Google要被拆分？+ Seasteading回潮 + 工具时间

Casey, I miss you. You're in New York this week.

I'm in New York. I thought it was time to finally come back to the offices of The New York Times and see what I could find out about your performance review.

Our studio in San Francisco feels very empty without you. What are you up to there?

It's so empty, you can actually fully extend your legs while standing inside the studio, which is not always true when there are 2 hosts present. So I saw Titanique last night. Have you seen this show or heard of it?

It's pronounced Titanic?

It's Titanique.

Okay.

It is a sort of fever-dream retelling of the Titanic story by an actress playing Céline Dion. It's kind of like a Céline Dion jukebox musical set in the world of Titanic. It's absolutely hilarious. I just thought, as a theater kid, you would love it because this is an Off-Broadway show, and I swear to God, the singers were the best singers I've ever seen live. One of them was the understudy. These people are incredible.

Yes, the talent density of singers in New York is unbelievable. I realized this when I was 22 and moved to New York and entered a karaoke contest at a local bar. I wasn't making much money, and they had a $500 cash prize.

I was like, “I sing. I'm an okay singer. I might be able to win this thing.” So I show up to the bar, and the karaoke contest is about to start, and I just hear these voices doing warmups in the hallway. They are the best singing voices I've ever heard. These are professional singers. I'm sure they were Broadway actors and actresses ready to sweep in and win this contest. I just decided, yeah, I'm not going to enter the karaoke contest anymore.

Yeah. I don't want to go up against Patti LuPone tonight.

1. Google Faces Two Antitrust Trials

Well, Casey, the big news we have to start by talking about this week is what the heck is going on with Google, which I understand has had a very busy week in antitrust land.

Yeah, so this was a huge week for Google. In fact, I think looking back, we might even come to see it as one of the most important weeks in the company's history because, on one hand, you had another antitrust loss where a judge has said that the company has an illegal monopoly, this time in ads. And across town in Washington, D.C., another judge is in the middle of a remedies trial trying to figure out what to do about the company's illegal monopoly in search.

Yes, and I have found this whole thing very complicated and hard to follow, partially because Google's ad business and the various mechanisms that it uses for these online ad auctions are very complicated, but also because there is this very strange naming thing that is happening because the judge in one of Google's antitrust cases is named Judge Mehta.

Mm-hmm.

M-E-H-T-A.

Mm-hmm.

And so every time someone talks about how Mehta is doing at its antitrust trial, I just have no idea whether they are talking about Meta the corporation or Mehta the judge.

Yeah.

Anyway, carry on.

And it's made even more confusing, Kevin, by the fact that the judge in the Meta case is named Judge Google, and so—

You just can't win.

It's been crazy. You can't win, yeah.

So the last time we talked about Google's antitrust woes on this show, we brought up the fact that there are 2 distinct cases against Google that the federal government is bringing. One of them is over this issue of bundling, whether Google is allowed to make these business deals that involve paying billions of dollars to companies like Apple in order to bundle its search products with Apple's iPhones and make them the default on Safari and things like that.

There's the other trial that I understand has something to do with Google's search ad business and its auction process, and various Byzantine structures within that. So maybe let's start with the one that I understand more about, the bundling case that is about these exclusivity deals. What is going on with that case?

Great. So it was last August, Casey, that Google lost that case. That is the case that's being presided over by Judge Mehta, and he ruled just what you said: that Google has maintained an illegal monopoly in search engines. A primary way that it has done that is by investing billions and billions of dollars every year to make sure that Google is the default on your iPhone, on a lot of Samsung phones, and various other things they call OEMs, or device manufacturers.

So in August, Casey, Judge Mehta says, “Hey, that's illegal.” A few months later, the government unveils its proposal for what they think ought to be done about it. And this week, Casey, Google is back at trial, and all of this is now being litigated, where the government is saying, “Here's what we think you should do,” and Google is trying to fight very hard against that and say, “We think that you should actually do much less.”

And how is that going for people who haven't been following that part of this antitrust trial?

Well, it's difficult to say, of course, because we're not inside the head of Judge Mehta. This is an interesting aspect of our legal system: This will be decided by 1 person, Judge Mehta, who's heard this whole case. It will sort of be up to him.

What we typically do in those situations is listen very closely to the questions that the judge is asking, and try to guess: Does he seem really skeptical about this? Does he seem more interested in that? Maybe we should start, Casey, by just laying out what the government has said should happen.

Yes.

2. The Chrome Breakup Proposal

So we've discussed this on the show before, Casey. I would say there are 2 big pieces of what the government wants. One is that it wants Google to spin out the Chrome browser. Chrome, of course, is the web browser that is the most popular browser in the world. Google built it in-house. The government is saying, “We want you to give this up.”

The government also wants Google to expose its proprietary data in some really interesting ways. It wants the company to license its search index—the index that it has made of the entire internet. Maybe even offer up some sort of API that would let the company's rivals, like DuckDuckGo, peek under the hood and see what everyone is searching for on Google right now.

And at trial this week, there was a witness from OpenAI, former Hard Fork guest Nick Turley, who said, “You know what? We would actually love to buy Chrome, and we would love to get a look at Google's search index, because that would make it much easier for us to build our own search engine and compete with them in the market.”

Hmm.

So those are the 2 biggest ideas that are being bandied about at the trial.

That's really interesting. I have so many questions about what OpenAI would do with Chrome and how they would use that as a way to bootstrap some improvements in their AI models. But we can save those for a later episode, or when Nick Turley returns from testifying in court and comes back on to tell us.

But what is Google trying to do to push back on this proposed remedy? Presumably, they don't want to give up Chrome at all. That's a very big, popular product for them and a big way to funnel people toward their search engine.

That's right. They have said that that is a really extreme measure and would sort of place a chill on innovation if the government could just step in and start tearing parts of Google out just because Google happened to make a very successful browser.

But what Google is really saying at the trial is essentially, the government should do as close to nothing as Google can convince them of, right? So, for example, Kevin, the third big thing that the government is asking for is, we want to put an end to these default-placement deals. That was how this whole trial started: The government was noticing, wow, Google is spending $20 billion a year just to be the default on iOS. Maybe we should stop them from doing that, and that would introduce some oxygen into the search market.

Google has come along, and this week they're saying, “No, no, no, no, no. Don't do that. Let us make these deals, but just make them nonexclusive. So maybe we pay Apple to be one of the search choices, but Microsoft also gives Apple a bunch of money, and then Bing can be a choice as well.”

Right, Apple's free to see other search engines.

Yes. And of course, I think this is a really flimsy approach. I've asked a Google representative this week directly, “How is this actually going to eliminate the monopoly?” The government has ruled that there is a monopoly in this case. If you just let other search engines bid to be on these devices, do you really think you'll lose any market share whatsoever?

The person I was talking to waved their hands and said, “Well, look, the government hasn't even really given us a target to go for, so we don't even know what the market is supposed to look like after all this is done.” I regarded that as hand-waving. What the government wants is obvious. They want there to be a search engine that doesn't have 90% of the search market, and one way of doing that would be to get rid of these deals.

Yeah, it sort of begs the question: If they did enforce this remedy, break up Chrome and Google, and force it to sell to someone else, and then they came back in a year or 2 and Google still had 90-plus percent of the search market, would they have considered that a failure? Because, for what it's worth, I think that is the very likely outcome here.

I don't think it's just lock-in and these exclusive-arrangement deals. I think Google Search is actually better than other search engines. And so I think even if you give people the choice or stop these bundling arrangements, many, many people are so accustomed to going to Google when they have a question that they'll just continue doing that.

3. AI Search Threatens Google's Lead

So I think that up until maybe a year ago, I would have agreed with you completely and said, “Maybe it's pointless to even try something like this, Kevin.” But then along comes AI search, along comes ChatGPT, along comes Perplexity, along comes all of these other chatbots that start to build a product that is not a one-for-one replacement for Google Search, but for many early adopters, it becomes much better than Google at lots of things, as we have discussed frequently on this show, right? You and I are both now using Google less because we're using AI products more.

And that's what I found so interesting about Nick Turley's testimony this week. What he was saying was, “Look, we're not here to compete with Google one-on-one. We want to build our own search product. We think search is a key pillar of what we're trying to build. But we can't build it in the way that we want to because there is a giant monopolist out there who won't work with us.” Something else interesting he said this week is that OpenAI went to Google last year and said, “Hey, can we make some sort of partnership with you? Maybe get access to some of your search index?” And Google said, “No, you can't.”

So what could happen in this case, Kevin? Well, maybe the government says, “Actually, Google, you have to go play nice with OpenAI. You have to go play nice with DuckDuckGo and Perplexity. You have to expose some data to them.” We can't say for certain that those companies are going to make perfect use of that data and make incredible products that are instantly used by billions of people, but it would give them a fighting chance. And I do think that, at least around the margins, it probably would continue to corrode Google's market share, particularly with ChatGPT, right? Because we've already seen it doing that even before that company has access to any of the data that I'm talking about.

Yes. Okay, so I have some thoughts about that, but first I want to hear about the status of the other Google antitrust case. Or as you could call it, Google Toogle. That's not a good joke.

No, that's a great joke. We're keeping that. Producers, make sure everyone heard Kevin say Google Toogle.

4. Google's Ad Monopoly Faces Trial

So over in what they call Google Toogle—that, of course, is the second lawsuit that has been filed—this one has to do with the online advertising market. And I think the reason that people know less about this case, Kevin, is that this market is incredibly arcane and difficult to understand, and that is on purpose. Because I think if more people knew how this market worked, they would have said, “Well, obviously, that's hugely anticompetitive.” And that is actually what a judge found last week.

So Judge Leonie Brinkema ruled in a 115-page ruling that Google has maintained a monopoly in 2 out of the 3 parts of the online ad market: tools for online publishers—newspapers and other people making webpages and trying to make money from them—and the software that publishers use to try to make money on their webpages. And that's a really big deal because that is the money engine at the heart of Google. That is the source of its wealth, and a judge came in and said, “This is a monopoly.”

So this second case, Google Toogle, is now going to proceed to the remedy phase that the first case is already in. What do we think will happen there? What stands to change about Google's ad business as a result of this judge's ruling?

Yeah. Well, the issue here is that Google just kind of owns every side of this market. And according to the judge, it has illegally tied those together. So in order to use one, you have to use the other. That is considered a classic antitrust violation by many competition scholars.

And so the thinking here, Kevin, is that the government is going to come in and say, “Well, first of all, you have to stop tying these 2 things together, and maybe you're actually going to have to unwind part of this operation. Maybe you're not allowed to own every side of this market. You're going to have to spin that out.”

And so, were that to happen, and were the government to also be successful in the web search case, all of a sudden you're looking at a Google that has way fewer searches, right? A Google that is sharing data with competitors, that is helping them grow, and that has less money because the core of its advertising engine has been disrupted. So are we at least a couple of years away from some of that stuff starting to happen? Probably. But if it does, Kevin, that would be the biggest change to the economics of the web in, I would argue, more than a decade, maybe 2.

Yeah. I think it's potentially a really big deal, and I've been convinced by you and others that this is actually something that I need to pay attention to because it actually might result in some major changes to the way Google operates. For me, the question I'm thinking about is whether the biggest effect of all of this antitrust litigation for Google will just be that it is distracted.

Mm-hmm.

What we saw during the last big government antitrust case against a tech company, Microsoft, was that the actual penalties and the remedies were not what ended up making things hard for Microsoft. It was just that it was so all-consuming and distracting for the company to be embroiled in this very high-profile litigation, to have all these lawyers peering over everyone's shoulders all the time, and to have all these reviews that had to go in before anything could really get shipped.

If you talk to people who were at Microsoft at the time, they'll just tell you that the net effect was that Microsoft became this sclerotic, slow-moving organization. And I wonder if what we're seeing at Google is some version of that. We've talked many, many times on this show about how Google's AI efforts just have not been nearly as fast and robust as you would expect from a company with Google's resources and talent.

And I wonder if what's happening is that they're just taking their eye off the ball in the new game that they are involved in while they're trying to litigate the old one.

5. Gemini Repeats The Search Playbook

So here's the thing: I don't actually think this is happening at Google, and here is why. Something else that came out at trial this week, Kevin, is that Google has been seeking default-placement deals for Gemini, which is its AI chatbot app. Gemini, if you look at it one way, is the sequel to the Google app, right? This is the app that is the chatbot; this is Google's ChatGPT competitor.

Google looked at the market and said, “You know what would be great for us is if, instead of people just going out and downloading ChatGPT, they buy a Samsung phone and Gemini is already on it.” And they showed some slides in court this week that suggested that before the ruling in one of these antitrust cases, Google was actually thinking about trying to make these deals exclusive, right? Essentially doing the same thing with the Gemini app that they were just going to be caught doing with the core Google app, the core search engine. But because the government came in and said, “No, no, no, no, no,” they went out and made nonexclusive deals.

So that Samsung, you know, maybe will get paid a lot of money to put the Gemini app on a Samsung phone without you having to install it, but Samsung can make that same deal with other people. Why is all this interesting? Google knows that it is at risk of being distracted and losing out on the next generation of search, and it went out and tried to do the exact same thing it had done with search, which was to lock up the market by investing its monopoly profits in creating these default placement deals.

Mm.

So they know what is about to happen to them, and they are using their money to try to prevent it from happening.

That's a really good point. It just seems like the corporate version of a guy who can't stop kiting checks or something. He's on trial for check fraud, and he just pays the bailiff with a fraudulent check because he can't stop himself. They are literally being investigated for these bundling deals, and what do they do? They make more bundling deals, or they attempt to make more bundling deals during the trial.

Absolutely. They are being dragged into change kicking and screaming.

Yeah. So, okay, Casey, net-net, do you think Google is in a different position this week than it was a week ago?

Yes. If you had asked me 5 years ago whether I was confident that Google would lose 2 antitrust cases related to web search and ads, I would have said no. I did think that they were going to lose the ad case. That one, to me, just looked like an absolute no-brainer. The web search case I thought was a little bit shakier.

But Google has now lost both of those. It is going to have to face remedy decisions in both of those. And while I'm sure the appeals will go all the way up to the Supreme Court, we have to remember that the current administration and the Supreme Court have not been particularly friendly to Google across many dimensions. So I think this is a case where Google has used up a lot of its goodwill, and it does not have a lot of friends in high places who can get it out of this jam.

Yeah. I'm glad you brought up the Trump factor in all this, because that's something that's been on my mind as I'm hearing you talk about these cases. We know that many people in Washington from both parties dislike Big Tech and want to see companies like Google broken up. But Trump and his allies seem to have a particular animus against Google. They really, really do not like this company. Can you help me understand why that is and why I think it's probably unlikely that Trump or any of his people will step in and try to save Google here?

Yeah. I mean, it's been a range of things. I think there has been criticism from conservatives about the way that search results are displayed, accusations of shadow-banning, showing results critical of a Republican congressman but not showing results positive to them. So that's been an aspect of it.

They made James Damore a cause célèbre. This was a former Google employee who raised a ruckus, essentially saying that the company was too woke. Damore wound up getting fired. He wound up becoming, again, sort of a cause célèbre for conservatives.

And then, when their AI systems came out, they had those embarrassing moments where, if you asked the chatbot to create the Founding Fathers, it would generate a sort of racially diverse crew and would not always depict them as white. These are some of the things that conservatives have tried to string together to paint a picture of a company that is essentially anti-conservative, and they have wanted to try to work the refs into turning Google into a company that they see as more favorable to themselves.

Right. We know that Mark Zuckerberg and Meta have been furiously bootlicking various people in the Trump administration, trying to get on their good side in hopes of making some of their antitrust concerns go away. Has Google or Sundar Pichai been doing anything similar?

Well, Sundar Pichai was on the dais at the inauguration behind Trump. I think that the company has made some steps to try to curry favor with the Trump administration, but I think it looks half-hearted compared to the complete surrender that we've seen at Meta, for example.

You and I have speculated over the past few months about whether Sundar Pichai will be able to keep his job through this, whether Larry and Sergey, who still run Google, will think, “You know, we need to bring in a new CEO, and we can sort of convince the Trump administration that we're making a clean break with the past, and we want to get on better terms with them.” We haven't seen any moves in that direction so far, but I don't think it's out of the question.

Now, Casey, I want to ask a macro question about all this. I have just become very jaded and cynical about these antitrust trials and cases actually resulting in real, meaningful changes to these companies. We hear all the time about these proposed remedies, these breakups, these split-offs, and then very few things ever actually end up happening.

So I guess my overarching question about this is: Is Google actually going to be broken up or forced to divest a key asset, or is this all going to get tied up in appeals and are they just going to run out the clock?

Well, the clock has been running for 5 years, and we're closer than we have ever been to some sort of meaningful action. So, look, I'm like you as well. I also get cynical about these antitrust actions. They take so long. They get strung out forever on appeal.

But look, 2 judges have now said that Google has a monopoly. There are now very real remedies that are being proposed. Google is fighting back against them, but remember, the people who decide the remedies have already decided that Google has a monopoly, so they're going to do something.

Now, it may be that Google eventually wins on appeal, but unless that happens, you can believe that the judges will require Google to take some sort of action that it absolutely does not want to take to try to address these monopolies.

6. Seasteading Returns To The Sea

Well, Casey, typically we call the second segment of our show the B segment, but today it's more of a C segment, S-E-A. That's a pun.

That's very good, Kevin, actually. I actually have no notes on that. That's exactly how I probably would have started the segment if it had been my turn.

So there was this piece in The New York Times Magazine recently that caught my attention. It was called “The Techno-Utopians Who Want to Colonize the Sea,” and it was by Mark Yarm, who's an executive editor at PC Magazine. Basically, it was about this group of people who want to build these pods out in the water and go live there.

Yeah. And we saw this, and we thought, “Is seasteading back?”

Yeah, so seasteading is something that we've both been interested in for a long time. This was a movement that kind of grew out of Silicon Valley maybe 15 or so years ago, that was funded by Peter Thiel initially. It was sort of part of this libertarian movement of people who were so fed up with regulations and big government and not being able to do what they wanted here on land that they were starting to hatch plans to build their own floating cities out in international waters, where they wouldn't have to obey any of our land-based rules.

Yes, and they were ridiculed at the time and didn't seem to be making a lot of progress. But, Kevin, as the years have gone on, we have repeatedly seen among the wealthiest members of our society this urge to leave the world behind, whether it is going up to Mars or going into space. There is just a lot of energy around removing yourself from society.

And so, when we heard that there were maybe some folks looking to do this once again on the sea, we thought, “We need to learn more about this.”

Yeah, I mean, what makes this story so interesting to me is not just the people who want to leave it all behind and go out onto the ocean and live there, which I think is a pretty small number of people.

But I think there is this desire among a lot of tech people right now for this kind of self-determination, this ability to break away from the existing strictures of society and start their own things. I think a lot of people, especially with these more engineering-oriented brains, look at the world as it exists today and think, “I could do better than that,” or, “My friends and I could set up our own thing that would have all these advantages.”

Yeah, or the one thing holding me back from achieving my dreams is being in a society.

Right. And so I think we're starting to see not just an obsession with seasteading or Mars, but these other, more experimental formats taking off. We're starting to see more experimentation with actual systems of governance. We've had these things called charter cities and special economic zones. We've talked about the people trying to build a new city for tech people in Solano County, California. And I think what a lot of these projects share is a dissatisfaction with the status quo and a desire to use technology as a way to maybe build something new and different.

That's right, Kevin. So I think it's fair to say that with this story, Mark really whetted our appetites to learn more. And so, to find out what's going on, I think we should bring him in.

Let's bring in Mark Yarm. Mark Yarm, welcome to Hard Fork.

Thanks for having me.

So, Mark, you just wrote this fascinating piece in The New York Times Magazine, and the piece opens on this guy named Rüdiger Koch, whom you describe as a German aerospace engineer and long-term Bitcoin investor. When you met him, he was 36 feet underwater, trying to break the Guinness World Record for the longest time living in an underwater fixed habitat. So what's going on here? What's this guy's story?

Well, basically, this was somewhat of a publicity stunt for Ocean Builders, which is the company that Rudi is one of the founders of. Ocean Builders has constructed 3 of what they call SeaPods in and around the bay in northern Panama, and I visited him on, I think, about day 118, about 46 hours before he emerged.

It's a real experience to go down there. Everything has a bluish-greenish tint, with schools of fish swimming by the portholes. It's a very confined space, but it wasn't super arduous because he had somebody come down to clean, somebody provided the food from the marina. It wasn't like it was all self-contained, and obviously he could get visitors like me.

You mentioned that this was a sort of publicity stunt for his company, Ocean Builders. They make these SeaPods. Is their idea that they are going to mass-produce these, and eventually lots of people will be doing what Mr. Koch is doing and just spending many, many days 36 feet below the sea?

Well, the actual SeaPods are mostly for above-water living, living on the water. This was kind of unusual. I don't think many people would voluntarily live in the below-water chamber of this particular model.

Maybe we should just describe what these things look like—

Yes.

Because for listeners who may be wondering what a SeaPod is, there's a picture in your story that makes this sort of look like if you took a dumbbell and turned it on its side.

Yeah.

It has 2 chambers. One of the chambers is above water, and then there's a little rod that goes down into the water—

Yeah.

—and there's, I guess, a spiral staircase inside that rod, and then you get to the lower, below-water chamber, and that's where Mr. Koch was. Correct?

Correct. For all intents and purposes, in Panama they are registered as houseboats. Ocean Builders stresses that they are not in the seasteading movement. This is more of a lifestyle brand. The company has its roots in seasteading. About 6 years ago, Chad and his now-wife, Nadia, had a seastead—a very primitive model of what we're talking about—about 14 miles off the coast of Thailand.

And Chad is Chad Elwartowski, one of the—

Yes.

—co-founders of this company with Rüdiger.

With Rudi and Grant, yes.

Grant. Yeah. Okay.

And they—

And he's another Bitcoin guy, is that right?

Yes, yes, yes.

Yeah. One fascinating detail in your story, Mark, is that the people who are involved with these SeaPods are very adamant that, despite the fact that these things cost about $6 million apiece to build, this is not just a hobbyist project for the rich. One of them actually says to you at one point, “This is not Elysium,” which is the movie where people are sort of orbiting the Earth—the 1 percent sort of orbit the—

Right.

—the ruined Earth in a spacecraft. So despite the fact that these are all very wealthy Bitcoin and other tech people who are doing this thing, you think there's a more innocent explanation that does not have to do with just wanting to escape the tyrannical governments that we all live under—

Yeah.

—and build their own thing?

Yeah. Grant Romundt, the CEO of the company, told me that many people in the Bay Area, of course, are interested in this concept of having a SeaPod. While I was there at the blue-carpet ceremony they had for Rudi emerging from his underwater chamber, they announced this project with the Maldives, which is obviously threatened by climate change and rising sea levels. They're going to have some SeaPods surrounding a very Venice-like, Technicolor city that they're scheduled to build in the Maldives. That's a very practical thing, I think, if a nation is going to disappear.

But the idea, Mark, is that there'll be clusters of these pods, and then you'll just be able to zip down in your Sea-Doo and Sea-Doo over to your neighbor's SeaPod for game night. Is that right?

I mean, that is one vision of it. What's currently going on in Panama is that there are 2 SeaPods, and then there's a third one, which is a prototype, not really used. Grant, the CEO, lives on one of them pretty much full time. He said that going out to shore was, in his words, “ghetto.”

Which, by the way, I think is a very revealing quote about how he feels about the rest of humanity.

It's up for interpretation, I suppose.

Yeah.

He seemed like a perfectly sociable guy.

Yeah.

But I think there was some sort of allure to being by himself on the water in this very high-tech smart home.

Yeah.

Now, Mark, I have to ask you about Rudi, the central character of your story.

Right.

Because on one level, he seems like he shares some qualities with some tech billionaires and other big thinkers in the world that we cover. On the other hand, he seems kind of crazy. There was this big piece in the FT in 2023 that described him as paranoid and vengeful. There were some allegations that he had hired a hit man to take care of some past wrongs that he says were done to him by the Thai government, which sort of broke up his previous seasteading attempt. So give me the rundown on Rudi and what we know about him.

I was nervous asking him about this because, obviously, it was reported in the press, but he denied it, as you can see in the story. We got both sides of the story. I don't know what's going on behind the scenes, but he did admit to having a certain level of paranoia because he was involved in that original Thai seastead, and they definitely did fear for their lives at one point.

I don't know what the truth of the matter is, but he was a very pleasant guy, though, and I did have a good time talking to him down in the underwater chamber.

But you have to say that because otherwise he'll hire a hit man.

Yes. You cannot see the gun to my head.

So, Mark, to me, so much of the appeal of your story is: Look at these fascinating, strange people who have decided, for some reason, to strike out on their own and try to build these—

Right.

—small communities at sea. But there is also this sort of commercial imperative. They want other people to join them in this lifestyle. How is that going? Is this idea getting much pickup right now?

As I said, in the Maldives, they have picked up on this idea. They're taking preorders in a very Tesla-like move. I think it's a limited market, but those who really want it would definitely seek it out, I think.

Yeah, I spent some time with some seasteaders maybe a decade ago.

Oh, really?

I was working on a TV project that never went anywhere.

Where was this?

Uh-huh.

And I was doing some research for that.

You were on a bit of a fishing expedition.

Exactly. And to me, there was just something about this movement that seemed so spiritually empty and sad. I know that sounds harsh, and some of the people involved in seasteading that I met when I was doing this research were quite kind. But there was something so empty about a world in which you become so dissatisfied with the place you live, the community around you, and the government that makes the rules that you follow, that instead of just trying to change the system or work—

Mm-hmm

within the system or reform the system, you just hit the eject button and go take your ball and try to start over in this place in the middle of the ocean that is very inhospitable to you being there. And I don't know. I guess I want to get both of your takes on that.

Mm.

Is there something sad and lonely about all of these tech people who want to go live on these floating cities or in these pods in the middle of the ocean?

To me, it would seem lonely and isolating. To you guys, perhaps, but I don't know. There are quite a few. As you know from doing that documentary, there are lots of people who are into the idea.

So, Kevin, like you, I also poked around the seasteading movement a decade or so ago, when it first came to the fore. It was a really novel idea, and I was interested to learn more.

I remember some of the folks at the time saying, “The reason to do this is that there has not been enough innovation in government. You look at the forms of government that we have in the world, and there aren't that many of them. Maybe if we want human progress to advance more quickly, we should just try more things. We happen to have this legal loophole, which is that if you move out into international waters, you can create these semiautonomous communities and maybe do a little bit of innovation.”

I have to say, part of me was with them on that point. I personally didn't want to move out to the colony in the middle of the sea, but if some people wanted to try it and come back with some new ideas, I didn't necessarily have any issue with it.

What I think is really interesting about Mark's reporting is that 10 years later, we don't really seem that much closer to building communities. We still have the same loners who are ensconcing themselves in underground chambers and trying to sell the rest of us on the idea, and I don't think it's working. I think it does actually seem way lonelier and way less ambitious, frankly, than it did a decade ago.

Well, Casey, I want to offer an alternative explanation here that I see in the failure of seasteading to really catch on beyond a small niche group, which is that the people who are influential and have a lot of money and resources in tech have just realized that they don't have to start their own civilizations, because they can buy the ones that already exist.

Yeah.

One of the biggest changes in our political climate over the past 10 years has been that the people who run Silicon Valley have decided to invest in trying to sway the government of the United States to make friendlier regulations and loosen up on some of the restrictions on their activities.

Mm-hmm.

If you are Peter Thiel or Elon Musk or someone else who wants to build your own civilization, you might actually have more luck trying to swing the existing system in your direction. So that's one idea that I have about why—

Right.

seasteading and movements like it have not taken off: There turns out to be a much more direct route to seizing power and becoming—

Right.

sort of semiautonomous. What do you think about that?

Arguably, Elon Musk has obviously seized a great deal of power, yet still proclaims to want to go to Mars. A lot of the people I spoke to for this piece talked about how we're going to colonize the oceans before Mars, that the oceans, as inhospitable as they may be, are much more realistic than Mars, which is perhaps beyond our technical capabilities, even if others would like us to believe that it is within our grasp.

Yeah. I have to tell you both this anecdote that's just coming back to me from the time I was reporting and researching seasteading a decade ago, which is—

Uh-huh.

I was at Burning Man—

Mm.

and I went to an event that some seasteaders were holding at Burning Man. They love Burning Man.

Yes.

When they did—

Mm-hmm.

manage to build these floating cities in international waters, they were dreaming about how things would go. And it was all men, and they were talking among themselves about how they were going to convince women to come out to the seastead. They just—I will never forget these guys at Burning Man saying, “Well, we could just helicopter them in for a week or two at a time, and then they could go back to land.” I remember thinking, “These people are insane.”

What is with the gender difference? Because there is a woman in your story, Mark, who—

Yes, Nadia.

lives on a sea pod. But I believe there's only one. It does seem like wanting to live by yourself in the ocean is an extremely male-coded activity.

Nadia, who now lives in suburban Indianapolis, was seasteading with Chad, her now-husband.

Yeah, it does tend to be pretty dude-heavy.

Yeah.

But I do want to take this idea of sovereignty and the tech world waking up to its own power more seriously, because I think there is a common thread running through your story, Mark, and a lot of the stories that we've seen about things like billionaires wanting to do space travel, the emergence of the tech right, and these charter cities and special economic—

Mm-hmm.

zones. I think we are at a moment right now where there are a lot of people on both sides of the aisle realizing that things have gotten kind of broken in the world around them.

Some people's instinct in a situation like that is to try to reform systems from within or try to—

Mm-hmm.

buy them or bend them to their will. And some people's instinct is just to say, “I'm out of here. I want to do my own thing.”

Right.

So I do think that is a split within the tech community on how best to engage with the world around them, or whether to engage at all, and I just think that's something we should keep tabs on.

Yeah, we should definitely. Obviously, as climate change gets worse and the land becomes more inhospitable, we're going to be looking—

I would never go to Mars. I don't know—show of hands: who else would? But it's a fantasy of many people.

I think you said, Mark, that the founders keep saying this is not an ideological project, but—

Right.

I think there just is something inescapably ideological about living by yourself out on the ocean, right? If you're a community builder or somebody who likes creating coalitions, you're probably not going to do it from your SeaPod, right? They really do seem to be optimizing for people who just want to remove themselves from society rather than people who want to rebuild it.

In theory, you could have a cluster of SeaPods and visit your neighbors and—

Casey—

Yeah, I mean—

I'll visit you on your SeaPod.

Okay. After I get canceled, and that's the only place that'll have me.

Yes.

And I'm—

Yes.

I'm trying to escape from the Thai Navy. You can come pay me a visit.

I'll see you there.

Well, Mark, thanks so much for your fascinating reporting here, and thank you for going where few reporters have dared to go before, onto the SeaPods.

Well, thank you very much for having me. I appreciate it.

Well, Casey, I'm underwater in email. When we come back, we'll do our Tool Time segment and talk about all the stuff we're using to try to become more productive.

Well, Kevin, I just checked my watch, and it's Tool Time.

Tool Time.

Tool Time, of course, is our segment where we run down the latest AI and other software products we've been using to improve our lives at work and at home. This week, we've got 3 different products that we've both started using in recent weeks. And as we so often do when we discuss AI tools on the show, we like to begin with a couple of disclosures.

Yes.

Now, I wouldn't call your boyfriend a tool, but that's on you.

Well, wow, did I come close to saying something I regret? But here's something you should know: Speaking of my boyfriend, Kevin, he does work at Anthropic.

Yeah, your Manthropic works at Anthropic.

That's right.

Also, I recently learned that people at Anthropic are starting to call him Manthropic around the office because of our podcast. And I would just like to say to those Anthropic employees: "Cut it out. Be nice."

Leave the boyfriend alone.

Yes.

He's very sweet. He's very sweet.

He is.

Now, do you have a disclosure, Kevin?

Oh, The New York Times is suing OpenAI and Microsoft for copyright infringement, of course.

7. o3 Pushes ChatGPT Forward

All right. Now that those disclosures are through, Kevin, let's get to the first tool we want to talk about today, and that one goes by the name o3.

Yes. o3 is OpenAI's newly available model. It became available through ChatGPT last week, and it made a big splash. People were very excited by some of its new capabilities. Tyler Cowen, the economist, said that he thought o3 was AGI—that essentially, it was as smart as we could expect AGI to be.

Lots of people disagree with that. People have been finding things that it's not all that good at compared to past systems. But Casey, give us the rundown of o3, and then tell me how you are using this thing.

Sure. So o3, as you say, is OpenAI's latest and perhaps most powerful model. It is what they call a reasoning model, so it gives a little bit of extra time after you enter your query so that it can do a bunch of cool tricks.

Most interestingly, Kevin, it is the first reasoning model to be able to combine every tool available to ChatGPT in a single model. You enter a query, and before it gives you the answer, it can search the web, analyze your uploaded files, analyze data with Python, reason over visual input—so if you upload a couple of images—and generate images.

Right now, we live in this nightmare world where you have to go to a drop-down menu and choose which of the 16 models you should use to do whatever your thing is. o3 is a step toward saying, "Hey, forget about that. Just tell us what you want to do, and we will give you the best possible answer."

It is the omni model. Yes.

Yeah.

And is this available to all ChatGPT users, or do you have to pay for it?

You have to pay for it, and the more you pay OpenAI, the more queries you will get per week. But I imagine that before too long, this sort of thing is going to come down to the free tier, at least for a handful of queries a week.

Yeah. So there are all kinds of benchmarks out there, and people will talk about, "Oh, o3 did this much better on this benchmark than the previous model." But I want your vibe-based evaluation of o3, Casey. What does this thing do for you, and why is it better than other systems you've tried?

The fact that it can do web searches and document analysis before it gives you an answer means that you can perform some super helpful tasks more quickly than you could if you were doing them in separate steps. For example, when these 115-page rulings come out of the Google antitrust trial, my first step is to upload them into o3, and then I can just chat with the document.

Maybe I'm trying to understand why the government thinks that this is the case, and o3 will pull out the quotes. Now, you may be wondering, do I then go back and make sure that those quotes are real and were not hallucinated by o3? Yes, I absolutely do. So far, I have not found a case of it making up a fake citation. I will continue to check its work.

But that has been super helpful to me, Kevin. I've also been using it for business ideation. I have a couple of things I'm thinking about doing with Platformer, and I basically sat down one Saturday morning and said, "Hey, here's what I'm thinking about. How might you go about this?"

And what did it tell you about starting an OnlyFans?

Yeah, that's right. Who's ready to see these feet?

The thing is, it's actually really good at this. My favorite way to use these AI models tends to be for things where I'm not asking them for critical life-or-death information that I then have to worry they have hallucinated. Instead, it's to say, "Hey, get me the first 10% or 20% of the way through a project."

As I think about doing these more creative tasks, I've found that o3 is a really smart partner. It does have good ideas. Of course, it has terrible ideas as well. But sometimes what you need to get unstuck on a project is for somebody to give you a terrible idea, and you think, "Oh, no, no, that sucks," but it does actually spark something about how you could do it better.

That's how I've been using o3 so far, and I would say I've been a pretty happy customer.

Yeah. I've been playing around with it too, and I've been using o3 sometimes for interior design. I used it to help pick out a rug for my office this week at my house. I used it to help me figure out what was going on with a car seat that was broken.

One of the most fun uses I've seen for o3 is people using it to play GeoGuessr. Have you seen this?

I have seen this.

GeoGuessr is a game where you are given a photo, and you have to guess where in the world it was taken using various visual clues, landmarks, and vegetation. There are people who are super good at this and do it competitively.

There's this guy, Rainbolt, who is the king of GeoGuessr. At least according to some of my friends, o3 is now quite good at this kind of visual analysis, where you can feed it a photo and say, "Where was this taken?" It will inspect various pieces and kind of do that for you.

I did try this on a few of my own photos. It did not get the right answers.

Hmm.

Some people are reporting that this thing is just incredible at that. So take that for what you will.

Let's say that that's very fun if you're playing a benign game of GeoGuessr. You can also imagine the worrisome privacy implications if, all of a sudden, people can upload any photo ever taken of you and say, "Oh, I know exactly where this person is at this moment."

Totally. So, Casey, who would you recommend o3 to at this point, and for what?

I think that I would still place it in the category of something that is cool, novel, and fun for an early adopter. If you really want to be on the bleeding edge, spend 20 bucks a month to use ChatGPT, play around with this, and see what it's useful for. If you find it super useful, maybe you want to go up to a higher-tier plan and use it even more.

If you're just a student or somebody who's casually dabbling with this stuff, I don't think there's anything about o3 that means you need to go out and upgrade immediately. But for folks like us, who are very interested in the state of the art, this was a moment where both of us said, "Aha, okay, we can see the frontier moving up quickly here."

Yeah. In fairness to other frontier AI companies, we should also say that a few of them have gotten upgrades since the last time we did a Tool Time segment. Claude now has the ability to search the internet, which is a feature that we had asked Anthropic CEO Dario Amodei about when he came on the show earlier.

People are also very excited about the new model inside Gemini. Gemini 2.5 is the newest version. All these models are taking big steps up as a result of these new reasoning functions.

Yeah, I would agree with that.

Okay, next tool. Casey, you want to talk about something called Tana. Now, what is Tana?

8. Tana Builds An AI Journal

I've been pronouncing it Tana, but it might be called Tana. Many folks who work on the team are from Norway, and I don't really know how they pronounce vowels over there. But I'm going to call it Tana for this segment.

It was built by some former Googlers, including one who helped build Google Wave, which was a product that I loved way back in the 2000s. It was an early step toward creating this kind of collaboration software. It borrowed elements from wikis and tried to create something really cool. I loved it, but it didn't really take off.

A lot of those ideas wound up surfacing again in what I'm going to call these collaborative workspace apps, like Notion, which I think is maybe the most popular one. I've also talked previously on the show about other similar note-taking and personal knowledge-management apps, like Obsidian or Roam. On one level, Tana is just another one of those, but I have been using it in a way that I've been finding cool and wanted to talk about.

All right, tell me.

What I've decided to do is essentially have one piece of software that I use as my AI journal. I'm a big believer in journaling. I think one reason why I'm constantly talking about note-taking and productivity is that I do believe everyone should keep a journal of some kind. I just think that it's good for you at work, it's good for you at home, and it's just a good thing to do.

But if you put everything into one journal, it can feel very crowded and cluttered, and it feels like it maybe doesn't have the utility that you want. I do most of my daily journaling in another great app called Capacities, which we've talked about on the show. I still really like Capacities. I still use it basically every day.

But once Tana came around, I said, "You know, I have this other idea." Every day, there's a ton of AI headlines, and I'm really trying to make myself as smart as I can about AI. So every day, I'm just going to take all the top headlines about AI, put them in one place, and add tags. Then I'm just going to revisit it throughout the week.

Maybe I'm writing about OpenAI this week. Now I just click on a tag, and I see everything that has happened in my world with OpenAI that week. I can quickly open up that story and figure out what I thought was interesting.

What I'm hoping is that, over time, I will have built this really interesting chronological log of the development of the AI industry that I can dip into at any time, that is very easily searchable, and that I'm hoping will make me smarter about the subject overall.

That sounds very useful—

Yeah.

—but I'm also remembering previous productivity apps that you've described to me that work in very similar ways—

Yeah.

—that allow you to create tags and kind of—

Yeah.

—keep track of topics. So how is this one different or better?

To me, it's useful in the same way that just opening a new browser window can be useful sometimes. You could have just opened the tab in the old browser window, but sometimes it's that blank sheet of paper that actually draws you to using it. That is how I'm trying to use this.

I guess what I'm really suggesting, or what I'm curious about, is whether you think there can be value in having a journal dedicated to something particular—a specific app that I do a specific thing in. Instead of trying to turn every app into a Swiss Army knife, sometimes you just want a hammer. So I'm trying to use Tana as a hammer to organize one particular thing.

Okay. Well, let me know how your experiment goes. I hope this is the productivity app that finally fixes—the gaping void inside you. Casey, how much does Tana cost, and are there any other things you'd like to say about it?

It's a freemium model. There's a fairly generous free tier. If you get really into it, there's a paid plan that starts at $8 a month or $96 a year.

I don't have any particular feeling of, "Oh, you have to rush out and try Tana." You could also try Obsidian, which is free to use in almost every case, and you could definitely do everything I'm talking about there. But I don't know—just experiment with having a single-subject journal and let me know how it goes. Or if you've done that in the past, email us and tell me what you learn from that experience. It could be useful to me.

Now, I believe you are also using an interesting new tool. Kevin, why don't you tell us about it?

9. AI Dictation Tames Email

If you'll remember, the last time we did a Tool Time segment, I said that I was drowning in unread emails. I was desperate for a company, a startup, or someone to come along and build me an app that would allow me to put my entire email inbox on autopilot—to have AI draft responses to all of my emails, let me push one button and send everything, and never have to think about email again.

You predicted that when we published that episode, a bunch of listeners would reach out and say, "Hey, I actually built that app. Why don't you try it?" And that I would try them and be disappointed by all of them.

Yeah. Well, what happened?

Well, Casey, you were right. After we published that episode, I heard from a number of listeners saying, basically, "We have this. It exists. Come try it out." I talked to a bunch of their CEOs, got some demos, and was very excited to try some of these email-autopilot products.

Mm-hmm.

And then I started running into walls.

Yeah.

One of the walls was that I realized my requirements for this program were going to be very hard to meet. My first requirement, as you will remember, was that the email-autopilot program not send any of my data to an AI company in a way that would require me to trust that they were handling it safely.

Right.

The second requirement, which I didn't know when we taped the show last week, was related to something that I'm enrolled in called Google Advanced Protection.

Mm-hmm.

This is a kind of high-security program for journalists, politicians, or dissidents who need a little bit of extra security on their Google accounts. One thing that Google Advanced Protection does is block access to third-party plugins that want to go into your Gmail and use that data, take it somewhere, or analyze it in some way.

That makes sense.

All of these demos suffer from the same problem, which is that in order to use their email-autopilot software, you have to essentially fork over your Gmail account to these companies. Many of them, I assume, are trustworthy, but I'm just not sure enough to give them 20 years' worth of my email.

Yes. So what happened?

Casey, my saga was not done, because I attempted some vibe-coding experiments to try to build my own app that would run locally on my machine. I tried so many different things.

I tried using o3 to build an extension for a mail program called Thunderbird, but it turned out the New York Times tech people didn't allow that either. I tried to build my own Chrome extension that would look at my Gmail as it was coming in and draft responses. That didn't work either because of some security issues.

I even tried creating something called an Electron app, which was terrifying. Eventually, o3 was getting so desperate with all of my crazy requirements, and wanted to help me so badly, that it actually suggested I essentially build my own malware: a program that I could install on my computer that would capture my screen, log all my keystrokes, analyze them, and use that to respond to my emails.

Hmm.

I thought that was a little bit overboard.

Yeah.

But this sad end to my email-autopilot saga is not where I leave you today, because, Casey, I have found a new path through the email wilderness.

All right. Give us some hope here, Roose.

I have become a dictator, Casey.

Yeah, our producers already could have told you that.

Not the tin-pot kind. I have become a person who uses AI dictation software.

Hmm.

We've talked on this show before about how I'm a guy who likes to talk rather than write. Now there are these AI-powered dictation tools that can do a lot of what I've been hoping something like Email Autopilot would do for me, which is to blaze through a bunch of emails very quickly just by talking to my computer. So I found a second route.

Okay. Yeah. So tell us, what app are you using here?

There are a bunch of them out there. There's Aqua Voice, there's Wispr Flow, and the one I've been using most is called Superwhisper.

We've had dictation apps and speech-recognition apps for a long time. They're built into probably every phone and computer now. But these new dictation apps can do more than just transcribe the words you're saying.

They're connected to large language models, so you can talk into them, and then they can capture what you're saying and transform it in some way. Maybe they can summarize it, create a bullet-point list, change the tone slightly, or just take out the filler words and format everything in a very context-aware way.

So if you're writing emails, it can sound like an email. If you're writing messages, it can sound like a message. If you're doing a draft of a piece, it can sound like a draft of a piece. And it can do this across any app. So anything that you're doing on your computer, you just hit your hotkey, start talking, and it types into whatever box you're typing into.

Right. And how much time does that save you, just dictating an email rather than writing it?

I've found it's saving me probably 50 percent of my email time—

Hmm.

—just being able to talk. I talk faster than I type. I think a lot of people do. I'm sure you've seen people going around dictating text messages, and they always have to say, like, “Tell Tom I'm going to the grocery store, period. Does he want cookies, question mark.”

Yeah.

And you don't have to do that.

They sound like they're sending telegrams.

Yes. And you don't have to do that with these new AI dictation apps. You can just talk naturally, the way that you would talk to a person.

Interesting. I have to say, I do not like sitting alone in a room talking to myself. I feel self-conscious about it. It also feels like I'm using more energy than maybe I would when I'm just running my little fingers over the keyboard. So I guess it doesn't surprise me that you love to talk to yourself, but I just don't think it would work for me in quite the same way.

I've got big muttering energy. Is that what you're trying to say?

Yes, like—

So that is my tool of the week. I like these AI dictation apps, and if you are a person who likes to talk, try one out.

All right. Well, those are a few tools for your consideration. And, listeners, if you've been using any cool tools recently that you want to tell us about, maybe even get our take on, email us. We're always looking for fun new things to try.

One more thing before we go. We told you last week that we were excited to tell you who our special guest is going to be for an upcoming Hard Questions episode, and we can now tell you who that person is, Kevin.

It's Pope Francis.

Back from the dead for one last score?

No, it's not. It's Ed Helms, the actor, author, comedian, podcaster, apparently banjo enthusiast—

Mm-hmm.

—and Hard Fork listener, most importantly. Ed is going to be joining us to help answer your ethical dilemmas and tech-related moral quandaries. So please, send them in.

And just if you're wondering what kind of quandary we might be looking for here, here's a couple we've had in the past. There was one that was like, “My top daycare choice is requiring a photo release and the rights to use my child's image in promotional materials as a precondition for registering. What do I do?” Or, “How do I tell my boss that sending ChatGPT-generated content to his team is both unhelpful and alienating?” So, stuff like that. We want to hear it. Send us a short voice memo or even a video of yourself. Keep it around 30 seconds if you can, and send it to hardfork@nytimes.com.