Hard Fork
Hosted by Casey Newton and Kevin Roose
Cryptocurrency is poised to have a huge year in 2025.
16,345 words · 7 speakers
Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.
Support for this podcast comes from Dell Technologies. It's time to put AI to work. Is your business ready? The challenge isn't getting started with AI. It's scaling beyond pilots while maintaining control of your data, governance, infrastructure, and costs. Dell AI Factory provides the integrated foundation you need to scale AI securely and efficiently with one trusted partner, Dell, AI on your terms. From idea to impact. Learn more at dell.com/yourwaytoAI. That's dell.com/yourwaytoAI.
Casey, how are you?
Well, Kevin, I couldn't help but notice you weren't at the Grindr Press Mixer last night.
I was not. I had a previous commitment, but how was it?
Were you invited?
Yes.
Oh, and you just said queer people's lives aren't important to me?
No, I figured you were gonna go. You wouldn't miss that for the world. And you would, you know, give me the rundown.
I, well, I went and I learned some exciting things on the horizon for the Grindr Corporation.
Such as?
Well, the big idea over there is they're like, we don't just want to be for hooking up anymore. They're like, we will always be for hooking up. But, you know, if you're queer and you're traveling around, we want you to open up Grindr and just be like, here is the coolest club in the city. And here is the most interesting thing that's happening right now. And here is maybe somebody who lives 50 miles away, but might actually be your future husband.
Hmm.
And I just want to say, I don't think any of those things are gonna happen. Have you ever met a gay man? These are not the questions they're asking when they're opening Grindr. But you know what? You gotta have a dream. And this is why I love tech industry, Kevin. You have to have dreams.
Wait, so they're, are they trying to branch out beyond queer people? Or they are just trying to expand the menu of options available to queer people?
First of all, I love that you are trying to make straight people on Grindr a thing. How dare you? Can we not just have one space to ourselves? Okay. But no, they are not trying to bring the straight folks in. They are just trying to say, hooking up, you know, it's a problem every app has. It's like you kind of saturate your, your, your first market. And then you're like, I'm a public company now. I have to show growth. What else can we do? What is the sort of next thing we can build? And the next thing you could build is usually something that is not as good as the first thing that you built.
Hmm, yeah.
Yeah.
So would you say they're trying to become more versatile?
I would say they're still grinding.
I'm Kevin Roos, a tech columnist at The New York Times.
I'm Casey New from Platformer.
And this is Hard Fork.
This week, how the election paved the way for a crypto congress. We'll tell you what the Trump administration is planning. Then, it's ChatGPT's second birthday, and head of product Nick Turley stops by to tell us where it goes from here. And finally, a listener wants to know, what social network should he be using in 2024? We'll give you our thoughts.
Well, Kevin, right now, there is a lot of uncertainty about what the next Trump administration will bring, but as we read the news this week, one thing seemed increasingly clear to us, and that is that cryptocurrency is poised to have a huge year in 2025.
Yes, this is one of the places in the tech industry that is celebrating the hardest about the Trump victory. A lot of people feel like this is the end of a brutal period of excessive regulation and what some people are calling a reign of terror that is now over.
Wow, did you experience it that way?
No, but that's what literally people are calling it, the reign of terror.
All right, well, let's talk about some of the things we've started to see as prospects were raised that maybe the reign of terror is over. What are some of the things we saw this week that made us say, oh, wow, crypto actually is one of the big stories coming out of this election?
So the first thing was just that the prices of all the major tokens have risen precipitously and started rising on election night as it was clear that Trump was winning the election. Bitcoin in particular has hit a new all-time high on Wednesday. When we're taping this, it's around $93,000. That's up from $62,000 a month ago. That is far higher than Bitcoin price has ever been.
Yeah, also crypto executives seem to be lining up to speak with the Trump transition team. So a friend of the pod and New York Times reporter David Yaffe Bellany reported this week that Brad Garlandhouse, the CEO of Ripple, a big crypto company, along with some executives at the crypto company Circle, had been in touch with people close to Trump and the transition team. And Coinbase CEO Brian Armstrong has also been trying to arrange a meeting with them. So a lot of people are Trying to get in Trump's ear.
Yeah, and it's not just that the prices of crypto assets are going up. These people also feel like they are making inroads to getting friendlier regulation, to getting more clarity around what the rules are for crypto. And they are also just having a lot of success at the more conventional special interest stuff that the crypto industry has been doing. A lot of the candidates that their groups and super PACs have been supporting won elections and are now going to be in Congress.
Yes, this is more of a crypto congress, I think, than we have ever had. And to put a bow on all of it, Kevin, Elon Musk just got named to co-lead something called the Department of Government Efficiency alongside the former Republican presidential candidate Vivek Ramaswamy. This Department of Government Efficiency is, of course, an acronym for Doge, which is Musk's favorite cryptocurrency and something he has long hyped. And in the aftermath of the announcement of this Department of Government Efficiency, the Dogecoin price popped 20%.
Which I was glad to see because my entire 401k is denominated in Dogecoin. I'm feeling good.
Seems like a smart choice, you know. Speaking of smart choices, nothing says government efficiency like having two people doing the same job. But anyway, today, Kevin, let's unpack what the crypto industry's growing political influence could mean. On one hand, this is a huge win for that industry, which might now be able to realize some of its wildest dreams. On the other hand, I'm worried it could introduce some significant new risks into the economy as the people who once mostly kept crypto separated from core banking and finance functions are about to be swept away and replaced by people who run crypto companies.
Yeah, let's talk about it.
Okay, so let's start with something very basic. How many pro crypto candidates won in the election that we just had?
So it's a little hard to say, like, who is a pro crypto or anti crypto candidate. But according to, there's a tracker run by an industry group called Stand With Crypto that sort of vets politicians based on stuff they've said about crypto in the past or positions they've taken on crypto-related bills or initiatives. And according to that tracker, as of Wednesday, 272 pro crypto candidates were elected to the House this year and 19 pro crypto candidates were elected to the Senate. Paul Grual, the chief legal officer of Coinbase, recently said that we now have the most pro crypto Congress in history, and I think that's true.
Yeah, and it sounds like Stand With Crypto is doing much better than its counterpart industry group, Sit With Regulation. They really got wiped out this year. So it seems like the crypto folks won some very tight races this year. Kevin, tell us about Bernie Moreno.
Yeah, so this was a race that a lot of people in the crypto world were paying close attention to in Ohio, my home state. This was an election between Bernie Moreno, who is the Republican Senate candidate, and the Democratic incumbent, Sherrod Brown, the long-term Senator, staple of Ohio politics for many decades. And Sherrod Brown is a sort of a centrist Democrat, but he's also been seen as a major opponent of crypto. He was an ally of Gary Gensler, the Chair of the Securities and Exchange Commission. He's also been antagonist of the crypto industry on a bunch of different bills that they have wanted. And he became a big target of the crypto industry and its lobbying and influence groups.
The crypto super PAC, Fair Shake, in particular, spent, reported $40 million to support Bernie Moreno in unseating Sherrod Brown, and it worked. Bernie Moreno, who is a longtime crypto supporter and actually started a blockchain company in 2018, he beat Sherrod Brown and will now be in the Senate.
And Kevin, with all of these crypto candidates winning, can I assume that Bernie Moreno and the others made crypto the centerpiece of their campaigns?
Well, not really. So Bernie Moreno is a longtime crypto supporter. He has a long track record of supporting this stuff. But what's interesting about the crypto industry's lobbying efforts and their donations to candidates in this cycle is that a lot of them did not play up crypto as a big issue in the campaign, right? They were just sort of funneling money to politicians who they knew were more sympathetic toward crypto, but it's not Like those politicians were turning around and making their entire campaigns about deregulating cryptocurrency or something like that.
Yeah, and this seems somewhat cynical to me. Jessica Piper had a great piece in Politico where she wrote about some of the ads that these pro crypto candidates were running. She writes, "In Ohio, a woman shares how GOP Senate candidate Bernie Moreno invested in her business making uniforms for female police officers. In Iowa, a spot highlights Representative Zach Nunn's military credentials and work to lower childcare costs. And in Colorado, a Spanish-language ad shouts first-term Democratic Representative Yedira Caraveo's support for more Border Patrol agents." What do these have in common, Kevin? These were all ads from pro crypto PACs, and none of them mentioned crypto at all.
Yeah, and I'm sure that was a strategic decision that they made based on focus groups and polling of how voters felt about crypto. But, look, I think it's fair to say that they had a very good election cycle. The crypto lobby got most of what it wanted. Most of the candidates that they supported were elected and some in very tight races. So whatever their tactics were, I think it's fair to say that they worked.
Yeah, they worked, but I think it's important for people to note that we are not saying this was some sort of crypto election where Americans stood up and said, you know, "Please let me go buy unregistered securities." We're saying that a lobbying group funneled money to people who supported their cause, and they sort of worked very quietly to get those folks elected without talking about what they wanted them to do.
Sure, but isn't that how influence groups in this country work in general? I mean, the oil and gas industry does the same thing. It's not like, you know, candidates show up suddenly talking about how much they love fracking. It's just that pro-fracking groups will give money to their campaigns. I'm not saying, I'm not saying this is like a good way to run a country. I'm just saying like this is basically just crypto catching up to what other industries have been doing for decades.
Well, I'm actually very disappointed to hear that, and I'm calling on the oil and gas industry to do better.
Now, Kevin, are all these pro crypto candidates the same? Like, do they all sort of share a kind of sort of common platform and set of common beliefs?
No, I mean, what they have in common is, in the crypto industry's eyes, a sort of track record of at least being open-minded about the possibility that crypto could be a kind of mainstream asset, that you could have
Bitcoin as part of something like a strategic reserve that you could have more clear guidelines that allow crypto companies to build stuff without worrying that the SEC is gonna sue them and try to put them out of business. Basically, they've assembled a group of legislators and candidates who they believe will usher in the kind of crypto regulation that allows this industry to become a mature, full-fledged part of the US financial system.
Okay, so we have a crypto congress that will be seated next year. Let's get into what we think that means, Kevin. And I think the first thing to say here is that the casino is back open, right? I feel like during the 2021 run that crypto had, folks were talking to us regularly about how their coins were useful in various new ways, enabled all sorts of new and exciting things, but let's face it, most of their popularity just comes from the fact that people love to gamble on them, right?
Well, so I've been talking with some folks in the industry since the election. And I'll just tell you what they're telling me without endorsing their beliefs or saying that they actually reflect reality. But what I'm hearing is that actually the status quo of crypto regulation and politics in the US has been part of what has created the casino atmosphere around crypto because it's hollowed out the market. And what I mean by that, or what they mean by that, is that you have, basically, the things that are considered kosher or legal to do with crypto right now include betting on Bitcoin, which is safe in their eyes from regulation. It's the oldest, most established cryptocurrency.
It's treated differently by regulators than some of the newer tokens, and so it has the special status when it comes to how it's regulated. And then you have the meme coins, the meaningless, frivolous tokens that are considered too unserious for regulators to care about, and so you can bet on the prices of those things to go up.
Like Dogecoin.
Like Dogecoin. But there's this middle of the market that they see as being full of actual, serious applications of cryptocurrency to things like real estate, to things like Music royalties, these kind of things that you'll hear people in the crypto industry talk about as being part of DeFi or Web3, these sort of attempts to move more parts of economic activity onto crypto blockchains and have them work in that way.
Right, so anything about which somebody might say, "Put it on the blockchain," you're saying is in this category?
Well, it hasn't been impossible to do those things, but it's been quite hard for institutional investors in particular to want to fund those kinds of activities because of this sort of uncertain regulatory environment where if you start a startup and it's trying to do something using crypto to do something with music royalties, for example, you don't know which government agency could show up tomorrow and try to put you out of business.
And this was the Andreessen Horowitz argument for why they backed President Trump, was essentially saying, because of this, our startups cannot get funding, they cannot succeed in the marketplace, and we want that to change.
Right, and so I think the people who I'm talking to in the industry, they believe that actually the casino part of crypto may become less prominent as a result of having a more crypto-friendly administration.
Well, maybe it will, but I'll hit you with just a few numbers from over the past few days, Kevin. So as of the time of this recording, not only is Bitcoin up above $93,000, as we said earlier, Ether is up more than 10%, Dogecoin is up more than 118%, and Coinbase's stock is up more than 40%. So while there might be other things you can do with crypto, to me, it seems clear that the gamblers are getting very, very excited. Now, let's talk about a second thing that you can be sure of in the next Trump administration, which is the crypto industry is going to be regulating itself now. Talk to us about what changes we're expecting to see over at the Securities and Exchange Commission.
So again, I'll tell you what people that I talk to who are plugged in in the crypto industry are expecting. This is what I would consider the consensus industry view, is, look, the most obvious thing is Gary Gensler, the head of the SEC, will be gone. He is not expected to serve a day in the Trump administration.
Wait, what if he puts out a statement later this week and he says, "I love crypto now," and President Trump, "Well, he's got some good ideas."
It worked for Jeff Bezos!
I think he should be polishing his resume.
All right.
I don't think there's any chance of that happening. And I think the person who will replace him will almost certainly be someone that the crypto industry knows and who they consider sort of a strategic ally. Some of the names that have been floated as potential replacements for Gary Gensler are people like Dan Gallagher, who's the chief legal officer for Robinhood, which is, of course, the big finance and crypto trading app. Chris Giancarlo, who's a former regulator who's served as an advisor to crypto companies. You know, none of these people are sort of shoe-ins, but I could see any of them being named to lead the SEC.
All right, so the foxes are going to be appointed to guard the henhouse industry. I also hear, Kevin, that they want to get some new bills through Congress.
Yeah, so there's a bill that has been sort of making progress through Congress called Fit 21. Fit 21 is sort of the crypto industry's attempt to create some of this regulatory clarity that they've been so desperate for.
And I have to say, it sounds like a diet that you would do before your wedding.
Yes, yes.
You look great. I've been on Fit 21. I'm sorry, go ahead.
So this is basically a bill that would create some rules that determine essentially the market structure of crypto. So which tokens, which crypto assets are considered securities, which are considered commodities, which are regulated by the SEC, which are regulated by the CFTC, basically creates some sort of rules of the road so that crypto companies who are doing these things know what the rules are and what lines they need to avoid crossing. And that's sort of the crypto way of describing this bill. Another way of describing this bill would be that it is a wish list of all the stuff that the crypto industry has wanted. But interestingly, there's been some reporting in recent days from some crypto news outlets,
including an outlet called Decrypt, that actually Fit 21 may not be ambitious enough for the current crypto industry because it was written during the Biden administration when this was sort of the crypto industry's last ditch attempt to save itself from sort of being regulated out of business by creating this bill that if it had passed would at least allow it to continue operating in some capacity. Now that Donald Trump has won the election, some crypto lobbyists are saying, wait a minute, we should go back to the drawing board and come up with a bill that actually includes all the stuff we want, and maybe we can get that passed.
We should write a Fit 22 or even a Fit 23.
Yes, exactly.
So we can expect to see that moving through the next Congress. And then, Kevin, there's also this one other idea that I've heard floated that I have to ask you about. Who are these people who want the government to establish what they call a strategic Bitcoin reserve?
Yeah, this is an idea that's been, I don't know, a pipe dream of many crypto boosters for years is that eventually cryptocurrency would become such a valued part of the financial system that countries, including the United States, would start stockpiling Bitcoin or some other cryptocurrency the way that we stockpile other global currencies as part of our strategic reserve.
Or like gold.
Yes, exactly. And, you know, this has not happened in any major economies. El Salvador has started to establish a Bitcoin reserve.
Kevin, it's not just El Salvador that's amassed a large store of Bitcoin. It's also Norway and Bhutan. So put some respect on their names.
Okay, I didn't know that. But basically, the crypto industry thinks that this would be a huge win for crypto for a couple of reasons. One is, obviously, if the government is buying something and stockpiling it, it's not going to ban it, right? It sort of gives them some protection against some of the more extreme regulatory scenarios that they are afraid of.
Right. Also, it's just a huge mark of validation, right? If the government thinks this has value, you should too, right?
Right. It creates an additional layer of legitimacy that makes it easy for lots of investors to say, well, if the US government is doing it, I guess we can do it too.
Do you think that if we established a strategic national Bitcoin stockpile, America could become the next El Salvador?
One can hope. I hear it's beautiful this time of year.
All right, so I want to ask three questions to finish this one out. And the first one is, will Web 3 make a comeback? Here's why I care about this. You know, I'm not a gambler. I never, you know, bought cryptocurrency to try to, you know, see if I could make a quick profit. But I was very interested in 2021 and the people who were saying, we think we can build cool new applications using blockchains. And we sort of want to remake the internet in the image of crypto. That proved very controversial, and I think I eventually became persuaded that that was a super bad idea. But one reason why it never really took off was because the Biden administration just hated crypto and tried to get rid of it wherever it could. So now that that is kind of off the table, do you think we're going to see the Web 3 comeback?
I don't know, but it's certainly possible. I think the thing that I can say with confidence is that crypto will now have the opportunity to prove itself, right? For years, people in the crypto industry have been saying, well, we would have all this cool Web 3 stuff and this decentralized internet if only the regulators would let us. Well, now the regulators are going to let you. So let's go build some stuff and let's see how it does. There's a possibility that it does come to fruition and that all the things that the crypto industry has been saying are sort of barriers to its growth and mainstream adoption turn out to disappear under the Trump administration. I also think there's a possibility that we have more blowups and people losing money as a result of scams or protocols that aren't as risk-free as they're advertised.
So actually, I think there's a real risk here for the crypto industry. I think in a lot of ways, they're kind of like the dog that caught the car. It's like, well, you're going to get everything you've been wanting. But now, if you fail, if you don't do what you have promised your believers that you would, that's going to be totally on you. There's not going to be a sort of evil regulatory regime to blame anymore.
I agree with you and at the same time, I sort of feel like this has already happened because a good amount of crypto applications did remain legal and possible during the Biden administration. You might remember the Constitution DAO, the group of people who came together to buy a real copy of the US Constitution and sort of manage its career using something called a decentralized autonomous organization. After that happened, I feel like we never heard about DAOs basically ever again. There was also some momentary enthusiasm for what they called play to earn games. These were games that pay you in crypto for playing them and some of the biggest ones turned out to be Ponzi schemes.
So while I think everything that you said is true, I tend to be a little bit more skeptical because I feel like the world got a pretty close look at these crypto applications a few years ago and they just decided these really kind of suck in a lot of pretty structural ways. Okay, question two here is essentially how far will Trump actually let the crypto folks go? Right? At the end of the day, the cryptocurrency is a parallel financial system that operates in some ways independently of the US financial system and away from regulatory oversight. And there are a lot of reasons why most countries don't want that to happen.
And if you're say, an incoming authoritarian leader, you might have even less interest in there being a parallel financial system that you don't actually control. So I'm curious if the optimism that the crypto folks are feeling right now might soon run into a hard wall, which is Donald Trump's self-interest.
It's possible. Look, I don't think that Donald Trump is, you know, an erudite scholar of crypto market structure and economics. I think he's basically doing what the people around him want him to when it comes to crypto. But I think there's this tension, right? That crypto at its core was designed to not require the kind of centralization and the planned market structure that traditional financial instruments do. It was created to be an alternative to the financial system. And now it's possible that with the loose regulation that the Trump administration is likely to put on it, that that whole crypto economy just kind of merges with the conventional financial
system and they become one and the same. And I think that would be a good thing for people who want to make lots of money doing crypto stuff and convincing mainstream investors to invest in it. But I think it also introduces a kind of risk to the whole crypto ecosystem. You know, we've seen crypto companies and projects blow up a lot in the past, and those have been big stories. We've talked about them on the show, but they have not sort of cascaded through the financial system in the way that, say, the 2008 financial crisis, the credit crisis at the big banks did. But as these two economies sort of become interlaced and interwoven more and more,
I think there's a real chance of contagion and of systemic risk in a way that they haven't been with these earlier crises, which were much more cordoned off.
Sure. So I mean, you know, I have to say, Kevin, nothing I have seen so far makes me feel like the crypto industry is taking this seriously. And I was glad that the regulators that we still have today do take that seriously, right? I do think that there is a real systemic risk here that comes with intermingling the crypto system ever more with what I think of as the real US financial system, right? As banks start to store more of these assets, as it becomes more of a tool in finance, if, as often happens in crypto, there is a sudden collapse in prices, we could find ourselves in these liquidity traps, you know, banks going under, retail
investors losing their shirts, and who is going to be left to protect us, right? We're talking about moving out of the realm of, oh, well, you're kind of a, you know, odd and quirky person, and you follow certain subreddits, and so you have some Bitcoin and Ether into, oh, no, this is now just a tool that your bank uses, and now your bank just went under.
Yeah, I mean, I am not as worried about systemic risk as it sounds like you are, but there is something else that worries me, which is that I think we are rapidly turning into Nation of gamblers. You know, this is broader than crypto. This is also the story of, you know, the, the rise of legalized sports betting, of prediction markets, of all of these new ways that people have of doing what is essentially speculation or gambling. We just are starting to see that legalizing a whole bunch of forms of what are essentially gambling has not been an entirely net positive thing for society. And that some of the people that it may be the worst for are the people who can least afford to lose money.
Well, now I feel bad about the way I wanted to end this segment, which was by asking you to make a bet.
What's that?
Will Sam Bankman Freed be pardoned under the new Trump administration?
I don't think so. He gave too much money to Democrats.
I'm going to bet that he's the Secretary of the Treasury within three years.
I'll take the other side of that.
Okay.
When we come back, a conversation with ChatGPT product head Nick Turley on the product's second birthday and where it's going from here.
Support for this podcast comes from Dell Technologies. It's time to put AI to work. Is your business ready? The challenge isn't getting started with AI. It's scaling beyond pilots while maintaining control of your data, governance, infrastructure and costs. Dell AI Factory provides the integrated foundation you need to scale AI securely and efficiently with one trusted partner, Dell, AI on your terms. From idea to impact. Learn more at dell.com/yourwaytoAI. That's dell.com/yourwaytoAI.
AWS AI is how the world's leading organizations are transforming their industries. Not in theory, but in production at scale right now. Like turning race day data into real time insights for Formula One fans. Reinventing the journey for every United traveler. And personalizing every trip with booking.com. However, big the question for your business or industry, AWS AI is how.
It's time to innovate at the speed of AI without compromising control. That's OutSystems. Take a look at what happened with one global retailer during a period of massive expansion. They used OutSystems to build AI agents to automate invoicing and translation processes, shifting 95% of their daily store operations to a new digital app, saving thousands of hours of manual work. OutSystems gives you the flexibility to deploy how you want and the governance to do it safely. Own your agentic future. Learn more at outsystems.com/nyt. That's outsystems.com/nyt.
Well, Casey, we have a birthday to celebrate today.
Oh, my God, I'm sorry. I didn't get you anything.
It's it's not my birthday. It's ChatGPT's birthday. The popular app is two years old as of this month. And I would say it had a pretty eventful first two years of its life.
Yeah, this was one of the biggest launches of a new piece of technology that we've seen in a long time. And I would argue is probably the biggest launch since we started this podcast.
Oh, for sure. And not only is it big in our world, it is also big in the world at large. According to SimilarWeb, ChatGPT is now the eighth most visited website in the world. It has millions and millions of users, and it has even become a verb. In fact, I recently heard a college student say that they were going to chat something rather than Googling it. So this is clearly a profound and important product. Whatever you think of generative AI, it has already had a ton of implications for the tech industry and for people's lives around the world. But we haven't actually dug deep into the product itself in a while, and it has been changing a lot. So today, we've invited on Nick Turley.
Nick Turley is the head of product at OpenAI for ChatGPT. He's been at the company since before ChatGPT launched, and he has a lot of interesting thoughts about how the product has evolved over the last two years and where it's going next.
Let's bring him in.
Before we do, we should make our traditional caveat. Every time we talk about this company, The New York Times is suing OpenAI and Microsoft for copyright infringement.
Nick Charlie, welcome to Hard Fork.
Thanks for having me.
So let's start by rewinding back two years to November of 2022. You're about to release this thing called ChatGPT into the world. What did you think was going to happen when it was released?
Yeah, we've been playing with ChatGPT internally. We had a version of it. We called it the chat playground and none of us thought it would be a product. So we had very, very tempered expectations toward this thing. And but we wanted to get it out. It was very important. I think we went sort of, it was a 10 day sprint from deciding to launch this externally to actually doing so. So everyone was pretty heads down. And the reason for that was that, you know, we really wanted the learnings back when we came back from the holiday. If we'd known it was going to be a takeoff product in any way, we would certainly not have done it right before everyone was leaving to, you know, go on break and see their families, et cetera.
So what was the moment that you first realized, oh, wow, this thing seems to be resonating with people more than we'd expected?
It was the next day because, you know, this is probably a classic story for any product that takes off. But I thought the dashboard was broken because we'd taken sort of these over or under bets with the team that, you know, how people would ever try it. I think the sort of outlier bet was 100,000 users would ever try it or something like that. And I was like, that's crazy. You know, I'm from a small 30,000 person town in northern Germany, and I remember going home two weeks later. And normally it's a very serene environment where, you know, that world doesn't interact with my Silicon Valley world and kind of like it that way. But this time, I remember the kids, the neighbor kids were talking about ChatGPT already. And it was two weeks later and it was unprompted, like I didn't even ask them. And I was like, wow, worlds colliding. This thing really is going around the world. So that's probably when I viscerally felt it, but I saw it in the data like on day, on day two.
So I think, you know, a lot of people that I talked to, they tried ChatGPT when it came out. It blew their minds, you know, they tried a bunch of different prompts. And then maybe it sort of didn't work its way into their daily routine. Maybe their usage has sort of trailed off a little bit. If someone's last experience with ChatGPT was back in, call it late 2022, and they went back today and revisited it, what do you think the biggest differences they would notice are?
Yeah, if I think about some of the big things that have happened. GPT-4 was like the first big update we made to the product, right? And what users noticed was it's just much more creative. And it just feels smarter. And then there's all these capabilities that we built on top of GPT-4 because GPT-4 was so smart. So for the first time ever, it could browse the web. It could actually run code, which is not something that most consumers care about. But what that means for you as a user is that it can actually crunch numbers, which historically the first version was really bad. It was bad at math, any type of analysis or calculation. Could suddenly do it because GPT-4 was smart enough to literally run very, very reliable code for you, which is a nice kind of connection of these large language models with traditional computers.
Later that year, we started venturing into multimodality, which is our term for giving it inputs other than text to play with. So it could start generating images, it could start viewing images. We launched our first voice mode, which allowed you to talk to the model. So I think year one was really the year of all these new capabilities that were unlocked by GPT-4 and related innovations. Year two was, we kind of saw this interesting consolidation, where all these new capabilities came into one model, which we call GPT-4O. And that model was itself, it was anything in, anything out. And the way that that feels is that you could just talk to ChatGPT in an incredibly natural way, like the new voice mode that we launched earlier this year. It's a completely different way of using ChatGPT because it's your companion.
You can have a real conversation with this thing. And then most recently, and I think this is maybe the rawest form of ChatGPT or the rawest evolution is, it starts to reason for you, which basically means it thinks before speaking, which we very much value in humans.
Rarely, it almost never happens on a podcast.
No, you get the raw chain of thought, as we call it, right? Yeah, that's right.
There have been a few studies recently of these kind of large sets of transcripts from AI chatbots, including ChatGPT. Basically, researchers trying to figure out what are people actually using these things for? And it seems like these researchers are finding that the most popular use cases are, number one, doing homework, number two, sexual role play, and then all the other use cases. Does that match your own understanding of how people are using ChatGPT?
For us, the order, and you know, obviously this varies about different products. For example, there's these like character products. ChatGPT is very much not that. It's, you know, I always describe our use cases as worky. They're not necessarily at work, though we, I think we are in like 90% of Fortune 500 companies or something like that, something crazy. But it's worky stuff. It's like transit, like I'm trying to get something done. So, you know, there's other products out there that, you know, will skew the, the distribution of use cases. For us, I think the ranking is question answering. That's how everyone gets started. I just ask, you know, and that just got better with search. So I think seen an increase in that. It's the most relatable use case for many people. Then writing that both, you know, can be like, hey, draft this tricky text that I don't know how to send. Or, hey, I wrote this email, but you know, English is my second language. Can you please edit it? Then it's coding.
We have a huge base of people doing technical things. That includes beginners who are learning how to program. And also, you know, experts who are looking to debug what they're working on, or they're running something Excel and it's not working.
And then it's a long tail of other things.
Have there been any popular use cases that you didn't expect or that really surprised you?
So many. Cooking was a really interesting one because I got into cooking just as, you know, ChatGPT was happening too. And I was like, this is amazing that you can just like sub out, you know, oh, I'm missing this ingredient and it'll just like slightly modify the recipe for you in a way that, you know, I probably would have called my, my parents. I still do, but I don't have to bug them anymore.
A lot of people use it as kind of a coach or self-improvement tool. This has been really striking to me, actually. You know, a lot of people use it to kind of process. I do this actually on my way to work every morning. I, you know, when I'm in the car, I talk to it and I just give it a bunch of unstructured thoughts of everything that needs to get done and have it repeated back to me. So it's almost having this like kind of second brain or coach thing that surprised me, both in my own usage and what I've been seeing with other people. Because it's really kind of helping people be a better version of themselves. And I think-
Say more about what you get back out of that. It's sort of like distills everything that you've told. I mean, are you doing that because you sort of want to get like a to-do list? Or like, what is it giving?
Kinda, I, I wanted to sort of help me structure my own thoughts. I- Kind of ramble at it. And by the time I finished the call with it, in part because it's getting me to talk, probably, and in part because it's actually repeating back to me. Okay, here are like priorities I'm hearing. And I'm like, no, that's not right. Like the the fourth one, I actually scratched that. I have this like ready to go list of what my day is going to look like. And so I think it's some combination of getting me to talk and some some and actually, you know, producing a tight list of stuff to do.
Yeah. So I have been talked about this a few times, but over the past month, I have been meditating more and I have been chatting with a chatbot and it has been coaching me. And it is incredible. It is this journal that talks back to me now. So it's not just me like processing my own thoughts. It is something that has a vast body of knowledge about this very particular subject and is able to say, here are some new things that you can try. So this is something that I feel like most people have not caught on to yet, but more people should try.
I want to talk a little bit about the trajectory here you've sort of described over the past couple of years, how ChatGPT has grown more capable, it can work in different modalities. But give us a sense of where all of this is going. Will ChatGPT always be a box we type into? Or where do you think it will be in two years from now, let's say?
First of all, what I'm about to say, I say with a lot of humility because I hope, you know, the
Also rare for a podcast, by the way, but go on.
Rare for a podcast. Yeah, I have been so wrong and I tend to underestimate how quickly things are moving, even myself.
I think there's going to be a few evolutions. The first one, we really saw, you know, the last couple of years, it's been an era of asking questions to the AI. And you can ask increasingly, you know, interesting questions, get increasingly better answers. You can ask them in new ways, including voice, including with vision. You can get different types of answers, including images and including voice itself, right? But fundamentally, it's been this interaction paradigm where I'm asking and I'm getting something back. But when I look at where the technology is going, we have really, really smart models now. And in a certain sense, I think chat under leverages what they're capable of. And what we're going to see instead, I think, is tasks. You delegate to this AI and it goes off and does stuff for you. And
You know, this idea isn't new. People have been talking about, you know, agents for a couple of years now. People even say they're building them and it's true. You can kind of build these things already in certain domains. And it reminds me of how building chatbots felt before we had the right technology where you could sort of do it, but they weren't really good and they weren't reliable. And I think by having these really smart models out there and, you know, our O1 model is incredibly smart model. I think you can build reliable tasks for the first time where it goes off and does something, whether or not it's planning a full trip for you, whether or not it's buying, finding you, you know, the best shoes under $300 that are, you know, made in the US, whether or not it's, you know, coding an application for you, running a data analysis for you. I think you'll be able to go off for 30 minutes, an hour, maybe two hours,
maybe even longer and do this stuff. And the reason that this is possible now versus a year ago is that the models are smart enough to deal with all the edge cases that they're going to run into, right? This is why it's so hard to build these things is you encounter things in the wild, like, you know, the shoes were sold out, or in the coding case, it didn't compile, it didn't run, and now what?
You're reminding me of the evolution of self-driving cars. You know, like a decade ago, when I would test a self-driving car, I would go down to Google headquarters and they put me on a track that a car had already gone around probably thousands of times before. They knew every single edge case in that particular parking lot, and it was super cool, but it could not really do anything that was not on that track. And then bit by bit, they started testing them on more and more roads. They started gathering more and more edge cases. And now, you know, when we leave today, we could order a Waymo and take it anywhere. It sounds like agents are just sort of a little bit behind in that process, but the basic logic is the same and you're driving toward a similar outcome.
Yeah, I think that's right. And I think that it's a really good analogy. And even even within, you know, if you accept that, you believe that that's what the evolution's gonna be, I think there's still the same kind of, you know, big fundamental questions that people have in self-driving cars, which is like, you know, is the future a world where, you know, this gradually gets better and it like takes over and more and more and I don't have to touch the steering wheels often anymore? Or is it this thing that is fully autonomous, but it's only going to work for certain use cases? Like, there's these different paths you can take. And, you know, we're trying to figure out too, what is the right path? Because, you know, we want the human to be in control. We want these things to be reliable and useful and safe. And I think you're going to have the same questions that you have in self-driving in terms of how to actually get there. But I think the end state is very, very apparent, at least to me. And I think many people working on this stuff, the question is how fast and what's the path?
We talked to a lot of educators, teachers, people who are excited about tools like ChatGPT, but also a lot of people who are worried about them. This was obviously a major early use case of ChatGPT was students using this to write their essays, do their problem sets, do their homework. So I want to ask you about that, but in particular, I want to ask you that the Wall Street Journal reported earlier this year that OpenAI had developed a tool internally for watermarking text, basically a system where you could take a piece of text and tell with something like 99.9% accuracy whether it had been written by ChatGPT. The Journal reported that that had not been rolled out after a lot of internal debate. Can you tell us more about that watermarking tool and why you haven't rolled it out?
I'm very much not the expert on this area. We have done research on this, on this stuff. In the case of images, we rolled out ways to detect whether or not images came from DALL-E and it remains an area of research. The thing that I will say, some of the considerations that go into stuff like this are, you know, on the one hand, you want to be able to tell if people use AI. On the other hand, there are many legitimate use cases. Maybe your English is your second language and you use it to edit your cover letter, in which case you wouldn't, maybe wouldn't want people to know that you use ChatGPT. I'm personally not sure, both about how good the detection tools can get and about what the right thing to do is in terms of who gets access to detection tools, et cetera.
But something we researched and I'm, you know, I'm on the product side and therefore don't, don't, don't know the latest.
Do you, I guess, just without getting into the specifics of that watermarking tool, do you feel like ChatGPT has any sort of responsibility to the people who use it and the people who may be trying to pass off AI-generated work as original work, given that this has become a big use case of the product?
Yeah. First of all, I and we feel a tremendous amount of responsibility to our users, our non-users, the broader world and the ecosystem, right? Everyone is affected by this technology. And we've generally found the best thing we can do, given that this technology will evolve with us or without us, is to release it iterally to the world and to start the right conversations and to be as gradual and thoughtful as we can in the way that we introduce this stuff. The, you know, the-
I very much believe that this technology is a tool for people. Right? We need to put people in control about how they want to use this stuff. That is true if you're using it for art. It's a tool for an artist. It's not an artist. And that's the same thing when you're writing stuff. This is not anything other than a calculator for words. And that is how I think we should treat it, which means that the onus should be on the people using the technology in the same way that we've treated any other tool.
Hmm. So the other thing that we'll hear from educators often is not specifically about cheating. I think a lot of schools are starting to move past the the kind of panic that ChatGPT is just going to lead to an outbreak of cheating and plagiarism. But there are some longer term concerns that people have about the fact that students may be using this kind of technology to short circuit their their thinking process. That if you have a chatbot that can, you know, teach you everything you need to know about Chaucer, you don't have to do the work of of thinking through it yourself. Is that concerning to you?
Yeah. First of all, it's something that, you know, absolutely concerns me. It's something I think about a lot. And the the way it makes me feel is I was in in, I think, middle school when Wikipedia came out. And, you know, it was a shock to to our teachers, far smaller of a shock than, I think, ChatGPT to grand scheme of things. But, you know, there's a lot of concern, you know, is going to the library dead? Are people going to lose their research skills? This is not a legitimate source. We ought to teach people the right way to do this. And, you know, I think that forward looking teachers at the time, they were started to bring Wikipedia into the classroom as, you know, a way of discussing, you know, what does it mean for a source to be reliable? Let's look at Wikipedia. Now let's look at reliable source. Let's find the gap.
And in the same way, I do see many teachers now bringing ChatGPT into their classroom. Some of the coolest examples I've seen is, you know, examples where students create their first draft without the help of ChatGPT. Then they submit another draft that, you know, has them edit it with ChatGPT and explain what they learned from the edit. And then there's the final version, right? And in the same way that I think we've integrated calculators, we've integrated other tools that were pretty disruptive and could be abused for, you know, undue shortcuts. I think chat GPT is going to end up being the same. The thing I want to acknowledge is that all this is happening so quickly that I think many teachers just don't have the opportunity to talk to other teachers to, you know, figure out, okay, how are we going to evolve our lesson plan to use this stuff in a way that is exciting and empowering?
Personally, I still believe that in net, this has been amazing for students because the thing you have to keep in mind is that there's people all around the world that don't have, you know, educational resources. Even for me, you know, in the, you know, German public school system that I grew up in, people rarely read my homework. And I often imagine what would have happened if someone had actually read my homework, I would have been.
Wait, your teachers didn't read your homework?
Our teachers were very, very busy and homework was oftentimes this thing you submitted. Um, and then, you know, it was graded on completion.
This doesn't sound like that good of a system to me. I hate to say it.
I came out okay.
Yeah, that's true. That's actually a good point.
I'm overall, overall very grateful to the system, but, um, but I'm just saying that I, I wish I'd had this as a kid.
Yeah. I want to ask about another debate that has, it seemed like it really heated up just over the past few days. And it is about whether, um, as companies like yours work to build these new models, they may be approaching the limits of what are sometimes called scaling laws. The idea that as you pour more compute and data into networks, systems like chat GPT get radically more capable. Do you have a sense that we may be approaching a plateau here in what we can get out of large language models just by making them larger? And is that a problem for you?
Yeah, this keeps coming up. Um, you know, this is my take as a product person. I'm very much, you know, not an AI genius, but I'm around the stuff. And I'll say this, this thought comes up from time to time. In fact, I think we have this. This thought when I first showed up at OpenAI, because I remember when GPT-4 wasn't useful and we were wondering, oh, man, is something go wrong? Then it didn't, you know, we figured it out. So this stuff is science. It's open ended. There's many scientific questions about how to scale these models. For us, we feel like we've stumbled on a new paradigm. The O1 model really changes the way that we think about scaling these models, because you now can just not, you can scale the amount of thinking it does as it responds. Like, for the first time, you can have this thing actually think really, really hard about something, give you a better answer.
And, you know, that's an entirely new paradigm that we're just beginning to explore. We're just beginning to productize and we're just beginning to scale up. So I'm actually very excited about where we are.
So is the basic idea that even if this one technique that had been the main technique for making large language models more capable over the past few years starts to slow in terms of the rate of improvement, you also have developed many other techniques for getting more out of these models that you are actually just starting to explore?
That's right. We're really at the beginning of the reasoning era, not the end. And there's so much to explore there.
On the reasoning point, can you give some ideas for, let's say, two writers for what they should actually be doing with this model? Because I have not been able to figure out anything to get the, to have the reasoning model do that I can't already get from the normal model. Is this, is it just none of my business, Nick?
No, it's everyone's business.
It's everyone's business. You know, this gets back to, you know, the same problem that we had when we first launched ChatGPT, where nobody knows what it's really capable of and we have to find out together. I say this in all earnestness, you know, we do so much before we launch these models now. You know, we red team them, we show it to experts. It's becoming a exploratory exercise for many of us to figure out what this stuff is capable of. For me, I found it to be really, really useful at things. And there's not very many of these things where I know more than most people. You know, a generalist, but I happen to know more people, more than some about, you know, jazz. So I had it, you know, reharmonize a song, which, you know, I'm
I won't bore you with what that is, but it's like something that I happen to be really interested in. And I realized it is much, much smarter than the average model.
Well, tell us a little bit about that because I'd like to hear what chat GPT actually does when you say, hey, reharmonize this.
Yeah, I give it the chords. I'm like, okay, what are other ways to express these chords that would sound more funky, more like modern jazz? That still work and like resolve at the end. So I'm having it, you know, kind of change the chord structure of the song and then I sit down at the piano and I try it out. So that's like one, you know, it's an area where I have a niche interest where I probably, you know, better than most humans at it. And I think the challenge with smarter than human models in many cases is that you got to find an area where, you know, you can actually assess whether or not the model is, you know, just equally good, worse, or really, really smart.
I have a lot of expertise in giving harsh feedback to Kevin. And so what you have me thinking about is could I actually use it to give even harsher feedback?
Well, I'll get ready for that.
Yeah, I'm gonna, I'm gonna use O1 to process your feedback and put it into a format that I can more easily accept without my ego shattering.
Sounds perfect.
So we can just have like a sort of battle of the AIs there. Nick, another question we get all the time from listeners and readers is about the environmental impact of chat GPT. This is a big question. There's some people who believe that this is, you know, you're pouring out a bottle of water every time you ask chat GPT a question or something like that. So can you like settle this for us? How bad is chat GPT for the environment?
I don't know. Very much not an expert on this field. The thing I do know is that, you know, the cost keeps coming down, the efficiency keeps going up. You know, we wouldn't be, if that hadn't happened, we wouldn't have been able to serve all the users that we have today. You know, you probably remember the time when chat GPT was down all the time. It was a very unique era for all of us, including the team. So it's gotten a lot more efficient and it will continue to do so, but I don't know the exact economics. We'd have to ask someone.
Okay, we'd like to close, Nick, with a game. We, I'm sure you've been seeing all of the people posting on LinkedIn and X and threads about their insane ways that you're, you're not getting enough out of chat GPT. And with these hacks, you could just become the ultimate chat GPT power user. This has become sort of a meme among the hustle bros of America. Nick, as the head of product for chat GPT, what are your top insane chat GPT hacks?
You could turn this into the most viral LinkedIn post ever. When we plan to, we're going to, yeah. Yes. You know, few things, really experiment with voice. It is a completely different way of using chat GPT. It's never existed before. It's like unlike anything else you will have tried in technology. Takes a little bit getting used to. The voice mode specifically. Yes, but yeah, put your AirPods in, go on a walk. You know, try to use in the car. I really recommend trying this out. Second, leverage memory. You can tell it to remember stuff. You don't have to wait for it to infer it about you. You can just tell it all the stuff that you want remembered. Can be your job, it can be your family situation. It can be your preferences on how it should respond to you. It can be, you know, your favorite food, stuff that you wanted to know.
The effort will be worth it because memory is just going to keep getting better. And it's already there in many cases. Third, if that's not enough, you can make a GPT. People actually do this for a variety of reasons. But in many cases, if you have this prompt you're reusing all the time, make a GPT for yourself. You don't even have to share it with others. It's super easy, takes seconds and it allows you to customize everything that you want about your chat GPT. Four, try uploading stuff, uploading files. It's pretty phenomenal. If you've got a really long, you know, paper to read, if you've got, you know, a handbook you're trying to process, et cetera. A lot of people don't know this exists and it's really, really powerful. And then fifth, images in, images out.
I find that there's text people and there's image people and they almost don't overlap. Where, you know, some people in my family think, you know, the text is the coolest thing about chat GPT. And some think that the image generation is the coolest thing. And I wish there was a bit more cross pollination between people because there's a lot of cool stuff. You're working on a birthday card, probably going to want to use both. Make a little custom image and work on a cute little poem. So you can, you can combine all these things in a way that I think most people don't understand. So those are a few. I don't think any of those are rocket science. I'm learning as much as you all are from observing what people do in the wild. So I always click on these posts. They get me every time and sometimes I do learn something new. All right, well, Nick, thanks so much for coming by. Thanks, Nick. Thanks for having me. Thanks. Bye.
When we come back, we swoop into action to help a listener.
Support for this podcast comes from Dell Technologies. It's time to put AI to work. Is your business ready? The challenge isn't getting started with AI. It's scaling beyond pilots while maintaining control of your data, governance, infrastructure, and costs. Dell AI Factory provides the integrated foundation you need to scale AI securely and efficiently with one trusted partner. Dell, AI on your terms. From idea to impact. Learn more at dell.com/yourwaytoAI. That's dell.com/yourwaytoAI.
Companies are pouring money into AI, but few leaders can explain how those AI investments work with the systems you already have. You know, the ones that actually run your business? Without Systems, you don't have to start over with AI. You can choose the projects where AI will have the most impact. And OutSystems gives you the platform to make it agentic on your terms with complete control every step of the way. So you can trust AI with the mission critical systems that run your business. Own your agentic future. Learn more at outsystems.com/NYT. That's outsystems.com/NYT.
AI spend is growing fast, but most businesses still can't see exactly where that money is going. Ramp gives businesses visibility and control across AI providers with realtime guardrails by team and model, helping ensure spend goes to the highest ROI areas. Stop guessing where your AI spend is going. Get started with Ramp. And if you're new to Ramp, you can get $250 off at ramp.com/hardfork. Go to ramp.com/hardfork. Ramp.com/hardfork.
Well, Kevin, when you and I had the idea for this show, we had one simple idea, which was to create parasocial relationships with hundreds of thousands of people and then monetize those on behalf of the New York Times Corporation. Is that what we decided? But somewhere along the way, we had a second idea, which is what if we could help people? And recently, an email came into our servers here at the Hard Fork operation, and I thought this might actually be a chance to do some good.
And it came to us from someone named Winston from Oakland, California nearby. And he writes, "Hi Kevin and Casey, big fan of the show. I'm wondering which social network I should pay attention to. I want to be engaged in the cultural discourse, but not constantly fed attention-grabbing clickbait. What social networks are you guys on? And what do you recommend for someone who wants to have a thoughtful, engaging, and positive relationship with social media? Thanks." And I guess the first thing to say is, that's not possible. But we do think that there are some interesting places on social media, places where you can learn some things, maybe have some good conversations. My expectation is, they will probably stress you out,
and you will probably want to put them away for long periods of time. But if it's been a long time since you've rethought your relationship with social media, Kevin, it does feel like in the aftermath of this election, we're seeing a bit of realignment.
Yeah, I'm getting a lot of texts and emails from people in my life basically asking the same question Winston is asking. I think people have this sense that the places they're spending their time right now, especially X, as it continues to change under Elon Musk, are maybe not the best places where they could be spending their time. But it's a big deal to invest time and energy into building a presence on a new app or a new social network. And so I think people are just a little bit confused, and maybe we can offer them a path through the wilderness.
I think we can. Look, you and I both really enjoy using social networks. Both you and I have written a lot about them over the years, and we have been taking a close look at everything that is happening in that.
I want to correct you there. I don't, I would not say that I enjoy using social networks. I say there are periods where I have good experiences, and then there are periods where I do it out of a sense of dutiful obligation and self-preservation.
And I actually do enjoy using them because I'm a fiend.
Well, good for you.
So with that in mind, let us take a stroll through the current state of the art. And I think we should focus in particular on the Text-based social platforms because this is where we're seeing the most change to the social media landscape. These are the platforms that people might be least familiar with and that are still changing a lot day to day as they build out their user bases and figure out their cultural footing. So here's how we're gonna do this. We're going to tell you a little bit about the network, who's on it, what the vibe is there, and some reasons why you might want to check it out.
Yeah, we're not gonna tell you which ones you should be on because that's a personal decision, but we'll give you sort of a lay of the land and you can make your own decisions.
All right, well, let's dive right into it. And let's maybe start with threads, Kevin. So Adam Mosseri, who runs both Instagram and Threads, announced earlier this month that Threads had crossed 275 million monthly active users. That is up from 200 million in August. And Mark Zuckerberg says they're signing up a million people a day. So this is a place where frankly, I have been spending most of my time in social networking, and I would love to tell you a couple of things about it. Tell us about Threads. So let's talk about who is there. You will find stars like Simone Biles and Sarah Jessica Parker.
Sarah Jessica Parker?
Did I say Sarah Jessica Parker?
You sure did.
I like that so much. That's what I'm gonna call her now.
Yes. So you'll find stars like Simone Biles and Sarah Jessica Parker. You'll find a lot of folks who have big followings on Instagram who want to give more text-based social networking a shot to the sort of creators and influencers. You will find a lot of reporters there. It's kind of how I follow the tech news day to day and see what the conversations are about it. And you'll also see a huge international community from countries including Brazil, Indonesia, and Vietnam. But that's who's on Threads.
Yeah. And what would you say the vibe of Threads is for someone who has not spent a lot of time there?
So I think it is candidly still trying to find its vibe. Friend of the podcast, Max Reed, has described it as the gas leak social network, which is sort of referring to the fact that if there is a gas leak in your house, you might sort of start hallucinating or behaving strangely. And this comes from the fact that some people, when they log onto Threads, just find themselves confronted with a lot of people who are just telling what seem to be pretty lengthy, random stories about their lives, having various problems. So it's essentially not clear what people are supposed to be doing there. I find reporters sharing articles and discussing current events.
But many reporters will tell you that the stories that they share there have their reach throttled because the Threads recommendation algorithms consider them too political and restrict their reach. So it's kind of a melange of things. But I will say that if you are somebody who just wants to use something that is a lot like Twitter to get your tech news, I do find that Threads works very well for me.
Yeah, I would describe Threads as a great social network for you if you want to learn who won the election 17 hours late. Like, it is so much like what Twitter used to be in terms of the features and the short length of the messages and, and sort of the general appearance of the platform. But it is chosen intentionally, and Meta has chosen intentionally, to de-emphasize news, especially breaking news on the platform. And so as a result, you just see a lot of times, things that are old, that are stale. It's sort of missing the, the sort of beating heart of current events that used to be at the core of what old Twitter was.
Yeah, particularly if you are using Threads on your phone, it defaults you to this "For You" page, which you can think of as a kind of TikTok for text. The Threads algorithm goes out, it tries to pick posts it thinks you might be interested in. And as you just mentioned, Kevin, it is not always selecting for the most recent things. And so, yes, one of the main criticisms of Threads over the past year has been, "The stuff here feels really old." I will say, if that's a problem for you, try visiting it on the desktop. You can set up different columns for Threads, including just a feed of the people you are following, and that feed will update in real time, showing you all of the latest things that people have posted. So that is the primary way that I use Threads, and that's the way that I get around the recency problem.
Yeah, and I am less optimistic about threads because I think that the company that is running threads, Meta, is just not at all aware of what actually made the old Twitter so good, which was that it did have this kind of beating heart of being able to figure out what was happening in the world before it appeared anywhere else. And news is not something that they have emphasized. In fact, it's something that they've sort of walked away from. They don't want politics on there, whatever that means. And I think that that is a fatal flaw for any app that is attempting to educate people and show them what is happening in the world around them.
All right. Well, for an app that has a fatal flaw, it seems interesting that 275 million people are using it every month.
Okay. How many of those people just clicked on something by accident while scrolling Instagram? That's what I want to know.
If they were on your profile, they definitely clicked on it by accident. Let's talk about Blue Sky. So, Blue Sky is another decentralized social network. We have had the CEO of Blue Sky, Jay Graber, on the show. And Blue Sky has had a real moment after the election. It confirmed that it gained more than 1 million new users in the past week and now has 15 million users globally. So, who's there? I would say you could find leftists, resistance libs, artists, sex workers, journalists, and, I'll say it, some of the most annoying people who ever had Twitter accounts. You know those people who, like, no matter what Trump had done, they would immediately have like a 70-part thread about it on Twitter?
All those people have Blue Sky accounts now.
Yeah. I mean, I have not spent that much time on Blue Sky since setting up an account there. I think I'm gonna start again, but, you know, one problem for any social network is you get there, you set up an account, and it's like, okay, well, who do I follow? And how do I make sure that I'm following all the right people? And Blue Sky has this new thing, Starter Packs, which is basically like, you can have a group of people that are all sort of connected thematically. Maybe you have a politician's Starter Pack, or a tech journalist Starter Pack, or an artist Starter Pack. And those are sort of a bunch of accounts that you can just follow all at once by clicking one button.
Yeah, it makes it easy to set up your network. So, what's the vibe? I would say, if you want to learn about current events entirely through people criticizing how the New York Times covered them, Blue Sky can
This is really a theme of a lot of the social networks today.
I would say the vibe is extremely horny art and not safe for work content. The furries are having a big moment over on Blue Sky.
Are they?
Yeah, and shitposting in general is just kind of a core value of the Blue Sky community.
Yep.
Do you know I've been canceled on Blue Sky twice?
For what?
Once I criticized liberals for spreading misinformation about JD Vance by saying he'd borked a couch. And two, I had a take about Elon Musk and Brazil that was not received well in Brazil.
Wow.
Now, the advantage of that was that most of the insults that came to me were in Portuguese, which is a language I don't understand. But let's just say I understood enough to know that I'm not welcome in certain parts of Rio de Janeiro. Now, some other interesting things about Blue Sky. It's also built on a decentralized protocol, but it is a different protocol than Threads and Mastodon. Theirs is called AT. And I have to say, so far, it seems like it is enabling a lot more creativity than Activity Pub. Recently, you mentioned the starter packs, which are cool. It also has a way for you to bring in your own recommendation algorithm. So if you don't like the kind of like the default feed, you can bring in your own. If you don't like-
What does that mean? Like, what's an algorithm that I could bring in on Blue Sky?
You could say, "I want you to show me more content about science." And the algorithm will work to find you more content in that vein. So you know, like on Threads, you're limited to the feed of people you follow and the feed of whatever Threads thinks you might want to look at. On Blue Sky, you can say, "I want a feed of just all the most popular Blue Sky posts from over the past day." Or, "I want a feed of posts that have been popular with the people I follow on Blue Sky." And you can get much more sort of granular and weird. And so of all of these networks, Blue Sky is the most customizable. Now, that also means that it kind of has the steepest learning curve. You don't need to do any of that stuff to use Blue Sky, but I would argue that to get the best experience, you do.
But it speaks to the creativity that is possible because the way they have built their network, and I think it is super cool.
And my understanding is that Blue Sky, which had this sort of invitation system early in its life, is now open to anyone. You can just go create- An account, right? You don't have to get an invite from someone?
Yes, absolutely. And something that the reporters are loving about it in particular is that unlike X and Threads, there is just truly no restriction on sharing news. If you're a reporter and you're in a story and it's about politics and you post it, no one at Blue Sky is going to try to reduce the reach of that post or make it take longer for the webpage to load like Elon Musk does.
Yeah, it seems like the only sort of major text-based social network that does not actively hate the blue link.
And is trying to like, remove that from the server.
Yeah, and of course, that is a core way that people use Twitter was to see what was happening in the news. And Blue Sky is quite happy to let that happen. Threads has been much more skeptical of that. So that is a core difference between these two. And if really all you are looking for, Winston, is what network should I use to replace Twitter? This is, I think, the question you have to ask yourself is, do I want a fire hose of news? Or do I want something that is going to be much more cautious about what news it shows me? So, who should check out Blue Sky? Well, I would say, if you hate capitalism, you're gonna love Blue Sky. Comes up a lot over there. If you love the news, you're gonna love Blue Sky. If you are a sort of protocol nerd who just likes to build stuff, or if you're horny,
I think you could have a great experience browsing the Blue Sky app, Kevin.
Well, I don't know any horny socialists, but if I do meet some, I'll direct them over there.
Let's move on and talk briefly about Mastodon. Mastodon is a smaller network, but in a way, kicked off this current realignment that we're seeing. Because while it was created several years ago, it has really ramped up after Twitter collapsed and was replaced with X. It had three and a half million users in November 2022, and it has almost nine million this month. So, who is there? I would say IT administrators, college professors who study arcane subjects, and bots. Those are sort of three of the big categories I've seen. Kevin, how about you?
I haven't spent much time on Mastodon, but in the times that I have gone over there, it's been pretty clear. This is a network for people who love being yelled at by Linux administrators.
Yes. If you ask me to describe the vibe of Mastodon, I would say, imagine a place where every reply starts with, actually.
And that's kind of Mastodon in a nutshell. So I would check out Mastodon if you want to sort of host your own federated server and send pedantic replies to well-meaning people who were just trying to share an opinion about something.
Now, I should say, I want to defend these social networks from some of the slander being perpetrated against them on this podcast. I think all of these networks have pockets of good in them, right? Like, it is possible to tweak your feed, to tweak your following list, to really spend the time setting yourself up on these networks so that you're seeing more of what you want and less of what you don't want. But that is true of any social network.
You're saying that my experience is not representative of all other people's experience?
That is what I'm saying.
That's highly offensive to me. Now, let's talk about the one other text-based network that some people might be considering, Kevin, and that is, of course, X, formerly Twitter.
Yeah. So who's there?
Still the biggest, by the way.
Well, there are a few folks. For example, Elon Musk, people who love Elon Musk, people who worship Elon Musk, and people who work for Elon Musk or hope to work for Elon Musk someday. And in addition to that, there are some reporters.
There's a lot more happening on X than this.
What else is happening?
So, for example, all of AI Twitter or AI X is still there. It's still a very robust and popular sub-community. There are also lots of other groups of hobbyists and people working in specialized fields who are on X. So I know that you are grinding an X here, and that's fine.
I'm grinding an X here.
But I also want to say, like, one of the narratives that a lot of people, I think, have been wishfully hoping would come true is that X is a dying ghost town. And I think we have seen just how untrue that is. It is, it may be changing, it may be morphing. You certainly hear, see more right-wing political content on there than you would have in the old days of Twitter. But I think this, this sort of wishful thinking among especially many liberals that X is dying off or would no longer be relevant very soon, we have just not seen a lot of evidence for that.
Well, I think it is dying. I don't think it's a ghost town yet, but I think it is well on its way. So What is the vibe on X these days? I would say it's kind of gloating about the election results. So if you want to read a bunch of posts by people who are excited about everything that is about to happen in America, this is a great time to check out X.
Okay, we get it.
Well, no, I have, I have one more question for you, Casey, which is like, there's been a lot of speculation about what X will become during the Trump administration, whether it becomes the sort of unofficial media arm of the Trump government, whether it draws back toward the center now that Elon Musk has accomplished his goal of getting Donald Trump elected. Like, what do you think the future of X looks like under the Trump administration?
You know, I read a great post that sort of predicted, made a prediction about this on Blue Sky today, Kevin, and it reads like this. This is from Lauren at rotatingsandwiches.com. Another cool thing about Blue Sky, you can bring your own custom domain into it. So anyway, Lauren says, "One day, you're going to see Elon posting, 'We just cut $200 million going to bull weevil outbreak prevention.'" Two cry laugh emojis. And then a couple of months later, the entire Midwest will be decimated by the first mass bull weevil outbreak since 1884. So that's what I think you can expect from X. There's going to be a lot of pronouncements from Elon about how he and his cronies are reshaping the federal government. I think it's going to feel like state media. It's going to be a sort of 24/7 Trump rally.
And yeah, that's kind of what I expect. What do you think?
So I think that's certainly possible, but I do wonder if liberals, if people who are opposed to Elon Musk have made a mistake by migrating off of X. Because some of the reason that you might be seeing more right wing content on X is because the knobs are being tuned in such a way that promotes that content artificially. But some of it is just that a lot of people who are creating other types of content have left the platform. And so now, if you are a person who forms your view of the world, or your view of politics, or your view of current events, by seeing what's going on on X, you might get an artificially sort of right wing seeming view, because that is sort of the user base that is most active there. And I'm
Starting to wonder if maybe it was a mistake strategically for Democrats to sort of stop investing in that platform and start going elsewhere. What do you think?
I think that that is a total trap. I think, you know, if you think that the elections outcome would have been different if liberals had just stayed and fought on Twitter, I truly do not believe that is the case. I think that so much of the power that X continues to have is the folks who are telling themselves, oh, no, I am there. I am fighting the good fight. But every day you look around and you have fewer and fewer allies around you and you realize the people that you are shouting at, you are not actually persuading them. What you are is you're providing a fig leaf that lets other people justify staying there because they're like, well, no, you know, there's still like a few of these other accounts. But in reality, this thing has just become a one-way partisan broadcast service.
Well, I think we've gotten a little far away from the the prompt here, which was about what social networks people should be on. But do you think there's anyone who should be on X, Casey?
Yeah, Trump voters.
No liberals?
No.
No, no journalists?
No policymakers?
What about people who just have a sort of an interest in knowing what the people with Donald Trump's ear are thinking and saying?
Yeah, that's actually why I still have an account, just to like login and see. If you want to see what other people are saying, that's fine. I like, I don't know what your expected value would be from participating in Twitter at this moment in 2024.
So Winston, we hope that helped you think through.
No, that absolutely did not help Winston.
Why was that not helpful?
We just gave Winston a laundry list of the pros and cons and jokes about various social media apps. But I want to actually try in earnest to answer the question that we received from our dear listener here, which is like, you know, without knowing more about Winston and what he uses social media for and is trying to get out of a social media experience, it is hard to say with any precision what which social network or social networks Winston should be using. But say that you are just an average consumer of social media. Maybe you want to see some jokes. Maybe you want to learn a little bit more about a topic you're interested in. Maybe you want to hear what people in your profession are doing. Maybe you just want to catch up on a little bit of news. Sort of the all-purpose social media use case.
What in today, in 2024, what is the the place, the social network, the app, that gives people the best chance of getting all of that in one place?
Well, here's the unsatisfying answer. I don't know that you can consistently find all of that in one place. I think the good stuff is scattered across every single network that we just talked about. And if you want to find all of the good stuff, you're going to have to spend a lot of time browsing. If somebody came to me and they said, I have a little bit of time for social media, where should I spend it? I think the answer is that most people, if you just sort of look at like how people are voting with their feet, they spend most of their time on TikTok and Instagram. Okay, so that would actually be my default suggestion. Most people don't want to post. Most people don't want to fight with strangers in the comments. Most people just want a little bit of entertainment while they are, you know, in at the dentist's office waiting for the dentist to, you know, come get them. So that's sort of what I would say. But I
think the hard fork listener Kevin is a little bit more sophisticated and a little bit more refined. I think they are very interested in the future. They want to see the debates that people are having about AI, about what stories are in the news right now. What is the conversation around those stories right now? And if you are one of those kinds of people, I would say, get on threads, get on Blue Sky and see which one you like better. You will probably wind up having a pretty decent experience either way.
Yeah, I just, I'm hungry for like something that just, right now, I feel like I spend about 10% of my day just like shuttling between various apps. And the like, the thing that I actually want is the is the one app to unite them all, right? The thing that will sort of use AI or something to sort of pick out the best 5% of posts from threads, from Blue Sky, from X, from Instagram and show them all to me on one beautiful interface. And how likely, Casey, would you say it is that something like that will be developed?
Well, this actually is one of the cool things about these decentralized federated networks is that you can download a Mastodon client and use it to browse threads. At the moment, the Activity Pub protocol that Mastodon is based on is not compatible with the AT protocol that Blue Sky runs on. But people are talking about potential Building a bridge that would let those two protocols speak together. And I think whatever else you think about, you know, this state of social media, to me, this remains a very exciting thing where in the future it will not be one giant monolithic company in control of a social network. It looks like it very well may be different social networks talking to each other over protocols.
And if and when that comes to pass, you will be able to read Threads and Blue Sky in the same place, and that could actually be pretty cool. Yes. Yeah. So Winston, we hope that helped. And if it didn't, well, it was free advice. Get what you paid for. You get what you paid for. Yep.
Companies are pouring money into AI, but few leaders can explain how those AI investments work with the systems you already have. You know, the ones that actually run your business? Without Systems, you don't have to start over with AI. You can choose the projects where AI will have the most impact. And OutSystems gives you the platform to make it agentic on your terms with complete control every step of the way. So you can trust AI with the mission-critical systems that run your business. Own your agentic future. Learn more at outsystems.com/NYT. That's outsystems.com/NYT.
AI spend is growing fast, but most businesses still can't see exactly where that money is going. Ramp gives businesses visibility and control across AI providers with real-time guardrails by team and model, helping ensure spend goes to the highest ROI areas. Stop guessing where your AI spend is going. Get started with Ramp. And if you're new to Ramp, you can get $250 off at ramp.com/hardfork. Go to ramp.com/hardfork. Ramp.com/hardfork.
Before we go, we have a special request. We want to learn more about you, our listeners, what you like, what you don't like. It will help us make a better show.
But don't just start talking to the radio right now. We won't hear it that way.
Yes. If you can, we would really appreciate if you filled out a quick survey. You can find the survey at nytimes.com/hardforksurvey. Your answers will not be published in any way. They will just sort of help us make the best show we possibly can and understand more about who listens to this show in the first place.
We want to know your most private desires. And what will we keep it to ourselves? I don't know.
Okay. Send your most private desires to Casey's email directly. But if you want to fill out our survey, again, you can find the survey at nytimes.com/hardforksurvey. We'll also drop the link in show notes.
Hard Fork is produced by Whitney Jones and Rachel Cohn. We're edited by Jen Poyant. We're fact-checked by Caitlin Love. Today's show was engineered by Alyssa Moxley. Original music by Diane Wong, Rowan Nemezto, and Dan Powell. Our audience editor is Nell Galogly. Video production by Ryan Manning and Chris Schott. You can watch this full episode on YouTube at youtube.com/hardfork. Special thanks to Paula Schuman, Hui Ying Tam, Dahlia Haddad, and Jeffrey Miranda. As always, you can email us at [email protected]. Send us your best ChatGPT hacks.
You can't undo everything in life, but you can undo prediabetes. More than 2 in 5 adults in the U.S. have prediabetes. Most don't know they have it. But with early diagnosis and action, you can delay or even prevent type 2 diabetes. Visit doihaveprediabetes.org to take the 1-minute risk test and start undoing the things you can. Brought to you by the Ad Council and the Centers for Disease Control and Prevention.
Page updated October 12, 2026 · transcript and summary are machine-generated
