Don't Scrape Me, Bro + The Activists Sabotaging Self-Driving Cars + How Reddit Beat a Rebellion 01:11:17

Welcome to the new era of PCs, supercharged by Snapdragon X Elite processors. Are you and your team overwhelmed by deadlines and deliverables? Copilot Plus PCs powered by Snapdragon will revolutionize your workflow. Experience best-in-class performance and efficiency with the new powerful NPU and two times the CPU cores, ensuring your team can not only do more, but achieve more. Enjoy groundbreaking multi-day battery life, built-in AI for next-level experiences, and enterprise chip-to-cloud security.

Give your team the power of limitless potential with Snapdragon. To learn more, visit qualcomm.com/snapdragonhardfork. Are you a cone guy or a cup guy when you get ice cream? Typically,

because a cone just feels too indulgent. But if you were to ask me, what is the best way to eat an ice cream cone? Obviously, it's in a cone. Too indulgent? You're eating ice cream. Well, that's the thing. It's like if you put it in a cone, then it's a hat on a hat. Oh, no, no, no, no, no, no. You are fooling yourself, my friend. You're engaging in a moral calculus that says if I have this thing in a sad cup...

with a sad spoon, then I'm not fully being bad. I'm being half bad. - No, do you know how many times people order a double cheeseburger with fries and a Diet Coke? Like that's what this is. - But that's self-deception. - Of course. - Get the cone. - Most of my life is self-deception. - Allow yourself to be happy.

I'm Kevin Russo, tech columnist at The New York Times. I'm Casey Newton from Platformer. And you're listening to Hard Fork. This week on the show, a growing backlash to AI that has more people saying, don't scrape me, bro. Then, organizers from the San Francisco activist group Safe Street Rebels stop by to tell us why they're disrupting autonomous vehicles by putting traffic cones on their hoods. After that, how Reddit crushed its user uprising. And finally, an update on LK99.

It's not super. Casey, we have talked on this show about generative AI and how many artists and writers and creative people are starting to get a little freaked out by all of these tools, some of the legal challenges to the companies that are making this stuff. And it has sort of been on a slow boil. But I would say this week, we really saw the first traces of

of an organized counter-movement to the use of generative AI in creative industries. Yeah, the backlash is popping up in many different places at once. So you suggested calling this segment Don't Scrape Me, Bro, which I love because it is like what I would call the first shreds of this counter-movement, which is like objecting not just to the existence of

generative AI and the use of generative AI in creative industries, but to the actual process by which these models are built and trained. Yeah, people want to have some say in how their data is being used to train this stuff. And the more that they learn about that, I think the more questions they're asking. Right. And this is something that, you know, in the early days of generative AI, like companies would tell you a lot about how they got the data to train their models. You know, they were

proud of how clever they were being going out and scraping all these websites to get this data to feed into their neural networks. And then, you know, as they sort of lawyered up and realized that people were going to be mad about this, they have stopped saying nearly as much about how they're actually getting this data. Yeah, I think ChatGPT's FAQ page just says that the data fell off the back of a truck, which was interesting. Yeah, how'd this end up here?

So this week we had, I would say, three incidents that I think illustrate what is going on in this war over AI data use. And I want to talk about them together because I think they're all related and I think they point to what is going to be a huge and long fight.

And they all involve issues around crawling and scraping. Now, Casey, what do you know about crawling and scraping? Well, if I have too many drinks at the club, that's how I get home. No, this is, of course, the sort of automated systems that gather data from the web. So if you've used Google's search engine, for example, that crawls the web. When somebody makes a new web page, for the most part, Google indexes it. And so that way, if somebody searches for it, they can find it.

Right. So crawling and scraping, they're things that internet companies have been doing forever. Basically, crawling is when you have a bot that travels around the web just visiting every link it can find. Scraping is related and sometimes goes hand in hand, which is that you actually go download some information from a website and use that for your own purposes. Yeah.

You know, it might be worth saying that scraping has been controversial. There are bad actors out there that will try to scrape websites for purposes that we don't like. So, for example, people try to scrape Facebook all the time to gather personal information about people for nefarious purposes. So Facebook has to build defenses against that. There's a case that went all the way to the Supreme Court involving LinkedIn with some firm that was trying to scrape LinkedIn. So scraping is not without controversy. Right. So I would say crawling is pretty controversial.

Very standard. Widely accepted standard. Scraping is not, and it's been a lot of controversy around that over the years. Yeah. So let's start this week with the story of Zoom. So Zoom, of course, is a tool that we used in 2020 to go to the worst birthday parties of our entire lives. Exactly. And play those, what were those games called that you could play? I refuse to say. No, it's a trauma response I'm getting when you mention those. Okay.

So Zoom, basically, they have some generative AI features that they have been experimenting with, including sort of automated meeting notes. If you're having a Zoom call, you can get like a text summary of what you talked about on the call. Right, it'll be like Kevin had his camera off the whole time, didn't even seem to be paying attention. Yeah.

Right. So this week, the publication Stack Diary ran a story basically calling attention to this update in Zoom's terms of service that appeared to give the company very wide latitude to sort of collect customer data from video calls to train its AI models.

And it didn't really specify what exactly Zoom was doing with that data, but it did sort of open the possibility that they were taking the contents of customers' video calls, scraping them, and feeding them into their AI models in some way. And this report set off a real firestorm. People were so mad. I saw people saying they were canceling their Zoom subscriptions. They were never going to use Zoom again. They were going to go back to using WebEx or whatever the hell we used before Zoom. Not Citrix. Not BlueJeans. Right.

So, you know it's serious when people are threatening to use WebEx. So...

After this, Zoom sort of responded. They said, you know, we're updating our terms of service to clarify that we are not using customer content to train our AI models without people's consent. Basically, like, if you don't mind us using this information, we will ask you to opt in. But if you don't want us to use this to train our models, we won't use it. Yeah. So very quick response from Zoom. People were very mad. What did you think of this? Well.

Well, I think people are right to be suspicious here. When you think about the kind of things that people do on Zoom, a lot of its paying customers are businesses. They involve sensitive conversations among coworkers, between businesses, people negotiating deals. And the idea that the contents of those meetings were being fed into an AI database and then could just be used for literally any purpose from now until eternity,

That's a really scary thing. So it makes sense that people called Zoom out, and I'm glad Zoom walked it back as quickly as it did. Yeah, I mean, I think the terms of service of a lot of major internet platforms actually already allowed them to do this kind of thing because they're written in such a broad way. You know, we may collect data for the purposes of improving our tools or our systems that would technically allow them to feed them into AI models.

So I think what happened with Zoom is going to happen with a lot of other companies. I think any company that sort of runs, frankly, any platform where customer data is being handled or transmitted, even in something as simple seeming as a video call, I think there's going to be a lot of attention paid when those terms of service change in a way that makes it more explicit that they're reserving the right to train an AI model using your face. Yeah, but good for everyone who complained about this because this was something that Zoom needed to address. Yeah. Okay. Okay.

So second story this week about the war over scraping is about a website called Prosecraft. So Casey, have you ever used the website Prosecraft? I have not, but I've been reading up on it and it does sound kind of interesting. Yeah, I would assume that you've never used it because your prose is not very well crafted. So this is a website created by a guy named Benji Smith.

And, you know, he says that he made this while he was trying to write his own memoir and that he was basically trying to sort of answer questions like how many words are in the typical book or, you know, what percentage of verbs are active versus passive in this classic work of literature or that classic work of literature. Yeah. And apparently he has like a background in computational linguistics. So like this has just been a sort of

pet interest of his for a very long time. Yes, and he also has a company that makes software for writers. So this was sort of in his professional and personal interest to do something about this. So he makes this website called Prosecraft, which he claims is dedicated to the linguistic analysis of literature. And we don't know exactly how he put this together, but essentially he got more than 25,000 books

And he fed these books into a kind of analytics tool, which allowed users to analyze them in a statistical way, like saying, you know, how much passive voice is used, you know, in Alice in Wonderland or which of the Harry Potter books has the most vivid writing, something like that.

And this went mostly unnoticed because this is a very small website run by a very small company. But this week, authors figured out that this was happening. Writers like Maureen Johnson, who's a big young adult fiction writer, Celeste Ng, who wrote Little Fires Everywhere, and Hari Kunzru objected to this, basically said, how did this website get the contents of my copyrighted books without my permission? And why is it using them for this kind of like statistical analysis? And so,

And basically pressured Benji Smith to take down the website. And in just a few days, that's what he did. He wrote in a blog post that was very apologetic. He said, Yeah.

So on this one, I have to say, I think that the reaction was a little bit overblown. Totally agree. Like, when you look at what he was doing, he was not allowing you to really do anything with the contents of these books other than to get these sort of short snippets via these little analyses. Yeah.

And, you know, it's like he was basically like showing you like word clouds, you know, like this was not the sort of use that if I had written a book that I would object to. And you've, of course, written books. And my guess is that you wouldn't really object to your book being part of this either. No, it's fine. And in fact, I've been asked, like, can I do something like this with your book? I'm, you know, building a database or something. And it's generally been fine with me, you know, in part because like what he was doing was not

generative AI. He was just basically feeding these books into a database and then running some very basic analytics on them so that you could see how many words should my book have or what's the right number of LY adverbs to have in this paragraph. So it wasn't really even

generative AI. He was not training a large language model on these books. He was basically just sort of giving you the chance to zoom out and like find some patterns in very famous works of literature. Yeah. Now, of course, at the same time, I do think that once people acquire a huge amount of data at some point, either they themselves or someone else might say to them, you know, that's actually pretty valuable and you could use that in a different way. That is actually basically what happened to Reddit over the past year, right? Where somebody said, hey, all this data on your website, that's

that would be really valuable if you actually sold it to these companies that are making large language models. And now I think Reddit's extremely interested in that as a business model. So the risk here, I think, was not so much what Benji was doing already. It's what he might be doing in the future. And because we don't have good data privacy laws, and because I think copyright law is still kind of unsettled on what

constitutes fair use in a generative AI context, there was some risk here. And I guess it's not surprising to me that at least some authors said, hey, get me out of that database. Right. I mean, it does feel a little strange and sort of misdirected all of the anger toward this one, you know, very small website. When the real enemy is capitalism. Yeah.

No, I think the real enemy, if you're an author who worries about AI, is like the companies that are training the large language models that can generate text that sounds like you wrote it and can fill up Amazon with like books that, you know, sound like you wrote them, but actually make money on that. So I think this was a little bit of misdirected anger, but I do think it is a symbol of how on alert people are, especially writers and other creative people we've

we're seeing this in Hollywood too. This is a place where a lot of people who write for a living are really having their first experience of being radicalized against some kind of technology that they feel presents a threat to their livelihood. Yeah, and also just that they are creating value that they are not capturing in any way, right? Yeah. So, but now it's,

down the website's been taken down benji smith has said that he may rebuild it in the future using permission from the writers and that's a good idea like that's what i will say even though i think the reaction was overblown i think if you wanted to restart this project and say i'm only going to include works from artists who've either consented or the many many thousands of books that are already in the public domain that's probably a good way of going about it right if you want to learn about books published in the 1800s um you can definitely go do that that is uh

you know, within bounds. How much of little women use the passive voice? I've always wanted to know. Okay. Third story of this week about the war over scraping is that OpenAI has now given people who run websites the ability to block its web crawler. This is sort of like a fun thing that harkens back to my days as a GeoCities webmaster. Have you ever

run across something called robots.txt? I've used robots.txt, in fact. Really? When I built my own website, sure. To do what? Yeah, you would make sure that Google could index it so that you could be in search results. Right, so robots.txt is basically a file that every website has that's basically like a set of instructions for the robots that are crawling that website. Where to go and where not to go.

Yeah. And now OpenAI has released a feature that lets you block their web crawler with just like two lines of code. And once you put this little two lines of code on your website, OpenAI can no longer scrape your data and they won't be able to use that to train their models. Right. And of course, this is a good thing. And I imagine...

A lot of the same sort of people that would get mad about Prosecraft will probably be adding this to their robots.txt files. At the same time,

OpenAI could have built this bot before it downloaded the entire internet, right? And so could any of these large language models. So there is a little bit of a too little too late argument against this kind of thing. Right. It's like, you know, someone ransacked your house and then they're like, well, you know, you could put up a security camera now. You know, I already have it.

Actually, we'll give you the security camera now. Yeah, yeah. So, you know, the cynical take on all this is that open AI can do this now because it has already crawled and scraped most of the open internet. And that essentially what it is doing now is just giving people sort of a false sense of security. Yeah. And so there has been some discussion this week about, well, for all the reasons that you said, how much does it matter that people can now do this when the entire internet has already been downloaded? And...

My curiosity is just around the fact that language evolves, right? The way that we talk to each other, the things that we talk about, how we might answer a question. When you just think about the slang that has leapt off of TikTok in the past year, you know, a year ago, nobody was saying Riz. And now Riz is everywhere. I always had Riz. You always had Riz, but you didn't call it that. Before there was even a word for it. Before Riz.

We called it Riz. We called it Mojo. And so there's a chance that if everyone added the chat GPT bot and you said, hey, what is this really charismatic person? Like, how would you describe him? And they would say, well, he's got a lot of Mojo. And you would be reading that in 2026 and just be like, I think he means Riz. Yeah, and all the teens would be laughing at you. And the teens would laugh. And everyone knows the most hurtful laughter there is is the laughter of teens. Correct.

My point here, and I do have one...

is that in order for these companies to have language models that service in the ways they want to, I do believe they are going to need ongoing access to new material, right? It's not just going to be about how did humans sound when all this data was originally being scraped over the past few years. It's going to be about how do they sound today. And for that reason, I think they're going to be incentivized to get as many sites as they can not to use this. Yeah, I mean, it's telling that they're not saying, you know, we're going to stop crawling.

or stop scraping by default. It's telling that they're forcing people to like put these two lines of code in their website if they don't want to get scraped.

So I think there still is going to be ongoing scraping, but I do think that it's become a much higher stakes proposition to go out and scrape a bunch of websites because the people who run those websites might get real mad. And now there's sort of more awareness that this is going on in the background. And so I think if you are an AI company, the best time to sort of gather your big data set was before everyone started getting mad. I think it's going to be a lot harder now. Yeah. Yeah.

Casey, why do you think this sort of anti-scraping movement is having a moment now? Is it just because people are learning more about how these tools work and the fact that they need to go out and ingest a bunch of data? Or is there something else happening?

I think that people have a pretty good intuitive sense about what constitutes a good trade of value between themselves and a tech company. So when you look at the previous generation of services, something like a Gmail, they

something like a Google search. The agreement was, hey, I know you're gonna be collecting a lot of data about me, but in exchange, I am going to get a really good free email service, I'm gonna get a pretty good web search service. And so I think people were mostly okay with that exchange. Same thing with a social network, right? Yeah, Facebook, you're gonna collect a lot of my data, but you're gonna let me keep in touch with all of the people I've ever made eye contact with and provide you with entertainment throughout the day, so it's okay that you wanna show me ads based on what you know about me.

I think with these AI tools, the value of the exchange is a lot less clear for a couple reasons. One is the fear that this data is going to be used to automate away a lot of jobs, potentially including the job that you yourself have as the user, right? So I think people are nervous about that. And then two, a lot...

of the large language models that we've seen today have this paid component where a lot of the AI tools that are on the market today want to charge you a subscription fee. And so now we're in this world where you're constantly giving up data often without your express permission to a large language model, which then wants to turn around and sell it back to you for $20 a month. And I just don't think that those things in combination feel like a fair trade to the average consumer.

I think that's right. I really like this theory because I think we do sort of sense as consumers when the trade has gotten less fair. I mean, I used to get a lot of questions about like, you know, why is everyone mad at Facebook over collecting data for ad targeting when Google does the same thing, but, you know, with your email?

And my response was always like, well, with Google, like you get Google Maps out of that. You get Google search. You get Gmail like you get all these free things. And so, yes, they're collecting your data. And on some level, you know that and it bothers you. But you also kind of like quiet that voice in your head because you're like, well, I'm getting all this free stuff.

With Facebook, I think that argument started to break down because all of a sudden people felt, you know, I don't know when this was, but like, you know, sometime in the last five years, people sort of said, well, I'm giving away all my data to this thing that is making me sad and angry when I open it up. Well, in particular, not just that, but the backlash started in 2017 when people thought,

the exchange is I'm giving them all of my data and people are using that to destroy our democracy. And that's not a fair trade. So that started to feel like less of a fair trade. And then so that it became like, well, what leverage do I have as the consumer? Can I delete my account? Can I restrict the information it gathers in some way? And so,

I think that's what we're starting to see with these AI models. But now I think you're starting to see not just people, but entire websites and corporations saying, well, wait a minute, this isn't a fair trade. It doesn't feel so fair anymore. And we want to actually be compensated for that. Absolutely. You know, particularly when there have been these cases of AI plagiarism where, you know,

even in a model that is supposed to be generating ostensibly original content based on the inputs that it's getting, people have been able to tie a lot of keywords and phrases to specific blog posts on the internet that were not that hard to find. And so again, if somebody wants to charge me a subscription fee so that I can read a free blog post, it's just not a fair exchange. So, okay, my other question about this anti-scraping backlash is, like, does it matter, right?

These companies that are building the largest language models on the planet, they are ingesting trillions of words worth of data. They are ingesting huge percentages of the internet. And, you know, something that you'll hear if you talk to them, because I've asked people at these companies, you know, what happens if people start, you know, saying, don't index my site or don't scrape my data? And they basically say like,

this doesn't really matter because A, we've got so much already and B, you know, the number of people who are going to object to that and pull their data out of the training sets if we allow them to is so small compared to the overall internet. See, that just smacks of hubris to me. It's like these people have their wax wings on and they're flying into the sun if they think that is the case. Like we...

already saw with Zoom that, yes, this stuff matters, that you can experience a massive customer backlash within a week that threatens your entire business, right? And then I think this idea that essentially, neener, neener, we've already downloaded the internet. There's nothing you can do. I don't think it's true. I think they need ongoing access to new material. And when they're telling you that, they are being extremely self-serving and short-sighted at the same time.

I like the idea that in the future there will be like two versions of the AI models, one that has been kept up to date with the latest teen slang and the other that is like the dad model that is just hopelessly stuck in like the year 2021. Like if you if you want to say talk in a way that's going to get you respected by your kids, you got to pay up for the new model with fresh data. Yeah.

Otherwise, they're going to be saying Riz in the year 2040, and won't that be embarrassing? Well, also, I think a lot of older people would probably like talking to a chatbot that reminds them of their youth. That's true. Yeah, a lot of our language models have been stuck in the past for a long time. You said the word mojo just now. I mean, that was going to update your model. When we come back, we'll talk about another tech backlash that is brewing. This one's about self-driving cars.

Beep, beep. Toot, toot. Your car is broken. It goes toot, toot. This podcast is supported by KPMG. Your task as a visionary leader is simple. Harness the power of AI. Shape the future of business. Oh, and do it before anyone else does without leaving people behind or running into unforeseen risks.

Simple, right? KPMG's got you. Helping you lead a people-powered transformation that accelerates AI's value with confidence. How's that for a vision? Learn more at www.kpmg.us.ai.

- Hello, this is Yuande Kamalefa from New York Times Cooking, and I'm sitting on a blanket with Melissa Clark. - And we're having a picnic using recipes that feature some of our favorite summer produce. Yuande, what'd you bring? - So this is a cucumber agua fresca. It's made with fresh cucumbers, ginger, and lime.

How did you get it so green? I kept the cucumber skins on and pureed the entire thing. It's really easy to put together and it's something that you can do in advance. Oh, it is so refreshing. What'd you bring, Melissa?

Well, strawberries are extra delicious this time of year, so I brought my little strawberry almond cakes. Oh, yum. I roast the strawberries before I mix them into the batter. It helps condense the berries' juices and stops them from leaking all over and getting the crumb too soft. Mmm. You get little pockets of concentrated strawberry flavor. That tastes amazing. Oh, thanks. New York Times Cooking has so many easy recipes to fit your summer plans. Find them all at NYTCooking.com. I have sticky strawberry juice all over my fingers.

Kevin, remember when we took a ride in that self-driving car? I do. That was so fun. And his name was Banana Slug. Yes. It was made by Cruise. Yes. And I think it's fair to say we had a good time. We did. But would it surprise you to learn that there are people who are seeing these cars in the streets of San Francisco and they're not having a good time?

It would not, in part because people in San Francisco can turn anything into a cause. That's right. We love our democracy here and we love to debate the issues of the day. And I think it's fair to say that the self-driving cars have become a hot topic of discussion as they have spread throughout more and more corners of the city.

And, you know, this week, the California Public Utilities Commission held an important vote on whether these things would be allowed to expand the hours of the day that they're able to drive or the number of neighborhoods that they're allowed to go into. And they ended up approving that expansion. Frankly, I had just been waiting for them to rubber stamp this and get on with my life so I can start taking a self-driving car to work. But other people said, we're going to put a stop to it.

Yeah, there's been so much activity around this particular vote because this is sort of seen as like a do or die moment for the self-driving car industry, at least in San Francisco. And, you know, the companies have been out sort of lobbying on this, but there are also like labor unions, like the Teamsters, who are opposed to letting these companies put more self-driving cars in San Francisco because they worry that it could displace jobs.

And then there's kind of been this grassroots activism that I really didn't see coming. And actually one of those, have you seen this thing called the week of cone where people are disabling self-driving cars using traffic cones? - I have not. Let's take a look at this. - Okay, so let me show you this video. - So what's the deal with the self-driving cars all over SF? These companies promise their cars will reduce traffic and collisions.

But instead, they block buses, emergency vehicles, and everyday traffic. So we've seen this video, self-driving cars sort of getting stuck in traffic. But we can do something about it. And what they're encouraging people to do is to pick up orange traffic cones from the street and put them on the hood of the self-driving cars. Which presumably disables them by blocking their sensors. Right. It's really fun and anyone can do it.

They're sort of encouraging this kind of guerrilla sabotage of these self-driving cars. Hooliganism, even. So, you know, naturally, the self-driving car companies hate this. They hate that this is going on. They think this is just kind of like vandalizing their property. And this is basically a misguided movement that doesn't understand that self-driving cars are actually going to be way safer than human drivers.

But this movement is here. It is happening in San Francisco. And ahead of this big committee vote, we've been seeing more and more cars with cones stuck on their hoods. That's right. And it's from this decentralized activist group called Safe Street Rebel. And I think it's fair to say that they're on the extreme end of the opposition, right? Right. They're basically for self-driving cars, what Greenpeace is for fossil fuel companies.

They do not like self-driving cars. They think there are all kinds of problems with them, safety and otherwise. And in addition to kind of doing the normal grassroots lobbying of like trying to get this vote to sort of strike down these tests by the self-driving car companies, they've also been mounting this sort of cone-based opposition campaign. Yeah. So...

Clearly, this is going to be a big fight, not just in San Francisco, but I would say in a lot of cities where self-driving cars are going to be appearing in the next few years. It sort of feels like what's happening in San Francisco now is a preview of some of the civic battles to come. And so I think it's worth digging in and actually getting a sense of what the stakes are here and what the opposition to self-driving cars looks like. So,

Today, we're going to bring in two organizers, Adam Eggelman and Ming-Wei Samuel from the group Safe Street Rebel. This is the group that has been putting cones on the hoods of driverless cars in San Francisco. And we're going to talk to them about why they're doing what they're doing and what they think the future of self-driving cars should look like. All right.

Adam and Ming-Wei, welcome to Hard Fork. Thanks for having us. So I want to start with just a basic question, which is just who are you and what is Safe Street Rebel? Sure. So we are organizers for Safe Street Rebel. Safe Street Rebel is the direct action and anti-carceral arm of the larger San Francisco bike and transit community. So we do all sorts of direct actions, sort of leveraging our power through shenanigans.

Things like bike rides, vigils for people killed by traffic violence, and other sorts of street theater. Yeah.

I think that sums it up pretty well. All right. Okay. I want to get into your mission and why you're doing this kind of direct action against the self-driving cars in San Francisco. But I just want to start with a basic question that I have, which is why cones? What do the cones do to sort of disable the self-driving cars? And like, why did you decide on that as your form of direct action? I'm not sure who...

originally came up with that? I think I did. Oh, yeah? There's one dispute, actually. One other guy who also may have figured that out in parallel. But, I mean, traffic cones are something that all drivers have to be aware of and observe. And part of the idea was, as we're biking around with my bike trailer, I could put a cone on that and see...

How would AVs respond to a cone moving around on the street? And then at some point we thought, okay, what if we just put it on the hood of the car and so it moves around with the AV itself?

And, well, it turns out that just makes the AV stop in place. And I'm pretty sure that's because it just gets in the field of view of the sensors. So it just doesn't really know what to do or maybe thinks that it's like intersecting into the cone. And what led you to decide to take action against self-driving cars? What are your objections to autonomous vehicles? So I think this all came about because a lot of us felt powerless.

With the influx of AVs in San Francisco, it's regulated by the CPUC and the DMV, both of which are not run by people who are democratically elected. So I think it came out of feeling powerless in sort of the face of these sort of obscure agencies and these very well-funded robo-taxi companies.

We fundamentally see this as these companies dumping more cars on our streets in an unregulated fashion with no sort of democratic means of accountability. I want to ask you both to respond. We reached out to Cruz and Waymo for comments before talking to you today.

And a cruise spokesman sent us over a comment. I just want to read it for you and have you respond. He said, quote, we have an active dialogue with people about AVs, including those with criticisms, as we try to make our service better and improve our technology. But regardless of your view on AVs, trying to disable cars like this in the middle of the street is an unproductive and potentially dangerous thing to do and something that has a negative impact on real people. So I guess who wants to respond to that?

it has a negative impact on perhaps the person waiting for the cruise to show up. However, it's a protest and there's this sort of idea, it seems like that comes up a lot that when you do a protest,

If you inconvenience anyone, you're doing something wrong. A protest is only good if you let no one know about it because no one's going to be bothered about it. So as part of a protest, you're going to turn some heads. And Cruz is doing a phenomenal job at proving how unreliable they are without cones. Yeah. They stall all the time without cones on them. So we're just adding a little bit to that.

Did you consider other sort of maybe less disruptive protest tactics? I mean, you could do things, you know, you could, I don't know, I don't want to give anyone ideas, but you could do a lot of things to drive attention to your cause without actively disabling the cars or making them stall out in the middle of the road. Did you consider other tactics too? I mean, I should say at the very least, we do go through like official channels, do public comments, and, you know, it's just not exciting though. The goal of this...

I mean, the cones are funny and it's great street theater, but the goal was to raise awareness and start a conversation, which we think we did. And then obviously turn up public comments. So we're pretty happy with how this went and not really concerned with the impact it has on these companies. Yeah.

Do you want to get more into your objections to automated vehicles overall? It's weird for me to hear because I can't... To me, it's just so clear that human-driven cars are so unsafe that if I could swap all of them out for AVs, I would do it tomorrow. We definitely agree that human-driven cars are a huge safety hazard. Our group is fundamentally anti-car, whether there's someone behind the wheel or not. Yeah. Personally, all the...

problems that AV companies claim they're solving, they really don't solve. The big one is safety and the normal conventional ways we measure how often drivers crash. Those don't really work on AVs because they fail in new and exciting and different ways.

And then all the other things like reducing car ownership, we're really just putting more cars in the road just like Uber and Lyft. Yeah, we see this as a repeat of what happened almost a decade ago with Lyft when they claimed that their services would render car ownership obligatory.

That's obviously not the case. And this is, you know, we're in a climate crisis. We need to get as many cars off the street as possible. So if the form factor for these autonomous vehicles was not cars that are designed to drive one or maybe two or three people around, if they were self-driving cars,

buses or vans that could carry multiple people, would you be less opposed to them? Is it the single person occupancy that bothers you most? Yeah. So if we imagine if we design some sort of transit system that was able to like guide these self-driving vehicles around and maybe we could string a whole bunch of them together. On like a route even? On a route, maybe a given route so people could just hop on.

Whenever they wanted hop off. We could even call it the bus. Yeah, the bus or a train. Yeah. But buses, self-driving buses would also displace. I get the point you're making, but they would also displace people from their jobs. They would also have all these other associated risks and they would be autonomous. So would that pass muster with you or would that still be something that you would object to?

I agree with that. And really, whether or not there's a human driver or a computer driving a bus really doesn't matter because a bus is an efficient form of transportation. So from a labor perspective, we'd object. But from a transit perspective, I think it's the same thing. It's a bus. It's great. Yeah. I mean, I think also we're seeing some pretty dire implications for labor with these robo-taxi companies. And we're hearing that a lot from the taxi unions. You know, they claim they are driverless and autonomous, but...

we see with the Cone campaign that someone has to retrieve these cars. There are people in call centers and the depots that have to respond to these incidents. So they may be self-driving, but there are still people involved in these operations and they're relegated to the background and they're more invisible positions, so they're easier to exploit.

I mean, taxi drivers were massively exploited for their entire existence. I mean, the medallion system was horrible for people, you know? I just, I'm having a hard time understanding why that should be considered a model going forward. Yeah, it was not like a utopia of labor rights and worker empowerment. Yeah. Yeah.

I can agree with that. We want to shift most trips to public transit. I think that's pretty clear. Unusual that we're aligned with the taxi workers on this one. Yeah, this is a rare case of solidarity. My enemy is my friend or whatever they say. I mean, even the fire department union we're aligned with, which is pretty rare for us. Yeah.

So your group, Safe Street Rebel, has a website. And on that website, you keep a long sort of tally of incidents that you describe as AV failures. I have it out here printed. Oh, you actually printed it out. Wow. So it's a very long list. I think there are more than 100 incidents that you have

cataloged everything from like this AV, you know, stopped at a stoplight and didn't go right away to actual crashes where, you know, people are hurt. Many of those have been, you know, attributed not to the AVs themselves, but to actually human drivers who bump into the AVs, you know? So I think the companies and the proponents of self-driving cars would dispute a lot of the things on these lists, but there is a big list.

And it strikes me that if you kept a similar list of human driver failures, it would be way, way longer, right? You would have no shortage of crashes caused by people looking at their phones or getting distracted or making a call or listening to music too loud or just being bad drivers in the ways that humans can be bad drivers.

And actually, there have been some studies that have shown that self-driving cars, you know, mile for mile result in fewer collisions than cars driven by humans, at least in the areas where these have been tested so far. So what do you make of the people who say that by opposing self-driving cars, you're actually making the streets of San Francisco less safe?

I fully agree that human drivers are very bad. But the thing about AVs is it's actually just increasing the total number of cars on the road. It's not eliminating human drivers, really. And also, the way that AVs fail, like I've said, they don't tend to crash. They're very good at almost crashing and creating dangerous situations that are hard to report on a

to a government organization where it sort of externalizes the risk to other road users like pedestrians or bicyclists by sort of stopping randomly in the road or just putting themselves into very awkward situations that sort of force everyone else around them to adapt. I mean, I imagine you guys are big cyclists. So, like, I mean, have you encountered this yourself? Like, what does an AV do that creates danger for you?

Yeah, so the most common thing they do that human drivers never do is they just stop wherever they are and turn on their hazards and do nothing for multiple minutes. In some cases, this blocks up the entire road, so you end up with a huge queue of angry car drivers behind that you have to go around when you're biking, and the bus gets caught up in it too. And then there's like a fundamental misunderstanding between...

These robo-taxi companies and the city, whenever a robo-taxi comes to a stop at an emergency situation or is about to enter a crime scene or an active fire scene, the robo-taxi companies want the first responders to follow their procedure and scan a QR code and call the dispatcher and sort of wait and babysit these cars, which is a huge waste of their time. And that also puts people at risk.

Got it. Yeah. Are you advocating for the AVs to be taken off the road altogether? I mean, we want to ban cars. Yeah. Something that really motivated me in this campaign was just the climate crisis angle. At the end of the day, we need to get cars off of our streets, whether they're electric, whether there's someone behind the wheel or not. Car is a car. You know, electric cars will still pollute through tire wear. That's like one of the main sources of plastic pollution in our oceans.

So at the end of the day, we need to get cars off our streets and more people on buses and trains. And I just see driverless cars as a huge distraction from that effort. I think we've ended up in a sort of classic situation where people who are fans of tech end up

Pointing to a new tech thing and saying this is going to solve all our problems when it's already a solved problem. And the solution is to design streets around people, fund transit, instead of designing them around cars. I mean, just to talk about transit for a second. I love transit. I took transit to get here today. I love taking community underground. It's wonderful, right? Yeah.

At the same time, there is disinvestment in transit. In part, that is because Uber and Lyft exist, and they sort of reduce pressure on the system. At the same time, every time I read a story about transit, it's like, well, they have no money, and the project is over budget. And it just seems like transit agencies, in San Francisco in particular, are just really struggling to make a good product.

So what do you see as some of the solutions to actually make transit better? So that's the problem. It's not a product. It's a public service. So...

Basically, the problem is transit has to compete with roads. And I mean, we pay for roads via taxes, but no one ever goes to the freeways and looks at what's their farebox recovery ratio for everyone driving on the roads. So basically, we put transit in this impossible position where they have to compete with cars and it just doesn't work out. I'm wondering if you can sort of expand on this vision that you have of a world in which cars, not just autonomous cars, but all cars are banned or

What does that look like? At least in cities. In cities. Paint the picture for us. Convince us that that would be a more desirable world than the one that we have today. It seems very inconvenient.

I mean, have you traveled outside of America? Yes, and it's usually more convenient to hail a car on your phone than it is to figure out how the transit system works. Not in cities like Amsterdam or Paris. A lot of cities right now are going through a process where they're rethinking what should we allocate our streets for. And when you deprioritize cars, you create cities that...

are quieter, cleaner, easier to get around, and they're safer for people who aren't in a car. I wrote a book about automation a few years ago, and as part of that, I researched how activists responded to the invention of the original automobiles back in the late 19th and early 20th centuries. And

a lot of them are sort of making the same criticisms that you all are making today. You know, they said cars are, you know, they're ugly. They would clog the streets and be dangerous for pedestrians. They would destroy jobs for, you know, stagecoach drivers. You know, those movements were, you know, obviously not successful. They didn't stop cars from proliferating. And,

And, you know, actually, if you look back at the history of technology, there aren't that many successful movements that are trying to sort of sabotage or undermine new technology, especially when that technology makes people's lives safer or more convenient in some way. So given the history of these kind of opposition movements, do you think that you may be fighting a losing battle here? Do you think that there's actually a chance that you will help turn the tide against autonomous vehicles? Yeah.

Yes. So that is a very good parallel. And part of it is that a lot of the stuff about cars taking over the streets in modern times now, we are realizing that

They had a point with freeways destroying communities and cars clogging up streets and creating pollution everywhere. People are now realizing in the sort of new urbanist movement that, you know, the places that did resist the expansion of cars everywhere actually had it right and have a lot better quality of life now because of it.

So in sort of the same way when like cars appeared, there were like campaigns by the auto manufacturers to introduce the idea of like jaywalking. Now you can no longer walk in your own street. It's a crime now. It's sort of the same thing with AVs showing up. And now first it was going to be, oh, they'll just be like any other driver. But now it's like, okay, if you're a firefighter, if you're a crossing guard, you should go take this training and start walking.

Making room for the AVs and when AVs and people get in conflict, it's always people need to get out of the way. They must have been doing something wrong. And that's just backwards. Somehow the burden is always placed on those outside of the car. Yeah.

You know, one place where I think I do agree with you guys is that you've pointed out that the AV companies partner with police and are often sort of just recording us constantly. And then that gets used in a variety of ways. Can you talk a little bit about what that relationship is and why that has concerned you? Yeah. I mean...

It's technically legal to record anyone in public, but if you imagine some guy with their phone holding it up, following you around, that'd make you pretty uncomfortable. And essentially, all the AVs are just cameras facing 360 degrees, filming everyone at all times. And that's not good for society. Yeah, I mean, it's a roving sort of network of always recording cars is not something, you know, synonymous with a democratic society, right?

Some people in the tech industry have criticized you guys as being Luddites, just sort of reflexively opposed to new technology. Does that bother you?

People who pay for Twitter really don't like us. I think it's funny that they call us Luddites without sort of understanding the full history of that term. Because the original Luddites, we're not really anti-technology. We're not anti-technology either. In fact, a lot of us are tech workers. We're more concerned with the labor and the potential for exploitation, privacy concerns. So do these second-order effects.

It's very easy to look out into the world and see technology that exists and say, oh, this technology survived, so it's good. And you forget about all the things that didn't survive because they're not around anymore. So, I mean...

Things like, you know, flying cars were always supposed to be a thing and vacuum trains and all sorts of ridiculous tech doesn't exist because it just doesn't make sense. And I think AVs are one of those things. Yeah, and there's this assumption that AVs through AI will...

automatically improve over time. That's not always the case and that's not, you know, you can't take that as a guarantee. They can regress. One of our friends was listening to an Adam Conover podcast episode from two years ago about autonomous cars and they mentioned how Waymo is back even two years ago or still didn't know what to do with when they saw Kone.

So there's no guarantee that they'll improve. I guess I'll adjust my statement a little bit. I guess personally, I think that eventually there will be AVs, but I think they're not going to solve any problems and they're just going to create more. But I imagine eventually when the tech finally... You don't think they're going to solve any problems? I don't think they'll hit people as much, but I think they will...

Isn't that a big problem? Well, there's also no guarantee that they will get people out of their cars. Yes. Like, we've seen this happen with left. Like, that was the promise that they would get people out of their cars. That has not happened. There are just more cars on the streets now. And they create different dangers is what I'm saying. And I can talk about specific examples if you want. Yeah, let's talk about it. Like, what's a specific example of where an AV is less safe than a human-driven car would be?

Yeah, so on, let's see, on Monday, there was, over near Golden Gate Park, there was a semi truck making a wide left turn. And any human driver in that situation would know, all right, give the semi truck some space. But AVs are very good at almost crashing. So in this case, a cruise drove up to the side of the semi truck and just stopped there because it saw it was crashing.

was in the way and the semi driver didn't notice. So the semi driver kept driving and hit the AV. So then the semi driver's at fault. So that's a situation where the AV created a unique danger just by not having intuition about how the world works. Another example is, is AVs just stop in crosswalks all the time. So when you're crossing the street, that means you are going to have to go out into traffic to get around the AV.

And then, of course, there's the cases of fire and police scenes where the AVs will just drive right up to the back of a fire truck or right next to the side and just stall there for multiple minutes. Or drive over a fire hose during an active fire. Right. But so far, as I understand it...

Despite the fact that these vehicles have been running in San Francisco for years now, there are no documented instances in which an AV is at fault for a deadly or serious collision. There was a dog that was killed by Waymo.

Okay. So a dog, a dog, but no humans. So it does. I mean, and in that same period of time, I'm sure if you went back and looked at the data about human driven cars, there, there probably were fatal crashes involving human drivers during that period in San Francisco. So I guess just from a pure safety harm reduction standpoint, shouldn't we be trying to get humans out of the driver's seat in as many cars as possible, as quickly as possible? Like,

Like, I feel like... What am I missing here? Dumping private robo-taxis onto our streets will not do that. The question... Making it harder to drive through, you know, infrastructure changes that prioritize transit and biking and walking will do that. Not dumping these cars on our streets without anyone's consent. Yeah.

Well, I mean, I think like the most cynical reading of your project is that the reason that you don't like AVs is that they make your overall project of getting cars off the road harder. And even though it could save thousands of lives, like you're opposing it anyway. I mean, I don't really buy that. And also I think that assumes our standards are just super low. Yeah. What do you mean? So we are fighting overall for street safety and safety.

The thing about AVs is they are a big distraction from actual solutions. But in your mind, the only actual solution is just that cars don't exist anymore.

Or more that our built environment prioritizes other modes of transportation. Because, you know, AVs require our communities and our built environment be designed around cars. Right. And I've seen transit groups arguing for, you know, protected bike lanes and things like that. And bus lanes. All of which I think are a good idea. I want to ask, like, of all of the contexts that we can imagine...

autonomous vehicles existing in. You know, busy city streets is just one of them, right? There are companies that are trying to do AVs for long haul trucking. Is that a situation where you would support autonomous vehicles if it weren't sort of in these busy areas in urban cores with lots of pedestrians and bicyclists around? I think the argument's certainly better there. In cities, it just makes no sense.

Got it. And I think the AV companies have been training in cities because that's where you get a lot of interesting data and that's sort of the highest challenge level for the technology is in a busy urban environment. So I think that's where they've been wanting to test. But I think ultimately these things will be appearing everywhere. So you're not going onto the freeways and putting cones on the autonomous trucks there, right? Not yet. Okay.

All right. Yeah. Well, thanks for stopping by. Really interesting to talk with you. And yeah, if you see Banana Slug, Banana Slug was our AV that we took. Just don't do the cone thing up and just spare Banana Slug. Banana Slug has done nothing wrong. Might be a little late there. Oh, boy. Banana Slug, send up a flare if you need help. Banana Slug's rolled over in a ditch somewhere. All right. Thank you both. Thank you. Thank you.

BP added more than $130 billion to the U.S. economy over the past two years by making investments from coast to coast. Investments like building EV charging hubs in Washington state and starting up new infrastructure in the Gulf of Mexico.

It's and, not or. See what doing both means for energy nationwide at bp.com slash investing in America. All right, Kevin, we have talked a lot today about rebellion and backlash. Yeah, it's been a big power to the people episode. Power to the people. And I want to talk now about a case where the people definitively lost. And that was in their uprising against Reddit.com.

And it's another story about folks caught up in the big fight over AI data scraping. So let's get into it. Yes. So we talked about this story on the show a few weeks ago when the rebellion was sort of in full swing. This was, you know, users of Reddit who protested these changes to the Reddit API that the company said were designed to ward off data scraping by AI companies, but that had the effect of sort of putting a lot of the third-party apps that people use Reddit with out of business.

Yeah, so just a quick recap. Earlier this year, Reddit said, hey, we're uncomfortable with these large language models scraping our site. We are going to make changes to our API. Previously, anyone could use the Reddit API basically for free to do a lot of things, including making accessibility tools, making tools for moderators, or even making a third-party Reddit client that didn't show ads. And people were selling these in the App Store, in some cases, making pretty healthy sums of money.

And Reddit shut it all down. And people really, really did not like that. And there was a huge protest in which subreddits were going dark or they were labeling themselves as not safe for work, which prevented them from being monetized. And it has been just a huge...

huge, huge thing over at Reddit. Yeah, and the company's CEO, Steve Huffman, stood firm against the protesters and basically said, like, we're going to wait this one out and eventually things will die down.

So what happened after that? Well, they waited things out and things died down. But I think how they did it is worth talking about because it was, I think, a pretty brutal crackdown. The first thing they did was they threatened to remove moderators. So the reason that all this was able to happen was that Reddit moderators had the ability to make their Reddit so that nobody could read them or make them so that they were labeled not safe for work.

And Reddit said, hey, if you do this past a certain point, we are going to remove you as moderators and we're going to install our own generals. Do they have that? Like, do they have like this army of like Vichy generals that they can put in to replace the moderators who volunteer to clean up the site? They essentially solicited for scabs and found them.

and then installed them at these subreddits. And one of the places I noted where they replaced the moderators was the subreddit devoted to canning, you know, like sort of preserving fruits and vegetables. Is this a big subreddit that you visit a lot? It's a huge subreddit that I visit a lot. And I just couldn't help but notice that the moderators of r slash canning were canned. And that just seemed like something worth pointing out. But, you know...

They started to reopen these subreddits under new management. There were some continuing protests. So, for example, at r slash pics, which is just a place to post interesting images, those moderators created a rule that you could only post images of the TV host John Oliver looking sexy. Wait, this was what...

I saw all these John Oliver pictures. This was part of an organized protest? Yes, they were trying to continue the protest by doing that. R slash awe, which I think is a subreddit devoted to cute things, only allowed pictures of John Oliver and a cute otter mascot from his show. R gifts only allowed gifts of John Oliver. So...

These are really big boards, and it was pretty disruptive. Yeah, they should have put cones on top of the John Oliver gifts. Well, I'm sure they'll be thinking about it after today's episode. There's also this thing that Reddit does every year called r slash place, which gives each Reddit user the ability to color on a huge digital art mural. And so people always make sort of elaborate creations by collaborating. This is like community, like crowdsourced graffiti on the internet kind of thing. Exactly. And so this...

This year, the makers of ArtPlace had a message that they spelled out with pixels, and that message was, Fuck Spez. Spez is the Reddit CEO's username, Steve Huffman. It goes by Spez on Reddit. And eventually, it got to the point where Reddit stopped people from posting or commenting on r slash place.

So, look, about 1,800 subreddits are still dark today, but it's mostly pretty small ones, and it does seem like the furor is subsiding. Did Reddit make any concessions to the protests, or did they basically just say, like...

we're going to replace you as moderators if you don't shape up and get in line. And that kind of worked. So I don't know if they would say this was a concession, but they did assure people that the API would continue to be able to be used for free for makers of accessibility and moderation tools. And that did seem to take a lot of sting out of the protests.

At the same time, those other third-party apps are shutting down, and Reddit has installed new and presumably more pliant moderators who will do its betting. What's amazing about this to me is that if you had asked me before this, like, what is the site on the internet where a user rebellion has the biggest chance of succeeding, I probably would have said Reddit because...

As social media platforms go, it puts a lot of power in the hands of these sort of local moderators. These people who, you know, are moderating the history subreddits and the pics subreddits and the canning subreddits. Like those people actually do have a lot of power. They are not normal users and they have leverage in particular, the ability to like shut down or sort of disabled their own communities.

So it seemed like it was heading for a pretty long stalemate because unlike the users of Facebook or Instagram or X, formerly known as Twitter, the people on Reddit, some of them do actually have real power. But it seems like that was not enough to get them what they wanted.

Well, because in the end, Reddit was able to take that power back, right? Because Reddit controls whether a forum is open or closed, who the moderators are, what they're allowed to do, what tools they can use, who can post and comment. And in the end, they decided, you know what, we are going to sort of take these unprecedented actions to remake Reddit the way that we see fit. Right.

So, you know, why did Reddit win this? Well, this is how you crush a rebellion, right? Like if you are invading a country, a lot of local politicians are going to be mad at you. But if you can capture and replace them with leadership that is loyal to you, over time, you will sap the energy of that movement because at the end of the day, most people just want to live their lives. So basically, Reddit made the most cynical bet possible and it paid off. But people are getting really annoyed at this. And I think,

And this kind of segues to what I really want to talk about here. This is why people care about, and we do hate this word, decentralization. I knew you were going here, and I've been dreading this moment in the conversation because, you know, I do think that

this issue of ownership or the lack of ownership on social media is a really big issue for a small group of people who want to keep, you know, their own subreddit, who want to keep their own handles. I don't think most people care about owning their social media data or information, but for a small number of people, it is a very salient issue.

Yeah, and I think that the existence of these people is a good thing for the internet overall. Because you're right. Most people, they don't care. They didn't care that much about the Reddit stuff. They're still going to keep using Reddit because they want to know how to key on their preserves this summer. But there are other people who are

really principled, who believe in good governance, and who want to be able to sketch out alternate visions of the internet. These are the people that buy their own web domains, and they build their own websites because they want to build a vision of the web that is outside of this kind of corporate control. And

As you say, this has been a pretty niche-y little cul-de-sac of the internet for a long time. But because people keep rebelling against these platforms, because they're keeping these moments where everyone gets so mad, it is pushing more and more attention to some of these alternatives. So in this case, there is an alternative to Reddit known as Lemmy, which you might think of as a kind of mastodon for...

Reddit, you can set up your own server for it. And I think the vision here is that if you want to run a canning forum that has a lot of Reddit-like aspects, but at the end of the day, no one is going to come along and say, well, you can't do this or that anymore, or you don't want someone to come along and start charging for an API or shutting down a tool that you rely on. Well, now you have that opportunity because you have built your own home on the web.

Yeah. And I look, I'm not opposed to decentralized social media. I think it is, in theory, a good idea to have less centralization on popular internet platforms and to give people actual ownership rights and sort of portability rights where you can, you know, if you stopped

agreeing with the way one social platform is run, you can take your account and all your followers and go to another social platform. I'm on board with that. Where I'm sort of left cold on this is just the idea that this is going to be the differentiator in why people would choose to use one product over another. And we saw this with crypto.

a lot where people would come along and they would say, you know, crypto solves all these problems for people. It means that the bank can't unilaterally seize your assets. You know, it means that Venmo can't blacklist you from sending certain transactions. And, you know, my response to that was always like, how many people actually have that problem?

Right? How many people are actually getting their bank accounts shut down? How many people are actually trying to send wires, you know, through Venmo and getting those transactions canceled because they're on some government list? Also, crypto assets are just seized all the time now. If I read about a crypto asset online, it's probably because it was seized. Right. So, crypto assets

This was one of the main points that proponents of crypto would make is like, this is a decentralized form of finance that is accountable to no centralized entity where it's censorship resistant. And that was the pitch that they made. And guess what? That pitch did not resonate with people to the extent that it did. The people were actually interested in getting rich. They were not interested in censorship resistant money for the most part. So I just think that it's very easy to overstate how big a deal this is to normal people. I think that if there

is a version of Reddit or X or Facebook or Instagram or any of these companies that is decentralized.

it could take off, but it will not be because it is decentralized. It'll be because it's a better product. Right. And I think the only thing I would say in response to that is I think that decentralization can make it a better product, right? So when you look at what is happening on Mastodon right now, Mastodon, as we've talked about before, is built on this standard called ActivityPub. Other services are testing out support for it, including Automatic, which makes WordPress the most popular blog software. Automatic...

also owns Tumblr, a social network of its own. There are publishers now, you know, the BBC has now said that it is spinning up a Mastodon server. So what that means is you have this user base, mostly thanks to Meta's Threads product, and if and when Threads ever decides to become interoperable with that broader, what they call the Fediverse,

then I think you just sort of have some interesting possibilities. Then you're able to build new social experiences starting from this big user base where people have a little bit more control over their little corner of the web. And I can't tell you exactly what it's going to look like, but the idea of all of these things being interoperable and individual users having a little bit more control, it just gets me excited. Yeah, well, you probably also like the California ballot initiatives where...

you know, millions of people are asked to vote on like, you know, the fate of dialysis centers. Like I do think there is such a thing as too much direct democracy. And when it comes to social media platforms, I think what we've seen with these decentralized platforms is that they just descend into chaos.

And there is no centralized force that is able to like stem the tide of entropy and like make it a pleasant and intuitive thing to use. Right. So you want to entrust your entire future on the internet to the whims of Mark Zuckerberg and Elon Musk. That's your pitch. Interesting. I didn't say that. I didn't say that. But I do think that there will be some combination of decentralization and centralization. I don't think it is all one or all the other. I think there may be

like you've described with ActivityPub, where you have Threads, which is owned by Meta, which is a centralized company that is making centralized decisions about content moderation and rules and things like that, but that is sort of building on this open infrastructure. And I think, you know, that will be a sort of a compromise position that a lot of companies may take if only, but then I think you get back into this question of what happened at Reddit, because if Reddit had been decentralized...

I think there's a good chance that the site would still be unusable today, that we would still be in this rebellion, that the company and the users and the moderators would have locked horns, and that there'd be sort of no way of resolving that. And as a result, we might lose Reddit. So what is your take on that? Like, has the Reddit rebellion proved that decentralization is necessary or that actually there is some benefit to centralization after all? Well, at the end of the day, Reddit is not a profitable company.

And so it's hard for me to say that a centralized network is the one that is going to win, at least in the Reddit space, until it starts making any money, right? I think you're right that a decentralized version of it will have struggles of its own. But like Reddit has existed for a really long time. It's still not making any money. And so this idea of like, well, but what a decentralized version of it, maybe we're

Maybe. I mean, Wikipedia is the sort of classic example of something that is sort of decentralized in both the way it operates and the way that users contribute to it. There's the Wikimedia Foundation, which kind of raises money for it and sort of oversees the technical infrastructure. But in all the important ways, like it is a sort of community decentralized project. Yeah. And it's worked really well. Yeah. They don't print the Encyclopedia Britannica anymore. Yeah. So that's how well that worked. Yeah.

But at the same time, Wikipedia is also full of drama. I'm sure if you've talked to Wikipedia editors, there's always some crazy drama going on in Wikipedia. Someone's edits are getting reverted. Someone is being asked to cite a source. It's not an uncontentious process. I mean, Kevin, let me tell you something about the internet. There's going to be dramas going on.

Okay, when people are speaking to each other on the internet, there's going to be conflicts and dramas. And that's going to be true whether things are centralized or decentralized. But look, coming out of the past decade that we had, I really do think a lot of people left it feeling like we have to reset the balance of power a little bit. And some of that power needs to go from these platforms back to the users. And if you believe that, I've got a cryptocurrency to sell you. And I have a cone to put on your head.

Indeed believes that better work begins with better hiring, and better hiring begins with finding candidates with the right skills. But if you're like most hiring managers, those skills are harder to find than you thought. Using AI and its matching technology, Indeed is helping employers hire faster and more confidently. By featuring job seeker skills, employers can use Indeed's AI matching technology to pinpoint candidates perfect for the role. That leaves hiring managers more time to focus on what's really important, connecting with candidates at a human level.

Learn more at indeed.com slash hire.

Christine, have you ever bought something and thought, wow, this product actually made my life better? Totally. And usually I find those products through Wirecutter. Yeah, but you work here. We both do. We're the hosts of The Wirecutter Show from The New York Times. It's our job to research, test, and bet products and then recommend our favorites. We'll talk to members of our team of 140 journalists to bring you the very best product recommendations in every category that will actually make your life better. The Wirecutter Show, available wherever you get podcasts.

So, Kevin, last week we talked about LK-99, the alleged room temperature superconductor, and I believe you have an update for us about that. So last week after the show, LK-99 got the hard fork bump. Everyone was talking about it. You put on threads that you were snorting LK-99 in the club bathroom over the weekend. I just wanted to see what would happen.

And I think it's safe to say it has not gone well for LK99. It's not been a good week. How did this happen? We were so close to having a room temperature superconductor. Well, we weren't actually that close. It looked from the videos like we might be close. But several labs have finished trying to replicate their LK99 creation process.

And all evidence that we have so far points to the fact that this is not actually a room temperature superconductor. No one has been able to replicate this finding. And the betting markets where the gamblers go to wager on whether this will be replicated or not has fallen quite a bit. So the good money is now predicting that this is not a room temperature superconductor. And it looks like we may have had a little bit of false hope. Well, that's the last time I'm trusting science. But you know, I read...

That this LK-99, while not a superconductor, there is evidence that it is ferromagnetic. And I guess my question for you is, what does that mean? I'm going to ask ChatGPT what ferromagnetic is. Yeah, let's see what it has to say. Okay. ChatGPT, explain ferromagnetism to me, an idiot. Okay. Okay.

Okay, this is already way too complicated. Oh, no. Okay. Did you make sure to tell it you're an idiot? Yeah, I told it I was an idiot, but it's now drawing like an extended metaphor about teachers and classrooms. Really? So imagine a classroom filled with kids. Okay. The kids are atoms. Every kid is holding an arrow that can point in one of two directions.

This is already, like, it's giving polarity. It's alignment of spins, magnetic domains. ChatJBT does not understand how big of an idiot I am. The neural network is confused. Well, I mean, that would require a lot more training parameters. We're going to need more GPUs for that. Hard Fork is produced by Davis Land and Rachel Cohn. We're edited by Jen Poyant.

This episode was fact-checked by Caitlin Love. Today's show was engineered by Sophia Landman. Original music by Dan Powell, Alicia Baitube, Marion Lozano, and Diane Wong. Special thanks to Paula Schumann, Pui Wing Tam, Nelga Logli, Kate Lepresti, and Jeffrey Miranda. You can email us as always at hardforkatnytimes.com. And we promise not to use your email to train our AI.

Don't make that promise. I am training a language model in my basement as we speak. One key cards earn 3% in one key cash for travel at grocery stores, restaurants and gas stations. So the more you spend on groceries, dining and gas, the sooner you can use one key cash towards your next trip on Expedia, Hotels.com and Verbo and get away from.

groceries, dining, and gas. And Platinum members earn up to 9% on travel when booking VIP access properties on Expedia and Hotels.com. One key cash is not redeemable for cash. Terms apply. Learn more at Expedia.com slash one key cards.