Google 搜索要凉了？+ 美国要建立加密货币储备？+ 你正在“凭感觉编程”什么

Well, Casey, it finally happened to me this week. I was radicalized by Elon Musk.

Radicalized in what way?

Well, not into shifting my politics.

Okay.

But I don't know if you saw, there was a big story in The New York Times a couple days ago that included, in its many revelations, a picture of Elon Musk's desk in the Eisenhower Executive Office Building next to the White House.

I have not seen this picture.

And I'll show it to you. Do you know what is on this desk? It is a DOGE sign, a Make America Great Again hat, what appears to be a big gaming PC, and then in the middle of the desk, a very austere setup, is a huge, wide, curved computer monitor.

Mm-hmm.

And I saw that photo, and I thought, "You know what? I want one of those."

Because you thought, Elon Musk, whatever you think of him, he does clearly get a lot done, and so maybe if you had a big, curvy monitor like he does, maybe you'd also get a lot done.

Well, yeah, I thought these were basically for gamers. I had never seen one in an actual office where people work.

Kevin, you have to come to my office. My monitor has curves and swerves like you wouldn't believe. It is so big and curvy. You would love my curvy monitor.

Really? See, I'm a flat-monitor guy, or at least I have been.

No.

But this week I decided to take the plunge, and so I went onto Facebook Marketplace and got myself a 2-year-old Dell mega-monitor that is curved. I didn't go for the biggest one that is a full 180-degree wraparound, full-IMAX experience, but I do now have a very large curved monitor, and I have to say, I love it.

How has it changed the way you use your computer?

I'm looking at so many emails. I am looking at, and not responding to, so many emails.

This week, Google is bringing AI to Search in a brand-new way. How will it change the internet? Then, The Times' David Yaffe-Bellany joins us to discuss the idea of a strategic crypto reserve, and why even some crypto supporters think it might be a bad idea. And finally, we asked you to share your experiments in vibe coding. Today, we're going to find out what you made.

Well, Kevin, every once in a while we get a preview of a very big change to the internet, and I think we might have gotten one of those this week.

Tell me about it.

1. Google Introduces AI Mode

On Wednesday, Google said that in addition to expanding AI Overviews to more users and upgrading the underlying model to Gemini 2.0, it is also introducing a brand-new search mode called AI Mode, and it seems likely to me that this may be the future of Google.

Yeah. This is, I would say, at a minimum, a much better name than we have seen for any AI feature that Google has released over the past couple of years. AI Mode: it's a mode, it does AI. Seems pretty straightforward to me.

Yeah. And where other companies might have called this X-73-Mini, Google just picked a normal name for it.

Yes.

Which is something that Google does a lot.

That's true.

So right now they're calling this an early experiment. It is rolling out now to paid subscribers to Google One's AI Premium subscription service. And even if you pay, you also have to opt into it through something called Google Labs, which is a feature that Google has to let you opt in to trying new stuff. So it's kind of hidden away. We have asked if we could use it right away, and they said no. So we don't have a direct report for you today, but what we can tell you is that AI Mode is Google's answer to bringing something like ChatGPT directly into Search, right?

So far, it has been dabbling. It's been sprinkling these little AI Overviews on top of the traditional 10 blue links. I'm sure you've all seen those. But while right now AI Mode is separate from Search, if it succeeds, I suspect it will gradually merge into the main Google Search results page. And that matters more than you might think, Kevin.

2. The Search Dilemma

Yeah. Before we get into AI Mode, let's set the scene here. Google: big company, most important search engine in the world. Ninety percent of search queries run through Google. It's used by billions and billions of people every day, and it has been an indispensable part of the architecture of the internet for the last several decades.

AI, large language models, ChatGPT comes along, and all of a sudden people, including us, think, "Why am I going to ask Google this question when I could ask it to an AI chatbot that might give me a better answer without requiring me to click on a bunch of links and navigate through a bunch of ads? I'll just get the answer I want right there in the LLM chat box."

Exactly.

So this is obviously what kicks off this process at Google of saying, "Well, wait a minute. We have this very profitable search engine. We want to keep it profitable and make sure that people aren't using ChatGPT or Perplexity or some other product instead. So we want to bring AI into our search product."

But that is complicated because, among other reasons, these AI Overviews that you just described got some things famously wrong, including telling people to put rocks on their pizza. Or was it glue?

It was telling them to eat rocks and to put glue on their pizza. Two different answers.

Okay.

Yeah.

So they've spent some time retooling this, but they have not backed down on their ultimate belief that people will use AI for search-related tasks, and that because some number of people will want to do that, or are already doing that, they need to build that into the core Google search engine.

That's right. And so, Kevin, this is really a story about Google being caught in between two imperatives. On one hand, it has to change with the times; it has to keep pace with the competition. On the other hand, it can't break its core business, which generated $54 billion last quarter. So that is a very tricky balance. How is it going to navigate that? AI Mode is the best evidence we have yet of how the company is thinking about it, so let's talk about it.

Okay, let's talk about AI Mode. What is it?

AI Mode is a new mode within Google Search where, if you ask a query that might be a little bit better suited to a chatbot, you can click over, and it is going to give you a slightly more chatbot-like experience. But it is different from something like the Gemini app, which is Google's straight-up ChatGPT competitor, in a couple of pretty interesting ways.

There are very prominent links to websites within AI Mode in what a user interface designer would call a carousel—you know, those little horizontally scrolling bars of links. And so as you move through the answer to your query in AI Mode, Google is going to say, "Here are some actual websites you could visit to get more of that information."

And you just said this is rolling out to a small subset of paying Google users who have subscribed to this Google One AI plan. Presumably, this is something that they're interested in building into the Google experience for everyone eventually, right? Or is this going to remain small?

So I talked with Google yesterday and tried to get a sense of that, and the company was loath to make too many predictions here. This really is an experiment, they told me, and if it goes super, super badly, they could presumably pull the plug. But I don't think that is going to happen.

I think Google realizes that this is a once-in-a-generation chance to reinvent the search experience, and that is going to mean a fundamentally different way of presenting information. That's what AI Mode is.

So I haven't tested out AI Mode on Google either. They have not made it available to me, but there are some screenshots floating around. I think maybe we should just describe what this is going to look like for people. They have a screenshot of someone asking the question, "Explain how déjà vu works and how it relates to memory." Now, I get a feeling I've heard that question before.

I thought we talked about that on a previous episode.

Okay, we made the same joke. Good for us. So in AI Mode, instead of showing the user a list of links or an AI Overview, it sort of writes them a little essay. It says, "Déjà vu is that strange and fleeting feeling of having experienced something before, even though you know you haven't. It's like your brain is playing tricks, creating a sense of familiarity with a situation that's brand-new," and it continues on.

And then it has a little thing at the top right under the query that says, "four sites," and you can click to expand, and it gives you little icons showing which sites it's pulling this stuff from. But basically, this is Google's version of what tools like ChatGPT and Perplexity have been doing for a while now, which is sort of replacing the 10 blue links altogether with this more bespoke AI response.

Yeah, and you can see that after each paragraph, there is a little icon of a chain link, which sophisticated Googlers like you and I will recognize as a hyperlink to a website from which Google derives information, and which many other people might just think is a fancy period.

You know what I mean? Those particular user interface elements, I would say, are not going out of their way to entice people to click.

Right, and we talked last year on the show about this idea that Google would do the Googling for you, right? They were very invested in trying to simplify the process of searching for information on the internet using AI, without requiring people to click through any of these blue links. I imagined at that time that they were trying to make this available to everyone, but it had some obstacles, including some of the mistakes that we talked about.

But it is expensive to run a search engine on a large language model this way. These queries require inference from these large models. It is not as cheap, easy, and efficient as just running a regular search engine. So did you ask them about that—how they've dealt with some of the cost concerns?

I didn't ask them about that this week, but they have said over the past several months, effectively, that the costs keep coming down here. They're figuring out more and more efficient ways to serve these queries. For what it's worth, I don't think that the slow rollout of this is primarily a cost issue. I think it's much more related to the fact that when they launched AI Overviews, they got a bunch of egg on their face because they were telling people to eat things that, unlike eggs, cannot be eaten.

Right. Do you feel like this is an acknowledgment from them that they are losing market share to companies like Perplexity, companies that are offering a more AI-native search-like experience, or that large numbers of consumers are already using ChatGPT and other AI tools for things that they previously would have Googled?

Absolutely. There was some interesting analysis that came out in January that showed that over the last 3 months of 2024, for the first time since 2015, Google's market share fell below 90%. Now, 90% market share is still incredible, and Google is just barely below it, but there is increasing evidence that these chatbots are starting to eat into Google's audience.

Some analysts have predicted that ChatGPT will have 1% market share in search by the end of this year. Bing, for what it's worth, which has been around forever, has something like 4% market share. So for ChatGPT to get from 0 to 1% in a little over 2 years is pretty impressive and speaks to why Google feels like it needs to do something.

Yeah. I'm finding that I'm using Google a lot less than I used to. I don't know exactly how much less, but I basically use Google now for what are called navigational queries, where I'm looking for this one train schedule or this one restaurant menu, and I know it's a link that's out there on the web. So I go to Google for that because it's right there in my browser bar, and it's very easy, and I can find the link.

I do not use Google anymore for things like product recommendations or advice about how to fix an appliance in my house or something like that. That is the kind of thing that I will now go to ChatGPT or Claude for, and I imagine that if I'm doing that, there are probably a lot more people out there doing that, too.

There really are. While I don't know that this statistic speaks exactly to what you said, Kevin, last month, The Verge published this survey that, among other things, found that 42% of people find that search engines like Google are becoming less useful.

There are a lot of reasons people don't like the ads in Google. I think, arguably, the quality of websites has probably declined. There's a lot of AI slop out there, right? But I also think part of it is what you just named, which is that we now actually have a superior technology to a problem like, “The faucet in my sink broke, and I want to fix it. I bet something like ChatGPT can just tell me how to do that directly.”

Yeah. So I read the Google blog post about AI Mode that came out this week, and it's sort of interesting as an artifact of cultural anxiety at Google. They clearly know that they have to do something big around AI. Their competitors are doing it. It's eating into their market share. But they also seem a little bit scared of it.

There's this paragraph in there that says, “As with any early-stage AI product, we won't always get it right. For example, while we aim for AI responses in Search to present information objectively based on what's available on the web, it's possible that some responses may unintentionally appear to take on a persona or reflect a particular opinion.” What do you make of that?

Look, Google is terrified of politics. To this day, if you ask questions about politics in Gemini, in my experience, it is more likely to refuse to answer your request than any other AI chatbot that I pay for, right?

They got in a lot of hot water last year, which we've talked about on the show, when, for example, they would not appear to depict exclusively white Founding Fathers when you would ask for that. In the wake of that, they got a ton of criticism, and so they tried to strip as much politics out of their products as they could.

Now they're about to put another thing out there where people are going to be asking it to give them opinions. Depending on what opinions are revealed, those people are then going to screenshot them and put them on social media, and congresspeople are going to see them, and it's going to trigger a whole cycle. So I understand why they are being cautious about this.

But I think it gets a bit silly because, ultimately, people are turning to these things for their opinions, and I think on some level, you want to have a product whose opinions you stand behind.

Yeah. So if you are a paying subscriber and you get access to AI Mode on Google, and you turn it on in your Google Apps feature, is every query that you type into a Google box going to be answered by AI from then on?

No. You are still going to have to opt into this, right? Think about how many hoops Google is making you jump through to do this. They are truly in that sort of beta stage where they just want to gather some data. They want to see whether there are any obvious, horrible problems that we can identify and solve before we release this to the entire user base.

But I can tell you, as soon as I get access to this thing, this probably is going to become my default way of using Google for a lot of different kinds of queries, at least for a while. First, I want to see how good the AI responses are, but also, I'm so interested in this question of whether I click any of the links that are in this mode.

Because if I don't, then it does seem to move us closer to that world where the entire economic foundation of the internet is changing.

3. Publishers Lose Their Traffic

Okay, well, you brought this up, so let's dive into this issue of how this could affect publishers—people who make things and put them on the internet and rely on Google as a source of traffic. What do we know about how AI Mode is being received by the broader internet? Are people freaking out about this like they did with AI Overviews?

There is a lot of nervousness among publishers in particular, right? A lot of people write stories about the news, and then people search for those news queries. In the past, they would click on links to read those stories, and then some of those people would see ads, and the publisher would make money. Some of those people would buy a subscription, and the publisher would make money. It seems like that is now happening less and less.

But Kevin, the impact is going beyond just publishers. There was this fascinating lawsuit that got filed last week by a company called Chegg. Do you know Chegg?

The homework cheating app.

Exactly. So if you've been in college any time in the past 15 years, statistically, you used Chegg to cheat on your homework. It offers a database of more than 100 million answers to test questions, and it has been charging students about $15 a month to use this thing.

Well, then along comes Google and its AI Overviews. Using who knows what methods, but I'm going to assume they did some pretty aggressive scraping of Chegg and other websites, they started to put the answers to all these homework questions directly in the AI Overviews. All of a sudden, there is no need to visit Chegg, and Chegg, Kevin, is now on life support.

This is a company that was riding high during the pandemic. It was valued at $12 billion in 2021, and now its stock is basically a penny stock, and it is exploring strategic alternatives, which is sort of corporate speak for “We're going to have to unload this thing in a fire sale.”

In their lawsuit, Chegg said, “That traffic rightfully belonged to us, and Google destroyed us with AI Overviews.”

“You really disrupted us in the homework cheating business.”

I realize as I'm saying this that Chegg is not a sympathetic company, and no one is going to be sorry to see it go who does not work at Chegg. But I'm telling you, this is a bellwether. Chegg is not going to be the last business to be completely disrupted by AI.

Yeah. So what do we know about how people who use AI tools for search-like tasks actually interact with links? Because every AI company that has rolled out something like this, whether it's Perplexity or Bing or Google, they've all said, “Look, people are still going to see links. Don't you worry, publishers. We're still going to put these little citations on it, or maybe we'll put some links below it, and you might get less traffic, but it'll be higher-quality traffic, and you'll still make a lot of money from it.”

So what do we know about whether those people were right or not?

I think there's basically 2 kinds of queries that you can make.

1 query is an answer where you just need something that is good enough, right? You need to satisfy your curiosity about something, and if it's a little bit wrong by 5% in either direction, you don't really care because you're just looking for the gist of something. I think this is the vast majority of all queries.

There is a 2nd kind of query, though, which is, "I need the actual information because my health, my money, my job depends on it." These are the cases where I think that people are actually clicking on the links. I'm a journalist. I'm constantly Googling for information.

I cannot rely on an AI Overview for 1 thing, because if I'm going to put a statistic or a historical fact into my newsletter, I need to know what the original source is, and I need to know that I trust the original source. If that link is going to The New York Times, The Wall Street Journal, or Bloomberg, I'm going to the original website. I'm scanning to that paragraph. I'm seeing it with my own eyes before I'm going to trust that it's real.

I think that's a really good thing that people should do in a lot of cases, but I'm also not naive. That is a minority use case for specialists. The vast majority of people, Kevin, do not need that level of clarity, and they are not clicking those dang links.

Right. There was a study that came out this week that I found totally fascinating and quite worrying. It was done by a company called Tolbit. They're basically an AI licensing company that sort of works as the go-between between publishers and companies making large language models.

They found in their study that the AI search interfaces they looked at—things that are similar to Google's AI Mode—delivered 91% less click-through traffic than standard Google searches. When they looked at pure AI chatbots, it was even worse. Those generated 96% fewer clicks to links than traditional Google searches. Casey, I'm no mathematician, but that's pretty bad.

It's pretty bad, and I should say I have some questions about the methodology here. Basically, Tolbit took the amount of traffic to its publisher sites that it could trace from known AI scrapers and divided it by the total number of times that those scrapers hit their sites overall.

I talked to Google about this. Google was like, "This is a really bad way of measuring traffic," because Google's crawler is constantly scanning websites. Every time the Google crawler hits a website, that is not an actual Google search. So there's some uncertainty here about what this really means.

But at the same time, Kevin, these numbers are intuitive because you and I both use chatbots a lot, and we know that we're not clicking those links a lot of the time, particularly in cases where we are not using the chatbot to do our jobs.

Yes.

Yeah.

No, and I think it's worth saying that even if these numbers were way off—say it was only half as bad—

Right.

—as TollBit says it is, and these AI search tools only produced about 45% less traffic than the Google searches that preceded them, that would still be a cataclysmic event for much of the media industry.

Totally. Now, Google has said, when I've asked about this, that the AI Overviews—which, again, is not AI Mode, but AI Overviews are sort of the warm-up to this AI Mode—have led people to search more in general, right? So Google says basically, "People like AI Overviews, and they see them, and they search more." That leads to maybe more searches in the aggregate, but the impact on traffic to individual publishers does not seem to have been positive so far.

Yeah. I mean, I want to play devil's advocate here for a little bit because you and I and others have been warning about these AI Overviews and these AI search products for more than a year now. We've been predicting that this would crater the traffic to publications and dramatically change the way that billions of people interact with the internet.

I think it's fair to say that, so far, that has not happened, at least at the scale that maybe you or I thought it might by now. There are some people who might be listening to this and saying, "You guys are panicking over nothing."

One thing that is true that we know is that people's habits change very slowly. There are still millions of people out there using AOL and Yahoo for their email, maybe Hotmail. These very old, what we would say are antiquated, services still have dedicated users because people are just used to going to their browser and typing in the same websites that they've been typing in for years.

So I think there is a case that I can imagine people making here that, yes, these AI features, these AI search engines, are going to appeal to power users in the Bay Area who spend all day looking stuff up online. But actually, they are not an existential threat to the internet or to publishers because most people are so used to going to Google that they are just going to continue to do that, even if a superior option exists.

4. Defaults Decide Google's Future

So that is true, which is why if 1 day AI Mode is no longer a little hidden feature and is just actually the front door to Google Search, then that's the ball game because people will not have to change their entire habits in order to have this experience. It's just going to be the default. We know the power of defaults, right?

So that's why I want to talk about this today, because this could be a preview of what Google is in 2, 3, 5 years, and it's going to be very different from the Google—

Yeah.

—we have today.

How do you search these days? What's your go-to if you have a question? Let's take it out of the realm of work. Say you're just looking for a good restaurant to go to. Where do you go?

For something where it's like finding a local business, to me, that is still an area where Google excels. It's like I need to find an eye doctor. I want to find a restaurant that's nearby. I'm going to go to Google or Google Maps, something like that.

But there's this whole set of queries now that I am turning to chatbots for. I've talked before on the show about how I just have a little keyboard shortcut that I type in, and I can query an AI directly. That's everything from how-to questions to essentially just random trivia.

I'm trying to remember, when did this album come out? Or how old is this celebrity? And crucially, Kevin, a lot of those queries are things that Google still does very well. But there is a convenience to just getting a 1-sentence answer and not seeing a sprawling webpage that's full of 6 different ads and various other widgets that I don't care about.

So it's a mix of things, but chatbot usage when it comes to search is on the increase for me. How about you?

Yeah. I'm using chatbots for, I would say, at least half of what I would previously have used Google for. Now, I am not a traditional user. I am sort of an early adopter of this stuff. I have a whole bookmark folder of AI tools that I open up every morning and start using them for browsing.

So I'm not conventional in that sense, but I do find that for the majority of things that I'm looking for, chatbots tend to give me better answers than Google.

When we come back, we have reserved some time to talk about the strategic crypto reserve.

5. Trump's Crypto Reserve

Well, Casey, there is a lot to say about what is happening in Washington right now. Elon Musk and his DOGE team are fanning out across the federal bureaucracy, slashing and burning wherever they go. But the story that I think we can really shed some useful light on here is about what's been happening with crypto and crypto policy.

Yeah, Kevin. We heard some news over the weekend that President Trump was interested in something called a crypto strategic reserve, and we thought that sounds just silly enough that it could be a Hard Fork segment.

Yes. And then on Thursday night, just as we were finishing up this episode, news broke that President Trump had signed an executive order establishing a strategic Bitcoin reserve. That announcement and the executive order attached to it clarified a few very important things.

1 is that this reserve will be filled with Bitcoin that is already owned by the federal government, that was seized or forfeited as part of criminal or civil asset forfeiture proceedings. The 2nd thing is that the Secretaries of the Treasury and Commerce, according to this executive order, will be authorized to develop what they call budget-neutral strategies for acquiring more Bitcoin. They say that this will not have an additional cost on the American taxpayer.

The 3rd thing that the executive order did is establish a separate U.S. digital asset stockpile, which consists of cryptocurrencies and other digital assets that are not Bitcoin. With respect to this stockpile, the executive order said that the government won't purchase any of these other non-Bitcoin crypto assets, but that if it gets some in the case of a criminal or civil proceeding, it will keep it and hold onto it as part of the stockpile.

We just want to note that that happened and that this conversation you're about to hear took place before the order was official, when a few more things were still up in the air. But I think it's fair to say, a really big deal.

It absolutely is. Those currencies, Kevin, are closely tied to people in the Trump orbit who have investments in those very currencies, and so this seems like a classic case of self-dealing among people in and around the president's orbit.

Yeah. I think there's an element of just shedding light on what is happening in Washington right now with regard to crypto policy that is important for us as journalists to do. But I also am growing more worried about what I see as an information and knowledge gap between the pro-crypto proponents and the anti-crypto skeptics.

Would you say that you have some reservations?

Yes. I have some strategic reservations—

Mm-hmm.

—about this. Back in 2022, I was covering crypto much more attentively than I am now. This was sort of at the height of the last crypto boom.

This is your Pudgy Penguins era.

Exactly. And I wrote this very long article, and I was basically arguing that people, whatever they thought of crypto, needed to understand it because this is something that could become strategically important if, for example, a bunch of pro-crypto partisans and activists were to seize control of some part of the federal government and start making big moves to enrich themselves and the people around them. And the response I got to that article was a lot of people saying, “Shame on you.”

Mm-hmm.

“Shame on you for taking this stuff seriously, for trying to explain it, for trying to understand it.” The mere act of wanting to know about crypto was sort of coded as being in the pocket of the crypto industry.

Yeah. And we're in a moment where just ignoring it is not going to make crypto go away. We know that now. And now that we see these really wild initiatives being proposed, I think it is time once again to see what we can learn and to engage with somebody who can hopefully explain what's going on.

Yes. I think it is time for a lot of crypto skeptics—and I would include myself in that group—to give what's happening in Washington more serious attention, because these are big moves with big consequences, and they may be hard to untangle even after Donald Trump leaves office.

Yeah.

So today, we're bringing in our friend and friend of the pod, David Yaffe-Bellany. He's a crypto reporter at The New York Times, and he has been covering all of the twists and turns of this latest saga over the crypto strategic reserve. Let's bring him in. David Yaffe-Bellany, welcome back to Hard Fork.

Thanks so much for having me.

So let's start with some basic questions about this idea of a strategic United States crypto reserve. When do you start to hear about this, and what is the basic idea?

This was an idea that popped up on the campaign trail last year. President Trump started talking about it, and some influential figures in the crypto Twitter sphere were talking a lot about it. But what it actually meant was pretty obscure then and remains pretty obscure now, because when people say Bitcoin reserve or crypto reserve, they really could be talking about 10 different sorts of ideas.

Right. And I want to understand this idea of a strategic reserve, because my basic impression is that this is not a new idea. Governments have long stockpiled assets like gold or oil or other foreign currencies to basically give themselves some cushion against economic downturns or inflation, or just maybe running out of oil during a foreign conflict. So is this similar to those kinds of strategic reserves? Why do people in crypto think that we need a U.S. crypto reserve? Because crypto can't power heavy machinery. It can't feed people. It's not useful in the way that some of these other assets are.

I think there are 2 key arguments that people in the crypto world are making. The more reserve-like argument is that, at the moment, the reserve currency of the world is the dollar. The world runs on the dollar, but maybe not so far in the future, the world will run on Bitcoin or on cryptocurrencies more broadly, and it would be in the strategic interests of the United States to have a giant stockpile of crypto.

What would it mean, though? Okay, so let's say we're living in fantasy land where the world now runs on Bitcoin. How does it benefit the United States to just own a bunch of Bitcoin?

It's money that you could borrow against. It's money that you could spend to fund all sorts of projects, either foreign or domestic. But you're asking the right question, because it's not totally clear how this would work.

Okay.

The other argument that people in the crypto world make, I think, aligns more with this kind of concept of a sovereign wealth fund, which Trump has also said that he wants to do. And that's the idea that crypto is just a good investment, and that if we bought a bunch of Bitcoin now at about $90,000, then 10 years from now, when it's worth $20 trillion, we'll be set up for success.

I'm curious how this idea of a strategic crypto reserve—

—fits with some of the other stated goals of the Trump administration, which is to cut costs and reduce the size of government, not invest a bunch of money in some new class of asset. The funniest post I saw about this said the following: “Sir, we have saved $12 billion that were spent on condoms for Gaza and transgender comic books in Brazil.” “Excellent. Let's buy some Cardano.” So how are people in the crypto policy world reconciling these 2 views of what the government should be doing?

Some of them just aren't reconciling it. They're just embracing the hypocrisy. And it's not the first time they've done something like that.

But there's also a huge split in the crypto world over whether this should be a Bitcoin-exclusive reserve or whether it should be a reserve featuring a bunch of different cryptocurrencies. On the campaign trail, at the Bitcoin Nashville conference, Trump said that he was going to start a Bitcoin stockpile, and all he said was Bitcoin. When he raised this topic again a few days ago on Truth Social, he named all these other tickers for various cryptocurrencies that the people around him have economic interests in and said, “Oh, it's going to involve all of these, too.”

And a lot of people in the crypto world said, “No, we love Bitcoin. Bitcoin is the most established of these assets. It's the one that has the longest-term potential. So don't fill your government coffers with Cardano.”

6. Crypto Splits Over Altcoins

Well, here is where it starts to feel truly silly to me. When Trump started talking about a Bitcoin strategic reserve, I had actually heard of this idea before. Am I right, David, that El Salvador also built a Bitcoin strategic reserve?

Yeah. Other countries are exploring this, too. There are various states where proposed legislation is pending that would create Bitcoin reserves, but it tends to be focused on Bitcoin.

Right.

Which, to be clear, is by far the most valuable cryptocurrency and the one with the longest track record.

Right. So I had at least heard of the idea before, but then in this post that Trump made on Truth Social on Sunday, he says, “We also want to put Solana, XRP, and Cardano into this strategic reserve.” David, can you give us a little bit of flavor about what Solana, XRP, and Cardano are?

These are 3 examples of what people in the crypto world call altcoins, which are cryptocurrencies that are a little bit outside the mainstream of Bitcoin. But these are big—

They're cooler. They play smaller clubs. They're not your Madison Square Garden cryptocurrencies.

But this isn't like some random coin that got invented a day ago. XRP is run by a company called Ripple, which has been a huge player in the crypto world for a long time and had this really important legal battle with the SEC. Solana is one of the top cryptocurrencies. It's a competitor to Ethereum. It's the platform on which the Trump and Melania meme coins were built, perhaps not coincidentally.

And then Cardano was created by a guy who has been a huge player in crypto for a long time and was involved with the early Ethereum movement. These aren't crazy coins, but it's a little bit weird to talk about putting them in a government stockpile.

Right. It's one thing to say these are altcoins that have been around a while, that they're in fairly wide distribution, and another thing to say that there is strategic value to the United States in accumulating a huge stockpile of these things.

Yes. With a lot of things that Trump says, it's not clear how serious he was because he says things sort of off the cuff. His first tweet named those 3 tickers, and then there was another tweet where he was like, “Oh, also Bitcoin and Ethereum will be in it, too,” as if it were an afterthought. So it's tough to read what he's actually saying.

The other thing that's not clear at all is that he said on the campaign trail he would create a Bitcoin stockpile built on Bitcoin that the government already owns, because we've seized a lot of Bitcoin from criminals over the years. It's in the billions of dollars. Historically, the U.S. has tried to sell that Bitcoin, and so all he said on the campaign trail was, “We're going to draw a box around the Bitcoin we already have and not sell it.” It's very different from saying, “We're going to hit up the open market and buy a ton of Solana.”

Yeah, how would that actually work in practice? Because I understand that the United States does have this Bitcoin that it's seized from people who commit various crimes and is presumably keeping in some secure wallet somewhere.

But to acquire new crypto assets, would they just go onto Coinbase and buy a bunch and keep it in a vault in the Treasury building? Or how would we actually go about this as a country?

Well, when you buy a big tranche of cryptocurrency as a huge institution, you're usually doing it as what's called an over-the-counter trade, where you're dealing directly with another person that you're buying it from. The idea is to try to limit the impact that a trade like that can have on the open market. So it's not like the Treasury secretary will log on to his Coinbase account and just start hitting the buy button.

But how would those purchases be funded? None of this has been fully explained, and obviously any government spending raises the possibility of a tax increase or spending some of these perhaps imaginary DOGE savings that we've supposedly been accumulating, so it's all really confusing.

Yeah. How are people in the crypto community responding to this idea of a strategic crypto reserve?

The response has been, I think, fairly negative from a lot of people who you might have expected to be supportive of it, and that's because there's a feeling in the crypto world that, “Wow, this might actually be a little bit of a conflict of interest.” It's like a bunch of people are discovering that that's a problem in the Trump political universe.

People are noticing that Brad Garlinghouse, the chief executive of Ripple, spent a lot of time at Mar-a-Lago before the inauguration and was a huge donor to Fairshake, the big crypto PAC, and suddenly the signature cryptocurrency used by Ripple, XRP, is going to be in the crypto reserve. And so people have become suspicious of those sorts of behind-the-scenes maneuverings. There's also a lot of loyalty to Bitcoin in the crypto world still, and a sense that we should be treating Bitcoin differently from this other stuff.

It's not all that different from if Mark Zuckerberg spent a lot of time with Trump at Mar-a-Lago and then President Trump announced that we were going to create a strategic Facebook stock reserve—of just shares of Facebook, right? I don't really think there's much difference between those 2 things.

Yeah, I mean, is that what is likely to happen if and when this strategic crypto reserve is created: that anything that is in the basket of stuff that we stockpile as a country will become more valuable just by virtue of the fact that we are stockpiling it as the United States government?

Yeah, I mean, this stuff became more valuable by virtue of the fact that Trump tweeted about it, right? The prices of these assets immediately went up, though I think some of those gains have been pared back since then. But yes, there's a concrete market effect. If someone starts buying a lot of an asset, the price will go up, so that will happen. But the symbolic statement of the U.S. putting its government seal on Cardano would be big for anyone who currently holds Cardano.

Right, and I was surprised to see some of the people in the crypto industry coming out against this proposal. Brian Armstrong, the co-founder and CEO of Coinbase, made a post basically saying that just having Bitcoin in the strategic reserve “would probably be the best option.” He didn't want all these other coins included. Tyler Winklevoss, 1 of the biggest Bitcoin investors in the world, also said that he doesn't think other crypto coins should be included in the strategic reserve.

There's a really interesting article by Nick Carter, who is a crypto analyst, trader, and voice on these issues.

Wasn't Nick Carter in the Backstreet Boys?

Different Nick Carter.

Okay.

Yes. But he is a stalwart defender of crypto, and he had an article called “Eight Reasons a Strategic Crypto Reserve Is a Bad Idea.” I just want to read you a line from this because I think it illustrates where some of these people are coming from. He writes, “A crypto reserve would transform Bitcoin from an apolitical asset into the plaything of the government, subject to Washington's political cycles. Bitcoiners were never ones to hitch their wagon to the government, and they shouldn't start now.”

So it seems, David, like there's essentially this coalition of more libertarian crypto supporters who think, “We actually don't want to be embraced by the federal government in this way,” because that could undermine some of the libertarian ideology of the initial crypto wave.

Yeah, I mean, as you said, Nick was opposed even to the idea of a Bitcoin reserve before this whole notion of a crypto reserve came up, and so it's definitely been divisive. Crypto—Bitcoin, in particular—started as this renegade economic movement. So to be begging the government to buy up a huge stockpile so that the price goes up seems sort of antithetical to a lot of those principles. Most people in the industry don't really care about those principles anymore, but some still do.

Well, let's talk about this more cynical take: that this is all just a self-enrichment scheme by the Trump administration. I know about the Trump meme coins. We've talked about those. I know that a lot of Trump supporters and friends are heavily invested in the crypto industry, but what do we actually know about the Trump family's personal interest in crypto beyond this small set of meme coins? I've seen Eric Trump tweeting on and off about crypto and various coins that he's interested in. Do we know much about the Trump family's overall exposure to the crypto markets?

We don't have a window into Eric Trump's personal crypto portfolio, but what we do know is that World Liberty Financial, the crypto business that the Trump family helped start last year and from which the Trump family profits directly, has, over the last few months, accumulated a huge stash of various cryptocurrencies.

It was supposed to launch some sort of crypto application, a DeFi project. That hasn't come to fruition yet, but this firm is sitting on a huge stash of tokens, so there's arguably a financial benefit for the Trump family in seeing those tokens rise in value.

Right, and some of the people that President Trump has tapped to lead crypto policy efforts in the federal government have also invested at various points in crypto stuff. David Sacks, 1 of Trump's advisors, has made some crypto investments over the years. He's said that he's divested from those to prevent conflicts of interest, but there are other people in the administration who presumably still do have active investments in these categories. So what do people in crypto think of the argument that this is all just a bunch of insiders trying to use the powers of the government to enrich themselves?

People in the crypto world are super cynical and love enriching themselves at all times, but they're also super paranoid and suspicious and are constantly convinced that there's some sort of conspiracy to screw them over. Those 2 instincts have clashed, and you can sort of see the internal battle playing out in a lot of crypto people who like Trump, where they ignore the conflict of interest that benefits them when it means SEC suits getting dropped, but they pay attention to it when it means that their favorite crypto didn't make it into the reserve but someone else's did.

And so I think that's sort of the conflict that's playing out in the industry, and why there's been some public backlash.

Yeah. So how likely do we think the strategic crypto reserve is to actually happen? Does Congress need to do anything? Can Trump just do this on his own by executive action? What are we going to see in the next couple of weeks and months?

Yeah, I mean, the way crypto people have made this distinction is: Is it a stockpile or is it a reserve?

Mm.

And if it's a stockpile, what they mean is we're just keeping the crypto we already have. If it's a reserve, that means we're buying more crypto on the market. I don't think there's anything inherent in those words that means those things, but that's how crypto people talk about it.

If the Trump administration goes the stockpile route and just keeps the Bitcoin it already has, that won't require any spending. It just requires a change in policy: Let's stop trying to sell this stuff. But if the administration goes the reserve route, and as far as we know, the government doesn't actually own any Cardano right now, so it would actually have to buy it to put it in the reserve.

If the government goes that direction, then yes, there's a question of how you would fund that. Congressional appropriation is the most obvious route, especially if it's a huge purchase. But since the summer, people in the crypto world have floated various out-there legal ideas about how Trump could push this through on his own.

7. Crypto Enters The Mainstream

Mm. Okay, so David, obviously a lot of what's getting attention in crypto these days is related to meme coins and strategic reserves and whatnot, but there's also some quieter crypto stuff that I understand is happening in Washington that could end up also being quite important, but may be flying under people's radars because it's not frankly that interesting. One of them is this stablecoin bill, which actually looks like it might pass with bipartisan support. Tell us about that.

So 1 of the big priorities of the crypto industry in this administration is to basically convert all the political goodwill it has into legislation, and there are 2 key pieces of legislation that the industry wants to advance.

One is this stablecoin bill, which would essentially create rules for stablecoin companies to operate in the U.S. It wouldn't do anything super crazy, but it would essentially create a regulatory framework for stablecoins that would, in some way, put a kind of government seal of approval on this branch of the crypto industry, which is super important. And so that's something that the industry has been pushing for, and which critics say, “Look, I mean, this could be a kind of gateway drug to the U.S. allowing other, even more dangerous crypto stuff to seep into the mainstream economy.” So that's the first priority.

The second priority is a market structure bill, and basically what that means is legislation that would strip power away from the SEC, which has obviously been super hard on crypto, and give it to the CFTC, the much weaker, less aggressive agency. So those are the 2 big legislative things that crypto people wanna do.

Hmm.

So I wanna bring this back to the question of the information gap that exists in crypto right now. My experience, and maybe your experience too, is that the people who understand crypto the deepest, who can talk about the L2s and the stablecoins and the DEXs, most of them are invested in crypto in some way, which is why they have taken the time to understand all of the extremely complex parts of the crypto ecosystem. But what worries me about that is that there is essentially no principled opposition left, it seems to me, in Washington, D.C.

The people who are making the policy, who are having the debates about the policy, they all come from the same universe of people who are bullish, by and large, about crypto. Maybe they have some disagreements about which crypto assets should be included in a strategic reserve or exactly how some regulations would be written, but there doesn't seem to me to be anyone left in Washington who knows what they're talking about and can stand up to some of these schemes from the industry. Is that an accurate read of the situation?

Yeah, partly this is a function of the changing administration. There were a huge number of people in Washington who knew about crypto and were in positions of power. Gary Gensler himself taught an MIT course on crypto. This wasn't some naive guy who didn't know the difference between Bitcoin and Ethereum. He was running the SEC. Now he's out, and the crypto people have the run of the town.

I also think there's an element of fatigue here that parallels the broader fatigue over what's going on with Trump. Some of the loudest crypto critics, people like the actor Ben McKenzie, who was super vocal during the last bull run and crypto crash, feel like they made their point. They were screaming about how dangerous crypto was, and then the whole thing crashed, and Sam Bankman-Fried went to prison, and they were proved right in a lot of ways. And yet crypto came surging back, and I think a lot of people don't have the energy to fight this fight all over again.

Maybe part of that is just the assumption that, when that crash happened, it seemed like the people who were affected were the ones that had decided to take the gamble on crypto, whereas people who just ignored the whole thing were basically fine. And so maybe that's why people are prioritizing their energy a little differently. What I think is concerning, though, is what you said, which is, look, if we pass a stablecoin bill in this country, if crypto becomes more connected to the main economy, then all of a sudden, if there's another crash, normal people might start to feel the pain.

Yeah, absolutely. Crypto used to be walled off from the real economy, and it still is in a lot of important ways, but that's beginning to change. The walls are coming down. We already have a Bitcoin ETF. We could have an XRP ETF soon, a Solana ETF. That just creates more avenues for people to put their traditional savings into crypto.

If your employer tells you that they're gonna start paying you in Cardano, look for another job. That's what I'm gonna say.

But I will say that I think there's also an element of complexity here that makes it very hard for people who are not spending hours a day trying to keep up with what's going on in crypto to make sense of it all. It reminds me a little bit of what happened on Wall Street during the financial crisis, where you suddenly had these assets that no regular people had ever heard of: mortgage-backed securities, credit default swaps, synthetic bonds.

These things that were so esoteric and hard to understand all of a sudden became quite important, and the people who understood them had been trading them for years, trying to make a bunch of money. But they had not been sharing that information because it was not in their interest to share that information. They didn't want us to know what was going on. And so my fear is that there's something similar happening in crypto right now, where the people who really understand this stuff are making a killing on it, possibly at the expense of the rest of us.

Hmm.

Yeah, and the complexity was a barrier to regular people getting into it for a long time. But with these ETFs coming onto the market, suddenly there's a way for you to buy this from your brokerage account, to have it factored into your retirement investments. It just increases the level of exposure.

Yeah. Well, DYB, thanks for coming on.

DYB.

Great to see you.

TYVM.

Thanks for having me.

Never actually heard anyone say that out loud. I thought it was just a text thing.

Didn't sound great.

I thought it sounded great.

Yeah. When we come back, we'll follow up on our episode about vibe coding and what you built with your vibe coding tools.

It's time to share with the class.

8. Vibe Coding Goes Practical

Well, Casey, a couple weeks ago on the show, we talked about AI vibe coding. This is the idea of using AI tools to build apps and websites and other things for yourself, even if you don't know how to code. I talked about some of the projects I'd been building. I built you a hot tub time machine for your hot tub, and we asked listeners to tell us what kinds of projects they were vibe coding. And Casey, we got a lot of responses.

We got so many responses. I'm told it was more than 60 in all, and it was so fun to read through these. Some of the people who wrote in had some amount of technical expertise, and it does seem like the more technical expertise they had, the further they were able to get. But we were also hearing from people who had never coded before and were just trying to see how far they could go with the tools that they have today, and in some cases, the answer was pretty far.

Yeah, and I wrote a column about this last week, about my vibe-coding experience, and I also got a ton of feedback—emails, comments, texts—from people who said, “I was pretty skeptical about these tools. Maybe I hadn't used them in a while, but after I read your article or heard your segment, I went out, I tried this stuff, I built something.” So I do think that there's something about having this firsthand encounter with this technology that is persuasive in a way that two guys yakking about it are not.

Yeah, and for what it's worth, after we had that conversation, I went back to my house, got out my laptop, and attempted to vibe code.

Yeah? What'd you vibe code?

I made a little platformer video game with a frog in a bathrobe jumping around various platforms, collecting coins. I realize that sounds absurd. The absurd thing was I was able to make that.

Yeah.

Because I just typed a few words into a box.

It works?

It works. Now, is it a very fun video game? No, but it exists.

That's amazing.

It is amazing.

Wait, can I play it?

Absolutely.

What's it called?

It's called Frog in a Bathrobe.

Okay. So before we hear about our listeners' vibe-coding experiments, I should also check on the status of the hot tub time machine. How's it doing?

Well, first of all, I have to say thank you. It was very kind of you to spend some time in your busy life creating a program that would attempt to instruct me when to put various chemicals and do other forms of maintenance on my hot tub. Hot Tub Time Machine did seem to get a little out of control.

Oh, no. What happened?

I will say that after our episode, it was emailing me basically every day and telling me to do quarterly maintenance after I'd owned my hot tub for a week. So fortunately, the emails have slowed down a bit, but it did introduce some questions into my mind about the reliability of Hot Tub Time Machine.

Yeah. Well, I'm sorry about that, and I have not looked into that because, frankly, I wouldn't even know how to start debugging something like this. This is one of the weird things about vibe coding: You can create something that works, sort of, but if it ever breaks, you have no idea what you're doing. You just have to poke at it and say, “Fix it.”

That's the exact approach I take to home maintenance. If it breaks, oh boy. I'm vibing in the house all day, but then—uh-oh—this stopped working.

Why can’t—why isn’t the ice maker—

Yeah.

—making ice? I don’t know.

I vibe-assembled an IKEA shelf the other night. Let me tell you.

Don’t stand underneath it during an earthquake. That’s what I’ll say about that shelf.

Okay, let’s get to our listeners, and let’s hope that they are doing a little bit better than we are with our vibe-coding experiments.

Let’s do it.

9. Listeners Build New Tools

Hi, my name is Mike Lapchick, and I live in Chicago’s Wicker Park neighborhood. First of all, the app I made is a speed reader for myself and for my dyslexic ADD friends who can’t read without jumbling up letters and words and losing focus altogether. How it works is you paste text in, say, from an article, and it displays one word at a time at around 400 words per minute, versus 250 words per minute for normies.

I’m not a coder, but I did run the product team at my last startup, and the experience was so similar to working with a developer. The better I defined the project upfront, the better the results. If I started with a loose definition, it could become really laborious, again, just like working with a real developer. Thanks for letting me share. Keep up the great work.

So, Kevin, how did you understand what Mike has built?

This is interesting because it’s similar to things that I think have been around for a number of years, which are these speed-reading apps that basically take blocks of text, whether it’s an article or a book or just a long email, and they basically do what is called rapid serial visual presentation, or RSVP. That’s basically flashing one word at a time on your screen very quickly, so that instead of reading left to right, you’re basically just getting the fire hose directly at you. You don’t have to move your eyes at all. It’s just one word after another.

People who swear by these apps say it helps them read much faster. I’ve never been a big user of them, but it could also be helpful for something like dyslexia, where it maybe helps you avoid transposing letters—

Hmm.

—and getting confused that way.

So I love this. We hear from listeners sometimes who want us to talk more about accessibility issues and technology, and maybe one reason why we don’t talk about it more is that it is a topic that platforms generally give short shrift to, right? We do not see huge investments being made in accessibility software, and so this feels like a perfect use of vibe coding.

You are scratching your own itch. You are building the thing that you cannot trust the company to build for you, and now you have a tool that is maybe going to be useful to folks who have dyslexia and ADHD.

Yeah. I should say, I also think this is a great idea, and I think it also demonstrates how disruptive vibe coding can be. There are speed-reading apps that work basically as Mike described that are out there on the market that you have to pay for. You could go download one of those, and maybe it’ll be a little more polished and have a few more features, or you could just make one yourself for whatever subscription vibe-coding product you’re using and build your own that you can use over and over again for free.

Yeah, very interesting. All right, let’s hear the next one.

My name is Lauren Buell, and I’m a full-time practicing anesthesiologist based in Hanover, New Hampshire. I started vibe coding with Claude back in November 2024 and created this app called ConsultCraft that uses adaptive AI to simulate anesthesiology case discussions and provide real-time feedback that’s based on case scripts and answers written by me.

I have no coding experience and just a very basic understanding of the command line, so it took me about 6 weeks in my spare time to make something that was functional enough that I could send it to my colleagues. Now it’s been about 3 months of adding new features and cases weekly, including a database so I can save transcripts for research.

I expected to hear mostly from my residents, who were using it to prepare for their oral board exams, but I’ve also gotten a lot of positive feedback from colleagues who say things like, “I haven’t really done that particular case in a while in the operating room, and this was a nice refresher,” which has been really great to hear and really unexpected.

Well, I gotta say, this anesthesiologist did not put me to sleep.

No, this perked me right up.

Yeah.

So explain what Lauren has built here.

Okay, so I’m looking at Lauren’s app right now. It’s very pretty. It’s called ConsultCraft, and it looks like basically a case simulator where you can say something that I don’t understand, like, say, you have a transcarotid artery revascularization for carotid stenosis.

I do have that.

And you can click on that, and you can go into a tutor mode where it explains the case: a 77-year-old man with a history of hypertension, something else I can’t pronounce, and diabetes mellitus presents to the ED with an episode of right-sided amaurosis fugax.

Wasn’t Omarosa in the first Trump administration?

That’s Omarosa.

Oh, that’s right.

Yeah. So it basically says, “What do you think of this case?” Then you can type your response and get some feedback, or there is a voice mode. This is very cool.

Yeah.

And I get why this took Lauren 6 weeks to build. It is not a simple tool, but I can see how this would be quite useful.

Absolutely, and I find this so inspiring because all of us have some kind of itch in our professional lives where we feel like, if only we had a tool like this, it would make our life so much easier. Up until now, if you couldn’t write software, you probably couldn’t do much about this, certainly not in the digital realm.

But Lauren figured out, “You know what? I’m going to apply myself. I’m going to spend 6 weeks, and I’m going to build something that now is benefiting not just myself, but all of my colleagues.” This is just a great example of the potential of vibe coding.

Yeah, and it actually illustrates something that I found when I wrote about vibe coding, which was that I had built all these software products that I thought were only useful for me, like this tool to help me recommend what to pack for my kids’ lunch. I put it out there in my story that I built this, and then I started hearing from people saying, “Can I get access to this tool?”

So it turns out that if you build something just for yourself, there are probably at least a few other people in the world who could benefit from it too.

So true. Let’s hear from our next listener.

Hey, Kevin and Casey. This is Matt. I’m a graphic designer, I live in Los Angeles, and I’ve been vibe coding for about 6 months. I created a project called Flavor Finder with Claude, and it’s a combination of the data set of The Flavor Bible, which is an excellent cooking reference, along with the UI of a color-palette generator.

You can go one by one and build the palette as you go, or you can just push Generate, and it will create a 5-ingredient flavor pairing where all of the ingredients pair well together. You can lock ingredients and substitute, move them around. It’s actually been a really fun way to discover new stuff in the kitchen.

Another exercise that I do with a lot of the LLMs is something I call Rochambeau. It’s a combination of the 2 worst games of all time: tic-tac-toe and rock-paper-scissors. I just say, “Hey, create a little playable web app for me combining these 2 games. They’re so bad. Make something fun.”

I have to say, unfortunately, it struggled to create anything that’s very entertaining, but it is an interesting way to observe how an LLM is thinking.

I should just start by announcing that The New York Times Company has acquired Rochambeau for a billion dollars.

Move over, Wordle.

So here’s Flavor Finder. If you see some ingredients that might work well together here—watercress, pineapple, pork, shallot, and olive oil—you can have it generate different combinations of things that fit the taste profile you’re looking for. Look, I am not a good enough home cook to take advantage of something like this, but I know a lot of people probably are.

I might try this.

I think you should.

The next step beyond vibe coding is vibe cooking. And I would like to challenge him to a game of Rochambeau.

All right, the next one comes to us from listener Zach.

Hi, Kevin and Casey. My name is Zach. I’m a photographer in New York City, and like Kevin, I was once a young nerd who bounced off of coding in exchange for photography and Flash and Dreamweaver and all that.

Lately, I’ve been using a combination of ChatGPT, Claude, and DeepSeek to help me write these little custom shell or Python scripts that help me with some mundane photo tasks. For example, I had a recent job where I had almost 1,000 images that needed to be renamed based on the subject, moved into subfolders based on that name, and exported from their layered Photoshop files into JPEGs and TIFFs in different file sizes.

It was a whole mess, and AI helped me write a script to automate this organization process. It even tried to make a Photoshop script to help with the export, but it was pretty bad at that language.

But, of course, with all the debugging, it probably took twice as long as it would've taken to do manually. But it was a lot more fun, and now I have this cool script I can tweak in the future. Thanks.

Yeah, so I feel like a question that we never quite answer enough is, what do we actually want AI for? To me, an answer to that question is that most people's lives are filled with unimaginable drudgery and things that take so long, are so tedious, and require zero creativity—really none of their human skills whatsoever. Zach found himself in exactly this situation, right? He has 1,000 photos. He needs to manipulate them. It's going to take forever, but can he just vibe code a solution that takes him twice as long as if he had just done it himself? I think that's inspiring.

Yeah, it is, but I also think he's being a little facetious about this, because it does seem like this is a process that he does frequently, and so this tool will help him save time in the future. Yes, there is some setup cost to building this tool for yourself, but if this is something that you're planning to do over and over again, it may actually make sense to build a tool rather than doing it manually every time.

Yeah, absolutely. And honestly, to me, the sort of BS I want to deal with is wrangling with the computer rather than the tedium of doing it myself. I'm happier trying to create the tool that hopefully has multiple uses in it than I am just being like, "Well, it'll be faster if I just manually rename 1,000 photos."

Right, and I think the part of Zach's response that I resonate with so much is the joy of it.

Yeah.

This is something that I got a lot of blowback on when I published my column. People said, "Oh, does any of this stuff actually work? Does it actually save you time? How hard is it to actually decide what to put in your kid's lunchbox yourself, you lazy jerk?" And I understand that. I think for me, the point of vibe coding is not pure efficiency. It is also discovery and exploration, and I find it very fun to watch the code fly past as the computer goes to work building something for me.

Yeah. To make a point that I think is sometimes underappreciated, it is fun to learn, and it is fun to make things.

Yes.

And these tools help people learn to make things—

Yes.

—and that's cool.

Yep. Thanks, Zach. All right, let's hear from our last vibe coding listener of the day. This one is more of a request.

Hi, Kevin and Casey. My name is Ashley. I'm from Southern California, and I am a working mom with 4 kids: a 9-year-old, a 6-year-old, a 4-year-old, and a 2-year-old. My vibe code idea is that I just need an app that goes through all of the social things that I am supposed to read, capture, and do something with across my kids' social calendars. So it's email, WhatsApp chat, Insta chat, text messages, Evites for the birthdays, even Facebook Messenger. I don't know if that's still a thing, but I just need something that's going to summarize, plan, give me a to-do list, and put it on a shared calendar, which I guess sounds like something that a wife would be able to do, but I'm not that interested in that. So if you could just vibe code me a bit of a wife, that would be awesome. All right. I love the show. Thanks.

Thank you, Ashley. Truly, I don't know if I've ever been so delighted on this show as I have been listening to Ashley's children stampede through her house in the background while she was recording this request, which I think is a great request.

Yeah. Also, 4 kids as a working mom—I'm tired just thinking about it.

Yeah. God bless.

God bless you.

God bless Ashley. Now, look, this one is a challenge. First of all, the idea is great.

Yes.

I want this, too, and I don't even have 4 kids. I think the challenge is that Ashley wants many disparate services to interact with each other, and right now, those services mostly do not have ways of interacting. But Kevin, as a more seasoned vibe coder, do you feel like you know an approach that could work?

Yes. I think we can do this.

Okay.

I don't think it is going to be very straightforward, because as you said, it does require the sort of interoperability of a number of different apps that don't natively work all that well together. But Ashley, I think we can get this done for you, and it is my pledge that we will do our best, because I honestly need this, too. I've only got 1 kid, but I'm telling you, man, the toddler's birthday-party circuit is out of control.

It never stops.

Yeah, it never stops. Kids are always having birthdays, and there are so many of them. Some say no gifts, and some say gifts, and some are at parks, and some are at zoos, and you have to keep track of all of it. My God, how does anyone do this?

How does anyone do it, and can it be vibe coded?

So, for my own sake and for yours, Ashley, I'm going to try to vibe code this app for you. I'll keep you posted.

And you're going to report back to us.

I will.

Okay. Well, it makes sense if the sort of narrative arc of this show is an AI trying to get you to leave your wife, to you mastering AI to vibe code a wife. That feels like a great sort of season finale for the Hard Fork show.

I agree.

Yeah.

I agree.

Yeah. Well, listen, thanks again to everyone who emailed us. All of the emails delighted us. We feel like you guys really like trying new stuff and showing us what you're trying, and it inspires us, so thank you.

One more thing before we go this week, Hard Fork is looking for an editor, specifically someone who can take the raw material of what Kevin says and try to make it make sense for the rest of us. So if you think you might have that skill set, please go to nytimes.com/careers and look for the Hard Fork editor job. We'd love to talk to you.

Hard Fork is produced by Whitney Jones and Rachel Cohn. We're edited by Rachel Dry. We're fact-checked by Ina Alvarado. Today's show was engineered by Chris Wood. Original music by Sophia Lanman, Diane Wong, and Dan Powell. Our executive producer is Jen Poynt, and our audience editor is Nela Galogly. Video production by Chris Schott, Sawyer Roquet, and Pat Gunther. You can watch this full episode on YouTube at youtube.com/hardfork. Special thanks to Paula Schumann, Puiwing Tam, Dalia Haddad, and Jeffrey Miranda. You can email us, as always, at hardfork@nytimes.com. And if you happen to be in Austin, Texas, for South by Southwest this weekend or early next week, Casey and I will be there. Catch us around town. Say hi.

Say hi.

Let's get some tacos.

Let's vibe code.