, Nogalod Lee, Kate LaPreste, and Jeffrey Miranda. You can email us at hardfork@nytimes.com. If you've made a room temperature superconductor at home, we want to hear about it. Please send us a sample.
1:12:24 Speaker 04 You can't undo everything in life, but you can undo prediabetes. More than 2 in 5 adults in the US have prediabetes. Most don't know they have it, but with early diagnosis and action, you can delay or even prevent type 2 diabetes. Visit doihaveprediabetes.org to take the one-minute risk test and start undoing the things you can. Brought to you by the Ad Council and the Centers for Disease Control and Prevention.
Previous Elon's X Machina + Crypto Orbs + A Visit to Google’s Robot Lab Next Don't Scrape Me, Bro + The Activists Sabotaging Self-Driving Cars + How Reddit Beat a Rebellion Page updated October 12, 2026 · transcript and summary are machine-generated