Casey, let's start the episode today with a little housekeeping. So we've talked on the show the last couple of weeks about the New York Times audio subscription. This is how you can connect your New York Times subscription on Apple Podcasts and Spotify to get full access to every episode of Hard Fork and all of the other shows from the New York Times. This subscription is now live. You may be running into it if you use those platforms and you should link your account. I did it this week. Very easy. Took me less than a minute. And now I can get all of my favorite
new shows and their whole back catalogs. We talked about A.I. Slop on the show last week and how hard it is to understand what's true and what's not true out there on the internet. After that show, I did actually subscribe to several different publications and I now pay for several different publications because I was like, if I'm preaching the value of good information out there, I got to be willing to walk the walk. You're being the change you wanted to see in the world. Yes. So if you want to be the change you wish to see in the podcast world, you can support
the new podcast. You can subscribe to the New York Times audio subscription. Do it. So that is our third and final pitch for the New York Times audio subscription. But there's another piece of housekeeping this week, which is that after this episode is over, you will hear a segment from a different New York Times podcast that'll sort of play just at the tail end of this episode. It's an episode of the New York Times newest podcast, the Wirecutter Show. And let me tell you this is an episode about laundry, which might not sound like it's on a traditional wheelhouse, but I'm going to tell you something. I went to a dinner party recently and a person
said, we got to talking and he said, have you heard the Wirecutter podcast episode about laundry because it changed my life? And so now we're giving that to you for absolutely free. So please buy a subscription to New York Times audio. If we sell the most subscriptions, if we sell more than Ezra and Michael Barbaro, we do get a free concert by the backstreet voice. And I'll say it. I want it that way. Casey, you are my fire.
We just got another email from somebody who said they thought I was bald. I have an apparently crazy bald energy. You have all the energy podcast. What do you think is a bald seeming about you? I think for me, they think of me as a wacky sidekick, which is a bald energy. You know, I think so. I don't think of I don't associate wacky and bald because I'm thinking Jeff Bezos. I'm like,
I know a lot of very hard core. So do you think that maybe people think that I sound like a sort of Titan of industry plutocrats? I would not say that's the energy you're giving is plutocrat energy, but really because I just fired 6000 people to show that I could. You did order me to come to the office today. I did I said there's a return to office and I'm actually immediately know what questions.
I'm Kevin news attack columns from the New York Times. I'm Casey Newton from platformer and this is hard for this week. Are we reaching the AI and game? A new essay from the CEO of Antwopic has Silicon Valley talking. Then Uber CEO, Dara Coast Roshagi joins us to discuss his company's new partnership with the New York Times.
The company has a new relationship with Waymo and the future of autonomous vehicles. And finally, internal TikTok documents tell us exactly how many videos you need to watch to get hooked. And so I did. Okay, brave God help me. Well, Kevin, the AI race continues to accelerate and this week the news is coming from Antwopic. Now last year, you actually spent some time inside this company and you called it the White Hot Center of AI Dumerism.
Well, the headline of my piece called it the White Hot Center of AI Dumerism. Just a clarify blame the headline. Well, you know, reporters don't often write our own headlines. So I just feel the need to clarify that. But the story does talk about how many of the people you met inside this company seemed strangely pessimistic about what they were building.
Yeah, it was a very interesting reporting experience because I got invited to spend several weeks just basically embedded at Antwopic as they were gearing up to launch an update of their chatbot, Claude. And I sort of expected it, you know, they would go in and try to impress me with how great Claude was and talk about all the amazing things it would allow people to do.
And then I got there and it was like all they wanted to do was talk about how scared they were of AI and of releasing these systems into the wild. I compared it in the piece to like being a restaurant critic who shows up at like a busy new restaurant. And all anyone wants to talk about his food poisoning.
And so for this reason, I was very interested to see over the past week, the CEO of Anthropic, Dario Amade, write a 13,000 word essay about his vision of the future. And in this essay, he says that he is not an AI Dumer does not think of himself as one, but actually thinks that the future is quite bright and might be arriving very quickly.
And then shortly after that, Kevin, the company put out a new policy, which they call a responsible scaling policy that I thought had some interesting things to say about ways to safely build AI systems. So we wanted to talk about this today for a couple reasons. One is that AI CEOs have kept telling us recently that major changes are right around the corner. Sam Altman recently had a blog post where he said that an artificial super intelligence could be just a few thousand days away.
And now here Amade is saying that AGI could arrive in 2026, which check your calendar, Kevin, that is in 14 months. And certainly there is a case that this is just hype, but even so, there are some very wild claims in here that I do think deserve broader attention.
The second reason that we want to talk about this today is that Anthropic is really the flip side to a story that we've been talking about for the past year here, which is what happened to open AI during and after Sam Altman's temporary firing as CEO. Anthropic was started by a group of people who left open AI primarily over safety concerns and recently several more members of open AI's founding team and their safety research teams have gone over to Anthropic.
And so in a way, Kevin, Anthropic is an answer to the question of what would have happened if open AI's executive team hadn't spent the past few years following apart. And while they are still the underdog compared to open AI, is there a chance that Anthropic is the team that builds AGI first? So that's what we want to talk about today, but I want to start by just talking about this essay. Kevin, what did Dario Amadeh have to say in his essay, Machines of Love and Grace?
Yeah, so the first thing that struck me is he is clearly reacting to this perception, which I may have helped create through my story last year that sort of he and Anthropic are just dooms, right? And so I think that they are just a company that goes around warning about how badly AI could go if we're not careful. And what he says in this essay that I thought was really interesting and important is, you know, we're going to keep talking about the risks of AI.
This is not him saying, I don't think this stuff is risky. I've been, you know, take it out of context and I'm actually an AI optimist. What he says is, it's important to have both, right? You can't just be going around warning about the doom all the time. You also have to have a positive vision for the future of AI, because that's not only what inspires and motivates people, but it matters what we do.
I thought that was actually the most important thing that he did in this essay was he basically said, look, this could go well or it could go badly. And whether it goes well or badly is up to us. This is not some inevitable force. Sometimes people in the AI industry, they have a habit of talking about AI as if it's just kind of this disembodied force that is just going to happen to us.
Yes, and we either have to sort of like get on the train or get run over by the train. And what Dario says is actually different. He says, you know, this is, here's a vision for how this could go well, but it's going to take some work to get there. It also made me realize that for the past couple of years, I have heard much more about how AI could go wrong than how it could go right from the AI CEOs, right?
As much as these guys get knocked for endlessly hyping up their products, they also have, I think to their credit, spent a lot of time trying to explain to people that this stuff is risky. And so there was something almost counterintuitive about Dario coming out and saying, wait, let's get really specific about how this could go well. Totally. So I think the first thing that's worth pulling out from this essay is the timeline.
Right, because as you said, Dario Amade is claiming that powerful AI, which is sort of his term, he doesn't like AGI, he thinks it sounds like two sci-fi, but powerful AI, which he sort of defines as like an AI that would be smarter than a Nobel Prize winner in basically any field. And that it could basically control tools go do a bunch of tasks simultaneously. He calls this sort of a country of geniuses in a data center. That's sort of his definition of powerful AI.
And he thinks that it could arrive as soon as 2026. I think there's a tendency sometimes to be cynical about people with short timelines like these like, oh, these guys are just saying this stuff is going to arrive so soon because they need to raise a bunch of money for their AI companies. And, you know, maybe that is a factor.
But I truly believe that at least Dario Amade is sincere and serious about this. This is not a drill to him. And Anthropic is actually making plans, scaling teams and building products as if we are headed into a radically different world very soon, like within the next presidential term.
Yeah. And look, Anthropic is raising money right now. And that does give Dario motivation to get out there in the market and start talking about curing cancer and all these amazing things that he thinks that AI can do.
At the same time, you know, I think that we're in a world where the discourse has been a little bit poison by folks like Elon Musk who are constantly going out into public making bold claims about things that they say are going to happen, you know, within six months or a year and then truly just never happen. And our understanding of Dario based our own conversations with them and of people who work with them is like, he is not that kind of person.
This is not somebody who lets his mouth run away with him. When he says that he thinks this stuff could start to arrive in 14 months, I actually do give some credibility. Yeah. And, and, you know, you can argue with the timescales and plenty of smart people disagree about this. But I think it's worth taking this seriously because this is the head of one of the leading AI labs sort of giving you his thoughts on not just what AI is going to change about the world, but when that's going to happen.
And what I liked about this essay was that it wasn't trying to sell me a vision of a glorious AI future right Dario says, you know, all or some or none of this might come to pass. But it was basically a thought experiment. He has this idea in the essay about what he calls the compressed 21st century. Yeah. He basically says, what if all AI does is allow us to make 100 years worth of progress in the next 10 years in fields like biology? What would that change about the world?
And I thought there was a really interesting way to frame it. Give us some examples, Kevin, of what Dario says might happen in this compressed 21st century. So what he says in this essay is that if we do get what he calls powerful AI relatively soon, that in the sort of decade that follows that we would expect things like the prevention and treatment of basically all natural infectious disease.
The elimination of most types of cancer, very sort of good embryo screening for genetic diseases that would make it so that more people didn't die of these sort of hereditary things. He talks about there being improved treatment for mental health and other ailments. Yeah, I mean, and a lot of this comes down to just understanding the human brain, which is an area where we still have a lot to learn.
And the idea is if you have what he calls this country of geniuses that's just operating on a server somewhere and they are able to talk to each other to dream to suggest ideas to give guidance to human scientists and labs to run experiments, then you have this massive compression effect on all of a sudden you get all of these benefits really soon.
Obviously, the headline grabbing stuff is Dario things we're going to cure all cancer and we're going to cure Alzheimer's disease obviously those are huge, but there's also kind of the more mundane stuff like do you struggle with anxiety? Do you have other mental health issues? Are you mildly depressed? It's possible that we will understand the neural circuitry there and be able to develop treatments that would just lead to a general rise in happiness. And that really struck me.
Yeah, and it sounds when you just describe it that way. It sounds sort of utopian and crazy, but what he points out and what I actually find compelling is like most scientific progress does not happen in a straight line, right?
You have these kind of moments where there's a breakthrough that enables a bunch of other breakthroughs. And we've seen stuff like this already happened with AI like with alpha fold, which won the frickin Nobel Prize this year in chemistry, where you don't just have a cure for one specific disease. But you have a way of potentially discovering cures for many kinds of diseases all at once.
There's a part in the essay that I really liked where he points out that CRISPR was something that we could have invented long before we actually did, but essentially no one had noticed the things they needed to notice in order to make it a reality.
And he posits that there are probably hundreds of other things like this right now that just no one has noticed yet. And if you had a bunch of AI agents working together in a room and they were sufficiently intelligent, they would just notice those things and we be off to the races.
And what I liked about this section of the essay was that it didn't try to claim that there was some completely novel thing that would be required to result in the changed world that he envisions, right? All that would need to happen for society to look radically different 10 or 15 years from now.
Indario's mind is for that sort of base rate of discovery to accelerate rapidly due to AI. Now let's take a moment to acknowledge folks in the audience who might be saying, oh my gosh, will these guys stop it with the AI hype? They're accepting every premise that these AI CEOs will just shovel it. They can't get enough and it's irresponsible. These are just stochastic parrots, Kevin, they don't know anything. It's not intelligence and it's never going to get any better than it is today.
And I just want to say I hear you and I see you and our email address is as we're client showing you at the time.
But here's the thing. You can look at the state of the art right now. And if you just extrapolate what is happening in 2024 and you assume some rate of progress beyond where we currently are, it seems likely to me that we do get into a world where you do have these sort of simulated PhD students or maybe simulated super geniuses and they are able to realize a lot of these kinds of things.
Now maybe it doesn't happen in 5, 10 years. Maybe it takes a lot longer than that. But I just wanted to underline like we are not truly living in the realm of fantasy. We are just trying to get a few years and a few levels of advancement beyond where we are right now. And Dario does in his essay, Mix some caveats about things that might constrain the rate of progress in AI like regulation or clinical trials taking a long time.
He also talks about the fact that some people may just opt out of this whole thing like they just may not want anything to do with AI. There might be some political or cultural backlash that sort of slows down the rate of progress. And he says that could actually constrain this and we need to think about some ways to address that. So that is sort of the suite of things that Dario thinks will benefit our lives. There's a bunch more in there. We think we'll help with climate change, other issues.
But the essay has five parts and there was another part of the essay that really caught my attention Kevin. And it is a part that looks a little bit more seriously at the risks of this stuff because any super genius that was sufficiently intelligent to cure cancer could otherwise wreak havoc in the world. So what is his idea for ensuring that AI always remains in good hands? So he admits that he's not like a geopolitics expert. This is not like a two of us.
Right. And there have been a lot of people theorizing about what the politics of advanced AI are going to look like. Dario says that his best guess currently about how to prevent AI from sort of empowering autocrats and dictators is through what he calls an on taught strategy. Basically, you want a bunch of democracies to kind of come together to secure their supply chain to sort of block adversaries from getting access to things like GPUs and semiconductors.
And that you could basically bring countries into this democratic alliance and sort of ice out the more authoritarian regimes from getting access to this stuff. But I think, you know, this was sort of not the most fleshed out part of the argument. Yeah. Well, and I appreciate that he is at least making an effort to come up with ideas for how would you prevent AI from being misused.
But as I was reading the discussion around the blog post, I found this interesting response from a guy named Max Tagmark. Max is a professor at MIT, who studies machine learning. And he's also the president of something called the Future of Life Institute, which is a sort of nonprofit focused on AI safety.
And he really doesn't like this idea of what Dario calls the on taught the group of this, these democracies working together. And he says he doesn't like it because it essentially sets up and accelerates a race. It says to the world that essentially whoever invents super powerful AI first will win forever, right? Because in this view, AI is essentially the final technology that you ever need to invent because after that, it'll just, you know, invent anything else it needs.
And he calls that a suicide race. And the reason is this. And he has great quote. Horny couples know that it is easier to make a human level intelligence than to raise and align it. And it is also easier to make an AGI than to figure out how to align or control it. Wow. I never thought about it like that. Yeah. You probably never thought I would say horny couple on the show, but I just did.
So Kevin, what do you make of this sort of feedback? Is there a risk there that this effectively serves as a starter pistol that leads maybe our adversaries to start investing more in AI and sort of racing against us and, you know, triggering some sort of doomspiral. Yeah. I mean, look, I don't have a problem with China racing us to cure cancer using AI. Right. Like that's not if they get their first like more power to them.
But I think the more serious risk is that they start building the kind of AI that serves Chinese interest, right? That it becomes a tool for surveillance and control of people rather than some of these more sort of democratic ideals.
And this is actually something that I asked Dario about back last year when I was spending all that time at Anthropic because this is the most common criticism of Anthropic is like, well, if you're so worried about AI and all the risks that it could pose, like, why are you building it?
And I asked about this in his response was you basically said, look, there's this problem of in AI research of kind of intertwining, right? Of the same technology that sort of advances the state of the art in AI also allows you to advance the state of the art in AI safety, right?
And the same tools that that make the language models more capable also make it possible to control the behavior of the language models. And so these things kind of go hand in hand. And if you want to compete on the frontier of AI safety, you also have to compete on the frontier of AI capabilities.
And I think it's an idea worth considering to me, it just sounds like, wow, you are really standing on a knife's edge there. If you're saying in order to have any influence over the future, we have to be right at the frontier and maybe even gently advance the frontier and yet somehow not accidentally trigger a race where all of a sudden everything gets out of control.
But I do accept and respect that that is Daria's point. But isn't that kind of what we observe from the last couple of years of AI progress, right? Like open AI, it got out there with chat GPT before any of the other labs had released anything similar. And chat GPT kind of set the tone for all of the products that followed it. And so I think the argument from anthropic would be like, yes, we could sort of be way behind the state of the art.
That would probably make us safer than someone who is actually advancing the state of the art. But then we missed the chance to kind of set the terms of what future AI products from other companies will look like. So it's sort of like using a soft power in an effort to influence others. Yeah, and the way they put this to me last year was that they wanted instead of there to be just a race for raw capabilities of AI systems, they wanted there to be a safety race, right?
Where companies would start competing about whose models were the safest rather than whose models could, you know, do your math homework better. So let's talk about the safety race and the other thing that anthropic did this week to lay out a future vision for AI. And that was with something that has, I'll say, kind of a boring name, the responsible scaling policy. I understand this, you know, this maybe wasn't going to come up over drinks, you know, at the club.
Yeah, but I think this is something that people should pay attention to because it's an example of what you just said, Kevin, it is anthropic trying to use some soft power in the world to say, hey, if we went a little bit more like this, we might be safer. All right. So talk about what's in the responsible scaling policy that anthropic released this week.
Well, let's talk about what it is. And the basic idea is just that as large language models gain new abilities, they should be subjected to more scrutiny. And they should have more safeguards added to them. They put this out a year ago. And it was actually a huge success in this sense, Kevin, open AI went on to release its own version of it. And then Google deep mine released it's a similar scaling policy as well this spring.
So now anthropic is coming back just over a year later and they say, we're going to make some refinements. And the most important thing that they say is essentially we have identified two capabilities that we think would be particularly dangerous. And so if anything that we make displays these capabilities, we are going to add a bunch of new safeguards.
And the first one of those is if a model can do its own AI research and development, that is going to start ringing a lot of alarm bells. And they're going to put many more safeguards on it. And second, if one of these models can meaningfully assist someone who has a basic technical background in creating a chemical biological, radiological or nuclear weapon, then they would add these new safeguards.
What are these safeguards? Well, they have a super long blog post about it. You can look it up. But it includes basic things like taking extra steps to make sure that a foreign adversary can't steal the model weights, for example, or otherwise hack into the systems and run away with it. Right. And this is some of it is similar to things that were proposed by the Biden White House in its executive order on AI last year.
This is also these are some of the steps that came up in SB 1047, the AI regulation that was vetoed by governor Newsom in California recently. So these are ideas that have been floating out there in the sort of AI safety world for a while. But Anthropic is basically saying we are going to proactively commit to doing this stuff even before a government requires us to.
Yeah, there's a second thing I like about this and it relates to this as be 1047 that we talked about on the show, something that a lot of folks in Silicon Valley didn't like about it was the way that it tried to identify danger. And it was not because of a specific harm that a model could cause. It was by saying, well, if a model costs a certain amount of money to train, right, or if it is trained with a certain amount of compute.
Those were the proxies that the government was trying to use to understand why this would be dangerous. And a lot of folks in Silicon Valley said we hate that because that has nothing to do with whether these things could cause harm or not.
So when Anthropic is doing here is saying, well, why don't we try to regulate based on the anticipated harm? You know, obviously it would be bad if you could log on to Claude Anthropic's rival to Chatsy BT and said, hey, help me build a radiological weapon, which is something that I might type into a club because I don't even know the difference between a radiological weapon.
And a nuclear weapon to you. I hope you never learn. I hope I don't either because sometimes I have bad days Kevin and I get to scheming. So for this reason, I think that governments regulators around the world might want to look at this approach and say, hey, instead of trying to regulate this based on how much money AI labs are spending or like how much compute is involved.
Why don't we look at the harms we're trying to address and say, hey, if you build something that could cause this kind of harm, you have to do X Y Z. Yeah, that makes sense to me. So I think the biggest impact that both the sort of essay that Dario wrote and this responsible scaling policy had on me was not about any of the actual specifics of the idea. It was purely about the timescales and the urgency.
It is one thing to hear a bunch of people telling you that AI is coming and that it's going to be more powerful than you can imagine sooner than you can imagine. But if you actually start to internalize that and plan for it, it just feels very different. If we are going to get powerful AI sometime in the next, let's call it two to ten years, you just start making different choices.
Yeah, I think it becomes sort of the calculus. I can imagine it affecting what you might want to study in college if you are going to school right now. I have friends who are thinking about leaving their jobs because they think the place where they are working right now will not be able to compete in a world where AI is very widespread.
So yes, you are absolutely starting to see it creep into the calculus. I don't know what else it could do. There is no real call to action here because you can't really do very much until this world begins to arrive. But I do think psychologically we want people to at least imagine, as you say, what it would be like to live in this world because I have been surprised at how little discussion this has been getting.
Yeah, I totally agree. To me, it feels like we are entering... I wouldn't call it like an AI end game because I think we're closer to the start than the end of this transformation. But it does feel like something is happening. I'm starting to notice AI is affecting my life more. I'm starting to feel more dependent on it.
And I'm also... I'm kind of having an existential crisis. Not a full blown one. But typically, I'm a guy who likes to plan. I like to strategize. I like to have a five year and a ten year plan. And I've just found that my own certainty about the future and my ability to plan long term is just way lower than it has been for any time that I can remember.
That's interesting. I mean, for myself, I feel like that has always been true. You know, in 1990, I did not know what things were going to look like in 2040. And I would be really surprised by a lot of things that have happened along the way. But yeah, there's a lot of uncertainty out there. Scary, but I also like... Do you not feel a little bit excited about it? Of course. I look. I love software. I love tools. I want to live in the future. And it's already happening to me.
There is a lot of that uncertainty. And like that stuff freaks me out. But like, if we could cure cancer, if we could cure depression, if we could cure anxiety, you'd be talking about the greatest advancement to human well-being. Certainly a decade. Maybe that we've ever seen. Yeah. I mean, I have some priors on this because like my dad died of a very rare form of cancer that was...
It was like a sub 1% type of cancer. And when he got sick, it was like, you know, I read all the clinical trials. And it was just like... There hadn't been enough people thinking about this specific type of cancer and how to cure it because it was not breast cancer. It was not lung cancer. It was not something that millions of Americans get. And so there just wasn't the kind of brain power devoted to trying to solve this.
Now, it has subsequently... It hasn't been solved, but there are now treatments that are in the pipeline that didn't exist when he was sick. And I just constantly am wondering like if he had gotten sick now instead of when he did, like maybe he would have lived. And I think that is one of the things that makes me really optimistic about AI is just like...
Maybe we just do have the brain power or we will soon have the brain power to devote world-class research teams to these things that might not affect millions of people, but that do affect some number of people. Absolutely. I just... I don't know. It really... I got kind of emotional reading this essay because it was just like... You know, obviously it's... I'm not someone who believes all the hype, but I'm like...
I assign some non-zero probability to the possibility that he's right, that all this stuff could happen. And I just find that so much... More interesting and fun to think about than like a world where everything goes off the rails. Well, it's just the first time that we've had a truly positive, transformative vision for the world coming out of Silicon Valley in a really long time.
In fact, this vision, it's more positive and optimistic than anything that has been like in the presidential campaign. Yeah. When the presidential candidates talk about the future of this country, it's like, well, you know, we'll give you this tax break, right? Or we'll make this other policy change. Nobody's talking about how they're going to freaking care cancer. Yeah.
Right? So I think, of course, we're drawn to this kind of discussion because it feels like, you know, there are some people in the world who are taking really, really big swings and if they connect, then we're all going to benefit. Yeah. Yeah. So we come back. Why Uber has way more autonomous vehicles on the road than it used to. Okay. See, one of the biggest developments over the past few months in tech is that self-driving cars now are actually working.
Yeah. But this is no longer in the realm of sci-fi. Yes. So we've talked, obviously, about the self-driving cars that you can get in San Francisco now. It used to be two companies, Waymo and Cruz. Now it's just Waymo. And there have also been a bunch of different autonomous vehicle updates from other companies that are involved in this space. And the one that I found most interesting recently was about Uber. Now, as you will remember, Uber used to try to build its own robot axes.
They gave that back in 2020. That was the year they sort of sold off their autonomous driving division to a startup called Aurora after losing just an absolute ton of money on it. But now they are back in the game. And they just recently announced a multi-year partnership with Cruz, the self-driving car company. They also announced an expanded partnership with Waymo, which is going to allow Uber riders to get AVs in Austin, Texas and Atlanta, Georgia.
They've been operating this service in Phoenix since last year. And that's going to keep expanding. They also announced that self-driving Uber will be available in Abu Dhabi through a partnership with the Chinese AV company We Ride. And they've also made a long-term investment in Wave, which is a London-based autonomous driving company.
So they are investing really heavily in this, and they're doing it in a different way than they did back when they were trying to build their own self-driving cars. Now they are essentially saying, we want a partner with every company that we can that is making self-driving cars. So this is a company that many people take several times a week Uber. And yet I feel like it sometimes is a bit taken for granted.
And while we might just focus on the cars you can get today, they are thinking very long term about what transportation is going to look like in five or ten years. And increasingly for them, it seems like autonomous vehicles are a big part of that answer. And what I found really interesting, so Tesla had this Robotaxi event last week where Elon Musk talked about how you'll soon be able to hail a self-driving Tesla.
And what I found really interesting is that Tesla's share price plummeted after that event. But Uber's stock price rose to an all-time high. So clearly people think that, or at least some investors think that Uber's approach is better here than Tesla's. It's the sort of thing Kevin that makes me want to talk to the CEO of Uber. And lucky for you, he's here. Oh, thank goodness. So today we're going to talk with Uber CEO, Dara Khauser-Shawi, Heetogover at Uber in 2017.
After a bunch of scandals led the founder of Uber, Travis Kalanick, to step down. He has made the company profitable for the first time in its history. And I think a lot of people think he's been doing a pretty good job over there. And he is leading this charge into autonomous vehicles. And I'm really curious to hear what he makes. Not just of Uber's partnership with Waymo, but of sort of the whole self-driving car landscape. Let's bring him in. Let's do it.
Dara Khauser-Shawi, welcome to HardFork. Thank you for having me. So you were previously on the board of the New York Times company until 2017 when you stepped down right after taking over at Uber. I assume you still have some pull with our bosses though because of your years of service. So can you get them to build us a nicer studio? I didn't have pull when I was on the board and I certainly have zero pull now. I've got negative pull, I think they're taking revenge on me.
Well, since you left the board, they're making all kinds of crazy decisions like letting us start a podcast. Oh my God. So we are going to talk today about your new partnership with Waymo and the sort of autonomous driving future. I would love to hear the story of how this came together because I think for people who've been following this space for a number of years, this was surprising. Uber and Waymo have not historically had a great relationship.
The two companies were in litigation and lawsuits and trade secret theft and things like that. It was a big deal. So how did they approach you? Did you approach them? How did this partnership come together? I guess it's time healing. When I came on board, we thought that we wanted to establish a better relationship with Google generally, Waymo generally.
And even though we were working on our own self-driving technology, it was always within the context of we were developing our own, but we want to work with their parties as well. One of the disadvantages of developing our own technology was that some of the other players away, most of the world, etc, heard us, but it didn't necessarily believe us. It's difficult to work with players that you compete. So one of the first decisions that we made was we can't be in between here.
Either you have to go vertical or you have to go platform strategy. You can't achieve both. We either have to do our own thing or we have to do it with partners. Absolutely. That strategic fork became quite apparent to me. And then the second was just, what are we good at? Listen, I'll be blunt. We sucked at hardware. We tried to apply software principles to hardware.
It doesn't work. Hardware is a different price, different demand in terms of perfection, etc. And ultimately that fork, do we go vertical? And there are very few companies that can do software and hardware well. Apple Tesla, or arguably one of the few in the world. And we decided to make a bet on the platform. And so once we made that bet, we went out and identified who were the leaders. Waymo was a clear leader. First we had to make peace with them and settle in court, etc.
We got Google to be a bigger shareholder. And then over a period of time we built relationships. And I do think there's a synergy between the two. So it just makes sense the relationship. And we're very, very excited to on a forward basis expand it pretty significantly. So this was, I feel like maybe your most consequential decision to date as the CEO of this company.
If you believe that AVs are going to become the norm for many people hailing a ride in 10 or 15 years, it's conceivable that they might open up the Waymo app, right? And not the Uber app. Waymo has an app to order cars. I use it fairly regularly, right? So what gave you the confidence that in that world it will still be Uber that is the app that people are turning to and not Waymo or whatever other apps might have arisen for other AV companies? I think first is that it's not a binary outcome.
Okay. I think that Waymo app and Uber app can coexist. We saw it in my old job in the travel business, right? I ran Expedia and there's this dramatic is Expedia going to put the hotel chains out of business or the hotel chains going to put Expedia out of business. The fact is both throught and there's a set of customers who have books through Expedia. There's a set of customers who books hotel direct and both businesses have grown and interactivity in general has grown.
Same thing if you look in food, right? Make Donald's has his own app. It's a really good app has a loyalty program Starbucks has its own app has a loyalty program yet both are engaging with us through the Uber Eats marketplace. So my conclusion was that there isn't an either or I do believe there will be other companies.
There'll be cruises and there'll be we rides and waves etc. There'll be other companies and self driving choices and the person who wants utility speed ease familiarity will choose Uber and both can coexist and both can thrive and both are really going to grow. Because autonomous will be the future eventually. So tell us more about the partnership with Waymo that is going to take place in Austin and Atlanta.
Who is actually paying for the maintenance of the cars does Uber have to sort of make sure that there's no trash left behind in the cars like what is Uber actually doing in addition to just making these rides available through the app? Sure. So I don't want to talk about the economics because they're confidential in terms of the deal but in those two cities. Waymo will be available exclusively through the Uber app and we will also be running the fleet operations as well.
So depots recharging cleaning if something gets lost making sure that it gets back to its owner etc. And Waymo will provide the software driver will obviously provide the hardware repair the hardware etc. And then we will be doing the upkeep and operating the network. For riders if you want to get in a waymo in one of those cities through Uber.
Is there an option to specifically request a self driving waymo or is it just kind of chance like like if the car that's closest to you happens to be a waymo that's the one you get right now the experience for example in Phoenix is that it's by chance.
You got one by chance and you can say yes I'll do it or not and I think that's what we're going to start with but there may be some people who only want waymo's and there's some people who may not want waymo's and we'll solve for that over a period of time it could be personalizing preferences or it could be what you're talking about which is only one waymo. Do the passengers get rated by the self driving car the way that they would in a human driven Uber. Not yet but that's not a bad idea.
What about tipping like if I get out of a self driven Uber is there an option to sip the car if it did a good job? I'm sure we could build that. Why not? Honestly, I don't know. I do wonder if people are going to tip machines. I don't think it's likely but you never know. It sounds crazy but at some point someone is going to start asking because they're going to realize it's just free margin. You know, it's like even if only 100 customers do it in the whole year.
I don't know. You know, it's just free money. I mean that the good news is tipping 100% of tips go to drivers now and we definitely want to keep that so we like the tipping habits but whether people tip machines is TBD. Yeah. And how big are these fleets? I think I read somewhere recently that waymo has about 700 self driving cars operating nationwide. How many AVs are we talking about in these cities? We're starting in the hundreds and that will expand from there.
I know you don't want to discuss the economics even though I would love to learn what the split is there. I'm not going to tell you. But you did recently talk about the margins on autonomous rides being lower than the margins on regular Uber rides for at least a few more years. That's not intuitive to me because in an autonomous ride you don't have to pay the driver. So you would think the margin would be way higher for Uber. But why would you make less money if you don't have to pay a driver?
So generally our design spec in terms of how we build businesses is any newer business. We're going to operate at a lower margin while we're growing that business. You don't want it to be profitable day one. And that's my attitude with autonomous which is again get it out there introduce it to as many people as possible. At a maturity level generally if you look at our take rate around the world it's about 20% we get 20% the driver gets 80%.
We think that's a model that makes sense for any autonomous partner going forward and that's what we expect. I kind of don't care honestly what the margins are for the next five years. The question is can I get lots of supply? Can it be absolutely safe? And does that 2080 split look reasonable going forward? And I think it does. I want to ask about Tesla. You mentioned them a little earlier. They held an event recently where they unveiled their I'd plans for a a Robotaxi service.
Yes. Do you consider Tesla competitor? Well they certainly could be right if they develop their own AV vehicle and they decide to go direct only through the Tesla app. They would be a competitor and if they decide to work with us then we would be a partner as well. And to some extent again both can be true.
So I don't think it's going to be an either or I think Elon's vision is pretty compelling, especially like you might have these cyber shepherds or these owners of these fleets etc. Those owners if they want to have maximum earnings on those fleets will want to put those fleets on Uber. But at this point it's unknown what his intentions are.
There's this big debate that's playing out right now about who has the better AV strategy between Waymo and Tesla in the sense that the waymo has many many sensors on them. Yes. The vehicles are much more expensive to produce. Tesla is trying to get to full autonomy using only its cameras and software.
And Andre Carpathi, the AI researcher recently said that Tesla was going to be in a better position in the long run because it ultimately just had a software problem whereas Waymo has hardware problem and those are typically harder to solve. I'm curious if you have a view on this whether you think one company is likely or to get to a better scale based on the approach that they're taking with their hardware and software. I mean I think that hardware costs scale down over a period of time.
So sure Waymo has a hardware problem but they can solve it. I mean the history of of computing hardware is like the cost come down very very significantly. The waymo solution is working right now so it's not very right and I think the differences are bigger which is waymo has more sensors as cameras has lied or so there's a certain redundancy there.
Waymo generally has more compute so to speak so the inference of that computer is going to be better right and waymo also has a high definition maps that essentially makes the problem of recognizing what's happening in the real world.
A much simpler problem so under Elon's model the weight that the software has to carry is very very heavy versus the waymo and most other player model where you don't have to kind of weigh as much on training and you make the problem much simpler as a compute problem to to understand. I think eventually both will get there.
But if you had to guess like who's going to who's going to get to sort of a viable scale first listen I think I think Elon eventually will get to viable scale but for the next five years I bet on waymo and we are betting on waymo. I'll say this I don't want to get into an autonomous Tesla in the next five years. I'm going to let somebody else can test that out. I'm not going to be an early adopter of that. That's pretty good. Have you used it recently? I have not used a really good. Yeah.
It's really good. Now again it's the for example the cost of a solid state lighter now is $500 or $600. Right. So why wouldn't you put that into your sensor stack? It's not that expensive and for a fully self driving specialized auto I think that makes a lot of sense to me. Now Elon has accomplished the unimaginable many many many times so I wouldn't bet against them. Yeah. I don't know. This is always you know my secret dream for you know obviously you should say it Uber as long as you want.
When you're done with that I actually do think you should run Tesla because I think you would be just as you've done Uber you'd be willing to make some of the sort of easy compromises like just put a $500 freaking light on the thing and we go much faster. So I have a full time job and I'm very happy with it. Thank you. Well the Tesla board is listening. That's good. I don't know if the Tesla board listens to you too. Good point.
I mean we're opening up the war meeting with an episode of hard fork everybody. I think I learned a lot from this show. What's your best guess for when say 50% of Uber rides in the US will be autonomous. I take close to eight to 10 years is my best guess but I am sure that'll be wrong. Probably closer to 10. What's that? You're okay. Interesting. Most people have overestimated, again it's a while guess the probabilities of your being right are just as much as mine.
I'm curious if we can sort of get into a future imagining mode here. Like in the year whether it's 10 years or 15 years or 20 years from now when maybe a majority of rides in at least big cities in the US will be autonomous. Do you think that changes the city at all? Do the roads look different? Are there more cars on the road? Are there fewer cars on the road? What does that even look like? I think that the cities will have much much more space to use.
Parking often takes up 20, 30% of the square miles in a city for example. And that parking space will be open for living, parks, etc. So there's no doubt that it will be a better world. You will have greener, cleaner cities and you'll never have to park again which I think is pretty cool. I'm very curious what you think about the politics of autonomy in transportation. In the early days of Uber there was a lot of backlash and resistance from taxi drivers.
Sure. And they saw Uber as a threat to their livelihoods. There were some well publicized cases of sabotage and big protests. Do you anticipate there will be a backlash from either drivers or the public to the spread of AVs as they start to appear in more cities? I think there could be. And what I'm hoping is that we avoid the backlash by having the proper conversations. Now historically society as a whole we've been able to adjust to job displacement because it does happen gradually.
And even in a world where there's greater automation now than ever before employment rates, etc. are at historically great levels. But the fact is that AI is going to displace jobs. What does that mean? How quickly should we go? How do we think about that? Those are discussions that we're going to have. And if we don't have the discussions, sure there will be backlash. There's always backlash against societal change that's significant.
Now we now work with taxis in San Francisco and taxi drivers who use Uber make more than 20% more than the ones who don't. So there is a kind of solution space where new technology and established players can win. But I don't know exactly what that looks like. But that calculus does not apply to self-driving. It's not like the Uber driver who's been driving an Uber for 10 years. That's their main source of income. Can just start driving a self-driving way. You don't need a driver.
It's not just that they have to switch the app they're using. It's that it threatens to put them out of a job. Well listen, could they be a part of fleet management, cleaning, charging, etc. That's a possibility. We are now working with some of our drivers. They're doing AI map labeling and training of AI models, etc. So we're expanding the solution set of work on demand work that we're offering our drivers.
Because there is part of that work, which is driving maybe going away or the growth in that work is going to slow down, at least over the next 10 years. And then we'll look to just. But listen, these are issues that are real. And I don't have a clean answer for them at this point. Yeah. You brought up shared rides earlier. And back in the day, I think when Uber X first rolled out shared rides, I did that a couple of times. And then, you know, I don't know.
I don't know. I like got a raise at my job. And I thought, you know, from here on out, I think it's just going to be me in the car. How popular do you think you can make shared rides? And like, is there anything that you can do to make that more appealing? Well, I think the way that we have to make a more appealing is to reduce the penalty, so to speak, of the shared rides. I think the number one reason why people use Uber is they want to save time. They want to have their time back.
And a shared ride would, you know, you would get about a 30% decrease in price historically. But there could be a 50 to 100% time penalty. Yeah. We're working now. You might end up sitting next to Casey Newton. That would be cool. That would be amazing. Although I would feel very short. Otherwise, I don't have no complaints. I people so far we've heard don't have a problem with company. It really is time.
And they don't mind riding with other people. There's a certain sense of that satisfaction with riding with other people. But we're now working with both algorithmically. And I think also fixing the product previously. You would choose a shared ride and you get a front discount. So you're incentive as a customer is to get the discount, but not to get a shared ride. So we would have customers gaming the system. They get a shared ride at 2 a.m.
When they know they're not going to be matched up, etc. Now you get a smaller discount and you get a reward, which is a higher discount if you're matched. So part of it is we're not customers aren't working against this and we're not working against customers. But we're working on tech. We are reducing the time penalty, which is we avoid these weird routes, etc. That's going to cost you 50% of your time or 100% of your time.
Now, in autonomous, if we are the only player that then has a liquidity to introduce shared autonomous into cities that lowers congestion, lowers the price. That's another way in which our marketplace can add value to the ecosystem. Speaking of shared rides, Uber just released a new airport shuttle service in New York City. It costs $18 a person. You book a seat. It goes on a designated sort of route on a set schedule.
I don't have a question. I just wanted to congratulate you on an inventing a bus. It's a better bus. You know, exactly when it's coming, picking up like just knowing exactly where your bus is. Pick up knowing what your path is real time. It just gives a sense of comfort. We think this can be a pretty cool product. And again, is bus going to be hugely profitable for us long term? I don't know, but it will introduce us to a bigger audience to come into the Uber ecosystem.
And we think it can be good for cities as well. If you're in Miami, by the way, over the weekend, we got buses to the Taylor Swift concert as well. So I'm just saying. Well, I mean, look, it should not be hard to improve on the experience of a city bus. Like, do you know what I mean? Why would I have a city bus? What is the last time you want a city bus? Well, I took the train here. So it wasn't a bus, but it was transit. It doesn't take share. He doesn't take bus.
I don't think it's like I like to ride public transportation. You're elitist. No, I would love to see a picture of you on a bus sometimes in the past five years. Because I'm pretty sure that's never happened. Let me ask you this. I think we can make the experience better. Here's how I've, you know, so far I've resisted giving you any product feedback, Dara. But I had this one thing that I have always wanted to know the explanation for.
And it says at some point in the past couple of years, you all when I ordered an Uber started sending me a push notification saying that the driver was nearby. And I'm the sort of person when I've ordered an Uber, Dara, I'm going to be there when the driver pulls. I'm not making this person wait. Okay. I'm going to respect their time. And what I've learned is when you tell me the driver is nearby, what that means is they're at least three minutes away and they might be two miles away.
And what I want to know is why do you send me that notification? We want you to be prepared to not keep the driver waiting. Maybe we should personalize it. I would love that. I think that's a good question, which is depending on whether or not you keep the driver waiting. I think that is one of the cool things with AI algos that we can do. At this point, you're right. The experience is not quite optimized. Because you get it. But it's for the driver. It's for the driver.
And if I were a driver, I would be happy that you were sending that. But you also send me this notification. Those are drivers arriving. And that's what I'm like, okay, it's time to go downstairs. But it's like we're making progress on that. I think I need an algorithm just likes you. Just wants to know that I love my rides. Yeah. Well, Casey has previously talked about how he doesn't like his Uber drivers to talk to him. And this is a man who doesn't share my ride.
Listen, this man likes to coast to his life. I thought so. I would do it. Bubble. Here's what I'm saying. If you're on your way to the airport in his six, 30 in the morning, do you truly want a person you've never met before asking you who you're going to vote for in the election? Is that an experience that anyone enjoys? By the way, I drive. And reading the rider as to whether they want to have a conversation or not, I was not good at the art of kind of conversation as a driver.
Were you too talkative? Hey, no, no. Hey, how's it going? You're having a good day going to work. And then I just shut up. Yeah. And have a nice day. That was it. But I don't know if that's. No, that's perfect. That's going to give you all the information. I'll be your driver. I'll be your driver in there. This case is real attraction to self-driving cars is that he never has to talk to another human. Look, you can make fun of me all you want. I am not the only person who feels this way.
Let me tell you. No, when I check into a hotel, safety, like, did you have a nice day of like, yeah, but where are you coming in front? I'm not going to do it. Yeah, just key. I would love to see you check you into a hotel. So once you do have a nice day and you're like, well, let me tell you about this board meeting. I just went to because the pressure I'm under. You don't want to hear about it. All right. Well, I think we're at time. Dara, thank you so much for coming. Thank you. Thank you.
Thank you. Appreciate it. It was fun. When we come back, well, AI is driving progress and it's driving cars. Now we're going to find out if it can drive Casey insane. He watched 260 TikTok videos and it'll tell you all about it. Okay, see, aside from all the drama in AI and self-driving cars this week, we also had some news about TikTok, one of the other most powerful AI forces on Earth. No, truly, yes. I, I, I unironically believe that. Yeah, that was not a joke.
Yes, so this week we learned about some documents that came to light as part of a lawsuit that is moving through the courts right now as people will remember. The federal government is still trying to force bite dance to sell TikTok. But last week, 13 states in the District of Columbia sued TikTok, accusing company of creating an intentionally addictive app that harmed children.
AI and Kevin, and this is my favorite part of this story, is that Kentucky Public Radio got a hold of these court documents and they had many redactions, you know, often in these cases, the most interesting sort of facts and figures will just be redacted for who knows what reason. But the geniuses over at Kentucky Public Radio just copy and pasted everything in the document and when they pasted it, everything was totally visible. This keeps happening.
I feel like every year or two we get a story about some failed redaction. Like, is it that hard to redact a document? I'll say this. I hope it always remains this hard to redact a document because I read stuff like this. Kevin and I'm in heaven. Yes. So they got a hold of these documents. They copied and pasted. They figured out what was behind sort of the black boxes in the redacted materials. And it was pretty juicy.
These documents included details like TikTok's knowledge of a high number of underage kids who were stripping for adults on the platform. The adults were paying them in digital gifts. These documents also claimed that TikTok had adjust its algorithm to prioritize people. They deemed beautiful.
And then there was this stat that I know you honed in on, which was that these documents said based on internal conversations that TikTok had figured out exactly how many videos it needed to show someone in order to get them hooked on the platform. And that number is 260. 260 is what it takes. You know, reminds me, this is sort of ancient. But do you remember the commercial in the 80s where they would say, like, how many licks does it take to get to the center of a Tootsie pop?
Yes. This to me, this is the sort of 2020s equivalent. How many TikToks do you have to watch until you can't look away ever again? Yes. So this is according to the company's own research. This is about the tipping point where people start to develop a habit or an addiction of going back to the platform and they sort of become sticky in the parlance of social media apps. It's a disgusting parlance of social media apps. It becomes sticky.
So Casey, when we heard about this magic number of 260 TikTok videos, you had what I thought was an insane idea. Tell us about it. Well, Kevin, I thought if 260 videos is all it takes, maybe I should watch 260 TikToks. And here's why I am an infrequent user of TikTok. I would say once a week, once every two weeks, I'll check in, I'll watch a few videos. And I would say generally enjoy my experience, but not to the point that I come back every day.
And so I've always wondered what I'm missing because I know so many folks that can't even have TikTok on their phone because it holds such a power over them. And they feel like the algorithm gets to know them so quickly and so intimately that it can only be explained by magic. So I thought if I've not been able to have this experience just sort of normally using TikTok, what if I tried to consume 260 TikToks as quickly as I possibly could and just saw what would happen after that?
Not all heroes wear capes. Okay. So Casey, you watched 260 TikTok videos last night. Yeah. Tell me about it. So I did create a new account. So I started fresh. I didn't just reset my algorithm, although that is something that you can do in TikTok. And I decided a couple of things. One is I was not going to follow anyone like no friends, but also no influencers, no enemies, no enemies. And I also was not going to do any searches, right?
A lot of the ways the TikTok will get to know you is if you do a search. And I thought I want to get the sort of broadest, most mainstreamy experience of TikTok that I can so that I can develop a better sense of how does it sort of walk me down this funnel toward my eventual interest, whereas if I just follow 10 friends and did like three searches for my favorite subjects, like I probably could have gone there faster. And so do you know the very first thing that TikTok showed me, Kevin?
What's that? It showed me a 19 year old boy flirting with an 18 year old girl trying to get her phone number. And when I tell you I could not have been any less interested in this content, it was aggressively straight. And it was very young and it had nothing to do with me. And it was not my business. And so over the next several hours, this total process I did about two and a half hours last night. And I did another 30 minutes this morning.
And I would like to share, you know, maybe the first nine or 10 things that TikTok showed me, you know, the assumption is it knows basically nothing about me. Yes. And I do think there is something quite revealing about an algorithm that knows nothing, throwing spaghetti at you, seeing what will stick, and then just picking up the spaghetti afterwards, and saying, well, what is it, you know, that it thought was interesting. So here's what it showed me.
Second video, a disturbing 9-1-1 call, like a very upsetting sort of domestic violence situation, skip. Three, two people doing trivia on a diving board, and like the person who lose has to jump off the diving board. Okay, fine. Four, just free booted clip of audition for America's Got Talent. Five, vegetable mukbang. So just a guy who had like rose and rose of beautiful multi-colored vegetables in front of them, who was just eating them.
Six, a comedy skit, but it was like running on top of a Minecraft video. So one of my key takeaways after my first six or seven TikTok videos was that it does actually assume that you're quite young, right? That's why it started out by showing me teenagers.
And as I would go through this process, I found that over and over again, instead of just showing me a video, it would show me a video that had been chopped in half, and on top was whatever the sort of core content was, and below would be someone is playing subway surfers. Yes. Someone is playing Minecraft, or someone is doing those sort of oddly satisfying things like that. This is a growth hack. I'm coming through a rug or whatever.
Yes. And it's like, it's literally people trying to hypnotize you. Right? It's like, if you just see the, oh, someone is trying to smooth something out or someone is playing with a lot of crime. They're cutting soap. Have you seen the soap cutting? Yes. So cutting is huge. And again, there is no content to it. It is just trying to stimulate you on some sort of like lizard brain level. It feels vaguely narcotic. Absolutely. It is like, yes. It is just purely a drug. Video number seven, and add.
Video number eight, a dad who was speaking in Spanish and dancing. I mean, it was very cute. Now, can I ask you a question? Yeah. Are you doing anything other than just swiping from one video to the next? Are you liking anything? Are you saving anything? Are you sharing anything? Because all of that gets interpreted by the algorithm. It's like a signal to keep showing you more of that kind of thing. Absolutely.
So for the first 25 or so videos, I did not like anything, but because I truly didn't like anything. Like nothing was really doing it for me, but my intention was always like, yes, when I see something I like, I'm going to try to reward the algorithm, give it a like, and I will maybe get more like that. So the process goes on and on. And I'm just struck by the absolute weirdness and disconnection of everything in the feed. At first, truly nothing has any relation to anything else.
And it sort of feels like you put your brain into like a Vitamix. You know, where it's like swipe. Here is a clip from friends. Swipe. Kids complaining about school. Swipe. Mickey Mouse has a gun and he's in a video game, right? Those are three videos that I saw in a row. And the effect of it is just like disorienting, right? Yeah. And I've had this experience when you, when you like go on to YouTube, but you're not logged in, you know, and like a new account.
And it's sort of just is just showing you sort of a random assortment of things that are popular on YouTube. It does feel very much like they're just firing in a bunch of different directions, hoping that something will, will stick. Yeah. And then it can sort of, it can then sort of zoom in on that thing. Yes. Absolutely. Now I will add that in the first 30 or so videos, I saw two things that I thought were like actually disturbing and bad. Where are they?
Like things that have never, should never have been shown to me. Was it a clip from the all-in podcast? Yes. Unfortunately, it didn't get that bad. But one, there was a clip of a, of a great in like a busy, busy city and there was air blowing up from the great. And the TikTok was just women walking over the great and their skirts blowing up. That seems bad. That's good. That was in the first 20 videos that I saw. Wow. With this video, okay.
I guess if you like that video, it says a lot about you, right? But it's like that. The second one, and I truly, I do not even know if we are going, we'll want to include this, our podcast, because I can't even believe that I'm saying that I saw this, but it is true. It was an AI voice of someone telling an erotic story, which involved incest, and it was shown over a video of someone making soap. Wow. Like, what? This is dark stuff. This is dark stuff.
Now, at what point did you start to wonder if the algorithm has started to pick up on your clues that you were giving it? Well, so I was desperate to find out this question because I am gay, and I wondered when I was going to see the first gay content, like when it was actually just going to show me two gay men who were talking about gay concerns, and it did not happen. Ever? No. It never quite got there at the, on this morning. In 260 videos. In 260 videos. Now, it did show me queer people.
Actually, do you know the first queer person, identifiably queer person that the TikTok algorithm showed me? Are you familiar with the very popular TikTok meme from this year? Very jameer, very mindful? Yes. The first queer person I saw on TikTok, thanks to the algorithm, was Jewel's LeBron in a piece of sponsored content, and she was trying to sell me a Lenovo laptop. And that was the queer experience, the thank God, in my rock through the TikTok algorithm.
Now, you know, it did eventually show me a couple of queer people. It showed me one TikTok about the singer, ChapelRone, who is queer, so I'll count that. And then it showed me a video by Billy Eilish, you know, a queer pop star. And I did like that video. And now Billy Eilish is one of the most famous pop stars in the entire world. I mean, like truly, like on the Mount Rushmore of famous pop stars right now.
So it makes a lot of sense to me, the TikTok would show me also incredibly popular with teenagers. Right. And so I liked one Billy Eilish video, and then that was when the flight dates opened. I was like, okay, here's a lot of that. But just from like sort of scrolling, it, no, we did not get to the, the gay zone. Now, I did notice the algorithm adapting to me. So something about me was because again, I was trying to get through a lot of videos in a relatively short amount of time.
And TikTok now will often show you three, four or five minute long videos. I frankly did not have the time for that. The longer I scrolled, the shorter the videos were that I got. And I do feel like the content aged up a little bit, you know, it started showing me a category of content that I call people being weird little freaks. You know, it's like somewhat, these are some real examples. A man dressed as the cat in the hat dancing to Sierra's song Goodies.
Okay. There was a man in a horse costume playing the Adam's family theme song on an accordion using a toilet lid for percussion. This is the most important media platform in the world. Yes, hours of day teenagers are staring at this. This one of us so we are so screwed. Yeah, you know, it figured out that I was more likely to like, contact about animals than other things. So there started to become a lot of dogs doing cute things, cats doing two things or, you know, other things like that.
But, you know, there was also just a lot of like, here's a guy going to a store and showing you objects from the store. Or like, here is a guy telling you a long story. Can I ask you a question? Like, was it any, any, in these 260 videos, were there any that you thought like, that is a great video? Um, I don't know if I saw anything truly great. I definitely saw some animal videos that if I showed them to you, you would laugh or you would say that was cute.
But there was stuff that gave me an emotional response. And I would say particularly as I got to the end of this process, I was seeing stuff that I enjoyed a bit more. But there, I did this morning. I decided to do something, Kevin, because I've gotten so frustrated with the algorithm. I thought it is time to give the algorithm a piece of data about me. So do you know what I did? I searched the word gay. Very subtle. Which like, in fairness, isn't insane search query?
Yeah. Because what is TikTok supposed to show me in response? Yeah. You can show me all sorts of things. But on my like, real TikTok account, it just shows me, your creators all the time. And they're doing all sorts of things. They're singing, they're dancing, they're telling jokes, they're telling stories. So I was like, I would like to see a little bit of stuff like that. Do you know the first clip that came up for me when I searched gay on TikTok to train my algorithm?
It was a clip from an adult film. Now, like explicit, unblurred, it was from, and I don't know this. I've only read about this, but apparently at the start of some adult films, before the explicit stuff, there will be some sort of story content, you know, that sort of establishes the premise of the scene. And this was sort of in that vein. But I thought it, if I just sort of set off-handed, you know, oh, TikTok, yeah, I bet if you just search gay, they'll just like show you like porn.
People would say like, it sounds like you're being insane. Like why would you say that? That's being insane. Obviously, they're probably showing you, they're like most famous queer creator. You know, something like that. No, they literally just showed me porn.
So I mean, it was like, I got so much of this process for me was like hearing the things that people say about TikTok, assuming that people were sort of exaggerating or being too hard on it, and then having the experience myself and saying like, oh, no, like it's actually like that. And alternative explanation is that the algorithm is actually really, really good. And the reason to show you all the videos of people being weird little freaks is because you are actually a weird little freak.
That's true. I will accept those allegations. I will not fight those allegations. So, okay, you watched 260 videos. You reached this magic number that is supposed to get people addicted to TikTok. Are you addicted to TikTok? Kevin, I am surprised and frankly delighted to tell you, I have never been less addicted to TikTok than I have been after going through this experience. Do you remember back when people would smoke cigarettes a lot?
And if a parent caught a child smoking the thing that they would do is they say, you know what? You're going to smoke this whole pack and I'm going to sit in front of you. And you're going to smoke this whole pack of cigarettes and the accumulated effect of all that stuff that you're breathing into your lungs. By the end of that, the teenager says, Dad, I'm never going to smoke again. This is how I feel after you're addiction. After watching hundreds of these TikToks.
So, okay, you are not a TikTok addict. In fact, it seems like you are less likely to become a TikTok power user than you were before this experiment. I think that's right. Did this experiment change your attitudes about whether TikTok should be banned in the United States? I feel so bad saying it, but I think the answer is yes. Not ban it, right? My feelings about that still have much more to do with free speech and freedom of expression.
And I think that a ban raises a lot of questions that the United States approached to this issue. It just makes me super uncomfortable with you can go back to our archive to hear a much longer discussion about that. But if I were a parent of a teen who had just been given their first smartphone, hopefully not any younger than like 14, it would change the way that I talk with them about what TikTok is. And it would change the way that I would check in with them about what they were seeing, right?
Like I would say, you are about to see something that is going to make you feel like your mind is in a blender. And it is going to try to addict you. And here is how it is going to try to addict you. And I might sit with my child and might do some early searches to try to precede that feed with stuff that was good and would give my child a greater chance of going down some positive rabbit holes and seeing less of some of the more disturbing stuff that I saw there.
So if nothing else, like I think it was a good educational exercise for me to go through. And if there is someone in your life, particularly a young person who is spending a lot of time on TikTok, I would encourage that you go through this process yourself because these algorithms are changing all the time. And I think you do want to have a sense of what is it like this very week if you really want to know what it's going to be showing your kid?
Yeah, I mean, I will say, I spent a lot of time on TikTok. I don't recall ever getting done with TikTok and being sort of happy and fulfilled with how I spent the time. There's a vague sense of shame about it. There's a vague sense that sometimes it helps me turn my brain off at the end of a stressful day. It has this sort of narcotic effect on me. And sometimes it's calming and sometimes I find things that are funny. But rarely do I come away saying that was the best possible use of my time.
There is something that happens when you adopt this sort of algorithm first, vertical video, mostly short form, infinite scroll. You put all of those ingredients into a bag and what comes out does have this narcotic effect. Does it say? Well, Casey, thank you for exposing your brain to the TikTok algorithm for the sake of journalism. I appreciate you. And I will be donating it to science when my life ends.
People will be studying your brain after you die. I feel fairly confident. I don't know why they'll be studying your brain. But there will be research teams looking at it. Can't wait to hear what you'll find out. Hard fork is produced by Whitney Jones and Rachel Cohn. I'm Ryan Manning and Chris Schott. From The New York Times, you're listening to The Wirecutter Show. Hey everyone, it's The Wirecutter Show. I'm Kyra Blackwell. I'm Christine Circlasette.
And I'm Rosie Garen. And we work at Wirecutter, the product recommendation site from The New York Times. Each week, we bring you expert advice from our newsroom of 140 journalists who review everyday products that will make your life better. Today's episode of The Wirecutter Show is called The Secret to Better Laundry. Hey y'all, it's our very first episode. Hurray! Love it. I'm an augure episode. Should we tell people a little bit about who we are and what we're doing here? Yeah, absolutely.
So, Christine, what is Wirecutter? What does Wirecutter do? Well, Wirecutter is a product recommendation service. We're part of The New York Times company. We have about 140 journalists who do rigorous product testing. And what we mean by that is they are very serious about the products they... What are we talking about here? So, I'll give a couple examples. We've done a guide to hiking boots. And for that one, our writers tried out 55 pairs of boots and hiked 1400 miles over seven years.
I mean, that's a lot of steps. You know, another example, this one's kind of wild. We did a review of FireSafes and our writer actually built a room and burned the room with all these FireSafes inside of it. I mean, that's the kind of lengths our journalists go to test things. Try it and true service journalism. Absolutely. Committed. And we're also independent, which that means we don't let companies pay us to review their products.
And we don't take freebies. Our recommendations are based only on what we really believe is the best stuff. It's a really fun place to work. And I'm excited to introduce listeners here to the amazing people I work with, like Kyra. Hi. Kyra, who are you? So, I'm a writer and I cover all things sleep and I test mattresses. People find it hilarious because I get to say that I sleep for a living. And I've probably tested about 100 mattresses at this point for work.
I've been at wire cutter for four years, but Christine here has actually been here longer than me. That's right. I'm OG. I've been here 11 years, nearly since the site was started. And I'm an editor. How about you, Rosie? Who am I? Yeah, who are you? I'm the show's executive producer. And I'm kind of going to be your producer's sidekick here to pose the questions that, you know, us normies have. Well, we're so happy to have you here.
Oh, thanks, Kyra. So to that end, why are we all here? Let's tell the people. I mean, what are we doing on the show? What's the plan? Look, we get tons of questions from people about problems they encounter or things that they're just trying to hack or solve. And our staff has a lot of expertise to share. Right. So we want to bring all of that to audio every week. We're going to sit down with a wire cutter journalist and talk through a topic that we think people could use some advice on.
We'll give product recommendations like we always do on the site. But this is going to be a lot more than that. We're going to talk through solutions, ideas, hacks, and ways that you can really make your life better. And sometimes we'll even invite outside experts to come on. I love it. And that's why I think today's episode is so good to start with because it's a problem that we all have. But maybe we don't realize that we have. And I am talking about laundry. Why are you looking at me?
No, you're right. I don't think I ever properly learn to do my laundry. Oh, you really hope my mom's not listening. Oh, Jerry. Rosie, I'm going to make the confession that I too don't think that I know how to do my laundry. It's like such good company. Yeah, I mean, I'm a lady. I'm a grown lady. You are a lady. I don't know how to do my laundry because I'm a lady. Yeah. And someone else should be a measure for me.
Yes, yes. The issue with like the entire time as an adult, I feel like, is that I've just never really been able to figure out how to get stains out. And so all of my entire family is just always walking around with like something on their clothes. The whole process, I will say, is very intimidating. Yeah. It's a myth. Don't even ask me. I don't know. It's a myth. My mom taught me how to like separate and lights from darks. And it's not the 1920s. I feel like we don't need to do that anymore.
And that's where I'm at right now. That's a hot take. But there's a lot that I don't know. Yeah. And you know what? That's why I'm really glad we have our special guest today. Because we are wire cutter. We do happen to have someone on staff whose entire job it is to think about laundry, test laundry solutions, and basically give great advice about laundry. Andrea Barnes is a staff writer and industry expert in all things laundry.
And she's going to join us to tell us the best laundry products and how to use them properly. Can't wait. And so listeners know we're going to be dropping links to the guides and products we talk about in this episode and all our episodes right in the show notes. So if you hear us talk about something and want to know more, just head over there. Does tide live up to the hype? Do we need to use hot water on our dirty clothes? Can anything get cooking oil stains out of my favorite t-shirt?
We'll find out in just a minute. We're back with Andrea Barnes. Andrea is wire cutter's premier expert on big home appliances like dishwasher and washing machines. But we're not even really going to focus on that today. All of her appliance testing has also let her to become our expert on all things laundry. She's written a ton of our most popular guides and articles in the category like for detergent, washing your tennis shoes, washing things by hand, and stain removal. Hi Andrea.
Hi, it's nice to be here. Nice to see you. So Andrea does a ton of laundry and tests out different washers and dryers and obviously detergents. And the wire cutter office is this just massive space in Long Island City. If you know there's like weird clothing hanging everywhere that's stained with like egg and lipstick and grass and obviously like lots of detergent all over the place on carts. And I don't know. It's like this weird, willy wankos chocolate factory of laundry over there.
I want to know what you're doing because I feel really undereducated on the topic. But first, you're testing our laundry IQ today with a quiz so you can judge our baseline on how much or little we might know about what's coming up. It's really like how little we know. We're going to come in humble. Yes, I have a little true false quiz for you. So question one, true or false, soda water is the best thing to remove a fresh red wine stain. True. I am going to, I was so enthusiastic about that.
So somebody spilled red wine all over me at my wedding and I was wearing a white wedding dress and my mom's best friend who's an amazing long dress. She like got it out with soda water. She's just hooked in and she is like, I got this. Is it true? Yes, she was like, we are doing this. Well, I'm glad that that worked, but I would actually say false according to our testing. Oh no, I'm sure that soda water can work, but we found that the best thing for fresh red wine stains is actually white wine.
What? Oh, yes. So you pour white wine on top of the red wine? Exactly. Exactly. It just cancels out. It cancels it out. Well, I mean, I did drink a whole bottle of white wine after that person spilled the red wine on me and the stain did come out so it did kind of work. That always worked. You were right. You believed for different reasons. All right. Question two, what do you got? True or false? A top load washer cleans better than a front load washer. True.
No, it's false. Even though they use less water, actually a front load washer is way better at removing stains than a top load washer. Huh. It doesn't have that little thingy in the middle though. I know. It's really confusing, but it's actually the friction that really makes the front load washer work. It's the friction, not the thingy. Well, yeah. Who knew science? Yeah. Okay, next. Okay. So you need fabric softener, true or false? False. False. You are correct.
It's actually just fat that's usually added to the end of a cycle. Oh. Fat? Yeah, it's fat. We actually used to use it in college to condition our hair. Oh my gosh. I wouldn't say do it now, but it worked really well. The olden times were wild. I can't put some curse guffs in my hair. I just want to be wild. I just want to be wild. Okay. So the last question, in order to get stains out, you need to wash with hot water. This has to be true. No, it's actually not. What if my cat pukeed on my bed?
Especially cat puke should be washed with cold water. Stop it. Oh my gosh. Really? Yes, because it has protein in it and protein needs cold water to be washed out. Wait, so have I just been baking it deeper into my comforter? You may have. Did the smell go away when you did it? The more you know, Tyrone. I would like to think so, but who knows, I have so many candles lit at all times. I mean, more water won't get it out in friction helps, but cold water is definitely more efficient.
All right. Well, I think Andrea proved our point. We have some laundry tips and knowledge that we need to acquire ASAP. Yeah. I think that's about right. Okay. So let's get into it, Andrea. I want to start with getting a better sense of how you even test something like laundry detergent. Oh, that's a great question. You know, initially when we, when I started designing the laundry detergent tests, the idea was to put common stains on t-shirts.
I pulled the wire cutter writers and asked them, what are the worst stains you've had? What are the things that have been hardest to remove? And most of them are pretty typical, like lipstick, ink, cooking oil. And we stained all these shirts. We very quickly realized that it's actually very hard to recreate difficult stains. And that most of the stains that you recreate, especially because they're only sitting for a day or two before you wash them. Almost everything was pretty easy to wash out.
And so when you say recreate, are you taking a bottle of Heinz and just motion it into some new t-shirt? Totally, exactly. And so with that in mind, I called one of my sources, who's a former washing machine developer from Whirlpool, and I asked him, what would you do in this situation? Because everything is cleaning out pretty easily. And he said, you've got to call chlorox and get their standardized test. And he described this swatch with 15 different stains on it that are all the same size.
And chlorox, of course, was not going to give me that, right? But I did find the place that chlorox buys it from, which is a German company that has a dealer in Pennsylvania. So to order these swatches, we actually have to email them and give them the credit card information. And all this, but we can't even do it online. It took forever. But this machine is awesome because it uses the exact same amount of whatever the stain material is.
And it presses them down under thousands of pounds of pressure. So you have a standardized stain set. And the stains are pretty typical stains you encounter in daily life. So there's tea, there's mustard, there's something that's called beta-carotene, which is more like a carrot or a sweet potato stain.
Then there are some that you're not necessarily encountering every day, like there's one that's used engine oil, which, you know, we joked about going to find a mechanic who would give us oil rags, but this seemed like a better solution. But the idea here is not to remove all of the stain. It's actually really just comparative data. What it does is it shows us how good a laudated herjian or stain remover is by how much stain they remove.
So the idea really isn't to see a blank t-shirt or a blank piece of jersey cotton at the end. It really is to see some stain left. I don't want to turn anybody stomach, but I am very, very curious because I think it's common, especially if you have kids. My kids fall and scrape their knees all the time. Yeah, blood. Oh yeah. Yeah, so what do you, I'm, I'm afraid to ask.
I became very good friends with the butcher is at Pisano's on Smith Street in Brooklyn where they sell both pork and beef blood for different recipes. And we, I bought, I think we probably between Launey detergent testing and stain remover testing. I think I probably bought six quarts of blood. Are you on some sort of list? Probably. I know. I've been wondering about that. Do they know what you want the blood for?
I probably just think I'm making some recipe, but I think for testing, we left several t-shirts over night for three or four days after being stained with blood. That was actually one of the best tests, like some things that could get those stains out were good detergents or good standard movers. I'm just seeing like a carry scene. That's a little known fact.
Something like that I personally, I don't know about the rest of you, but I, I just use liquid detergent because it's what I buy and that's what I use. And I don't really do anything else, but I, from reading your work, it sounds like there is a difference between liquid and powder and when you want to use those. So can you talk a little bit about that? Sure. I mean, I think for most people, liquid detergent is probably the best choice.
I mean, there are great products on the market and most people are removing body oils from their clothing. A lot of people aren't actually removing that many stains from their clothing. I mean, honestly, the water itself does a lot of the work too. If you have a job where your outdoors a lot and you are working a lot with clay or mud or particulate soils is what they would be called.
Powder detergents probably a better choice because powder detergent is going to work with those stains better because of the way it's formulated. And also if you just have a lot of stains in your clothes in general, powder detergents a nice choice because most of them have oxygen bleach or some other sort of non-chlorine bleach built into them. So there's a little bit more stain-roving power. And you can use those on all colored fabric.
Yes. Totally different than chlorine bleach, which is not safe to use on all sorts of fabrics, but oxygen bleach is, it's basically hydrogen peroxide. So you're going to want to test a small part of your clothing to make sure that it can withstand the oxygen bleach, but we have yet to run into something unless it's like a really poorly dyed item that can't be used with oxygen bleach. The most favorite oxygen bleach is like oxyclean, right? Yes. And we tested it. We didn't love it in testing.
Wow, that guy was lying to me in all those information. Really great infomercials. Watch how oxyclean unleashes the power of oxygen making tough stains disappear like magic. Oxyclean didn't do great in testing because it takes a really long time to dissolve. It really needs hot water to dissolve.
And it took a lot longer to remove stains than other detergents that we tested. Well, oxyclean's not a detergent, but we tested Tied Ultra Oxy, which is a powder tied with oxygen bleach, and it did a much better job with removing things that oxyclean did. I think I read in one of your guys that the thing that really picks up stains is this thing in it called enzymes, and that's what really eats away at the stains. So you don't really need bleach to get out of this thing.
No, for most stains, you don't need bleach. I say to use oxygen bleach when it's very specific kinds of stains. Like, basically anything that you would use as a natural dye if you were making dye in your own clothing. So like tea, coffee, fruits, things where the stain really changes the fabric versus just food stains, which are pretty topical, and enzymes are really good at getting those. So we can basically just throw out all of our chlorine bleach, and we don't need it.
No, not for your regular clothes, you don't need chlorine bleach, but it's still a good disinfectant. And oxygen bleach is for the stains that are really set in and kind of changing the color of the fabric, right? Like wine or something? Yes. Soaking an oxygen bleach will get the most stubborn stains out, and it's great for things like wine or tea or coffee that really do soak into the fabric and are very hard to get out.
So when you say that those oxygen detergents are basically just hydrogen peroxide, I know there's always like a DIY community and everything. When there's a product, there's a DIY product, but why can't you just make your own detergent? I mean, you could if you want to spend a lot of time making something that doesn't work as well. That's like a lot of DIY and junk. Exactly.
You were mentioning when we were talking earlier, this phrase that is kind of a short hand that sometimes helpful, is it something like likes, likes? Yes. It's a solvent term, like, likes, like. So for example, when you have an oil stain, liquid detergent works really well because of surfactants and liquid detergent behave as an oil, so they are better absorbed into oil stains. So basically the best way to look at it is that there's water stains and there's oil stains, right?
The solvents that you want to use to get rid of those stains are going to be similar in properties. So remember I said that if you're working around a lot of clay or dirt, powder detergent is better. That's why similar, right? So they behave similarly and they absorb with each other better and combine better. So this is why my kids like muddy, dirty clothing never gets back to like zero. Because it's just like I'm only using liquid and so it's not picking up the dirt.
Yeah. I mean, you can try pre-treating with a liquid detergent. I find that works really well, but you might just try powder detergent too. What exactly is the process of pre-treating? So pre-treating a stain is kind of how it sounds. You really are just putting the stain remover or the laundry detergent on the stain beforehand and it's usually somewhere between five minutes to 20 minutes before you put it in the washing machine.
For almost every stain, if you have laundry detergent on hand, you probably can just use that for pre-treating. We recommend a different stain remover, all-purpose stain remover for pre-treating called Amidex. But the reason we recommend that is because it removes other stains that laundry detergent is not always good at getting out. So those two are actually makeup and permanent ink.
But otherwise, I would say you can pre-treat with laundry detergent for most stains and then it is a good hack for laundry. So I don't know if any of you have ever had a marinara stain on your shirt and you don't pre-treat it, you throw it in the washing machine and you come out and you still... I've never done it. I've never not treated my stains. Never. Never have I forgotten. She's lying on her. Don't believe me.
Yeah, so I will admit to you all that when we were doing laundry detergent testing and one of my sources told me to soak oil stains in warm water and liquid laundry detergent, I was able to get oil stains. You know those dark oil stains you get on colorful fabric, then it'll be like the... It's just a slightly darker shade and what I think my daughter's goal in life is to have every article of clothing have these stains.
But I was able to get these stains that had been in three or four washes and dry cycles out from this soaking. So you know, there are a lot of different things you can do for a stain removal. Pre-treating is great, but it doesn't necessarily mean that you're totally in trouble if you still have stains coming out of the wash later. Got it. So I've heard you mentioned Tide a couple times.
Why, and I think both of the powdered detergent, which is which, which is that again the one that we like is... Tide ultra oxy. Tide ultra oxy, and then the liquid we like is Tide free and gentle, gentle and free. Gentle and free. Why? Why? Why do we like Tide? Why is Tide such... I mean like we're not shilling for Tide. There is no Tide commercial on this podcast, but what's the deal?
You know, it's interesting. I really went into testing wanting to not recommend Tide, because I had a lot of misconceptions over being allergic to it. And then the more testing that I did and the more research I did, I learned that being ingredient that most people are allergic to in laundry detergent, it's called MI or MCI. It's a chemical preservative. And it's in even quote unquote natural liquid detergents, almost any all of the ones we tested contain it.
And what's interesting is that people perceive that that's what they're allergic to. One, almost all of its rinse off by your washing machine. But the other is that it's the same preservative as probably in your shampoo, your shower gel, anything that you're using that needs to be preserved. So that's actually probably what's causing your allergy. And we talked in multiple dermatologists about this. So that said, we partly chose Tide free and gentle because it doesn't have that preservative.
And so they had all this money and all this R&D to go into creating a laundry detergent that lacks these allergens. It was the combination that had hypoallergenic detergent with the fact that it removed stains so well that made it go on on top. So in the end, Tide just kept, they're doing something really good there. They have a big budget, right? They have a big budget to spend on R&D, which is something that we find across categories at Wirecutter that we're kind of sometimes surprised.
But the big companies do come out ahead because they just have so much money. I'm kind of relieved when the big names are actually doing something right because then you're going to get it anywhere. If you run out, you don't get very accessible. Like meesh stuff sucks. Okay, so Wirecutter recommends Tide free and gentle liquid detergent for most laundry, for really stain stuff, dirt stain stuff, Wirecutter recommends Tide Ultraoxy.
And I trust these recommendations, frankly, base off all of the blood testing and stuff we discussed. Ooh, yes. You know what? I think we're learning. We're doing good, guys. Okay. But I do have some like graduate level questions I needed to throw in here. Like about dry cleaning and pause and just more questions about stains. I'm how to make eco-friendly choices at the laundromat. But first, a break. Welcome back to the Wirecutter Show.
Today, we're getting a masterclass in laundry tips from Andrea Barnes. Andrea is Wirecutter's staff writer on large appliances and all things laundry. So, so far, we've covered some detergent basics. But now we're going to get into some of the nitty-gritty details, literally. Okay, so what if you have really nice or vintage items like cashmere or silk? Do you have to get those dry cleaned? It's really going to depend on the item, but I would say the vast majority of the time know.
First of all, you most likely can use regular detergent, tide-free and gentle on a lot of your dry clean-only items if you use a mesh bag and a gentle cycle. I wouldn't do this with something that you really love. But there are a few other options. We recommend a hand-wash detergent called Soak, which is really great. And what you can do is actually hand-wash your garment and you soak the garment with this detergent, which is called Soak, and that's confusing.
And it's a no-rins detergent, so you don't have to handle the clothing for that long. So you submerge it for 15 minutes and then you just take the item out and press the water out. And it's ready to go. The detergent evaporates really quickly. And so you don't have to rinse it. So what if you have allergies? Isn't there something on it that will... So that's a great question. The preserve we talked about earlier is not in this hand-wash detergent, which is good.
And then we interviewed the owner of the detergent company and they did third-party testing that showed less than 0.005% of the residue is left on the... Because you also use a tiny amount. Like we're talking maybe two teaspoons for several gallons. So there's not much left to begin with. I would say that that's ideal for... You know, we feel the Casimir sweater that you've worn a bunch of times and it hasn't really been stained, but you want it freshened up. That would be a great option.
If it's something that's really... You got stains all over it or you bought a Vintagitam as is. What we learned is that the best thing to do is to use pure sodium per carbonate, which is oxygen bleach and soak for hours. So in that case, we wouldn't use something like tide because there's builders and fillers in it that you won't necessarily want on your clothing.
But pure sodium per carbonate is great for soaking your vintage items. And the product we really like for that in testing is called Restoration. And it's pure oxygen bleach. There's nothing else in it. So OxyClean has fragrance and other fillers in addition to oxygen bleach. And this is just oxygen bleach. How do you test these? How are you testing these vintage? So for this, we did... I'm not going to give you my keyword search, but... I don't think I want my keyword search.
But I looked for items on eBay that are stains. And we ordered a lot of used linens that looks like they had no hope left. And the craziest part... So we got this huge duvet bag filled with tablecloths and napkins and linens. And they had stains. But the best part, slash worst part, was that we realized when we opened the bag, when we un-zipped it, that the person was definitely a smoker. And it was smelled like stale cigarettes. Like really... Yeah.
Not a lot of competing bids. What did you say? What infected in 20 seconds after we ordered it? They were like, it's at your door. Check outside. We were praying you would buy this. So we found this and it was all delicates like lace and linen. Again, I called a source. This time I called our source, Miki Evans, who's amazing. And she is the assistant wardrobe supervisor at the notebook, the musical on Broadway.
And she told me about restoration, this oxygen bleach product that you can buy for soaking vintage items. So we tried it against, I think, 6 or 7 other oxygen bleaches. And this one was by far the best one. And we felt because it had no fillers was what I would call the least risky one. And it took out the cigarette smell. It did. And we tested a bunch of hand wash detergents with this lot of smokey linens. And there were definitely some that did not take the stink out.
Was that your keyword search? Smoky linens. Smoky linens. Smoky linens. Sounds like an eyeshadow color. Smoky linens. So one thing we actually haven't gotten into is laundry pods. Oh, yeah, we got to talk about laundry pods. We got to talk about the pods. So Andrea, what's the deal with laundry pods? I initially didn't really want to recommend pods because you can't pre-treat with them. It's so concentrated that it actually doesn't absorb well into stains for pre-treating.
So you have to add water and you're dealing with pod films. So all these things make it not a great choice in my opinion for pre-treating. But when we started having paid testers come to the office. And let me just interject. Yeah. Paid testers are people who are not on our staff that we bring in to test with us. And basically they're folks who have a variety of different abilities and body types.
And we like to bring them in and get their feedback to get a wider diversity of opinions about the products we're testing. Yes. And some of whom have limited mobility and limited grip strength. I observed very quickly that pods were absolutely 100% the best option for them when operating either washing machines or dishwashers. And so it made sense to me to make a recommendation for that. Yeah. That does make sense.
I actually had a little revelation because I don't have washer or dryer in my unit, but my partner does. And whenever I go over there, it's like I've never done laundry ever in my life. I will just toss a pod in the little compartment that they have in those washer machines. But I've learned recently that you're not supposed to do that. Why is that? It totally depends on the washing machine. But the way the dispenser works is that water goes through the whole thing.
And if it's not a very strong stream, it can be very hard for the pods to dissolve. Basically, you need the water to be strong enough to pierce the pod and start and start everything moving. Got it. All right. Do we have a rule? Okay. You talked a little bit earlier about temperature. And you told us earlier that you're actually supposed to be using cold water. But is there any time that you should be using warm water or hot water? I would say hot water only if you're sanitizing something.
Like if you had sheets from someone who was sick, or if you had... Like, it's about to potty train, for instance. We'll see. Now, pee stains, I would say you could do in cold water. Because urine is pretty sterile. But like a virus or something like that. Yeah. Yeah. Exactly. I would say hot water is appropriate then. But you know, one thing people always think they see blood and they assume, oh, I should wash them in hot water. And that's actually a great way to stain your clothing more.
You absolutely need cold water for removing blood. Oil stains need warm water to kind of melt the fat. Yeah, exactly. Well, the warm water emulsifies it, right? Yeah, and lifts it. What about like preserving... I mean, I know because I wash a lot of fabric to sew with. And I know that like I sometimes wash stuff in hot water. But I'm wondering, does that do anything to the color long term? If you do wash in warm or hot water? Warm and hot water can definitely degrade and fade. Dies faster.
Since I've switched to doing cold water washing, I don't see fading on my clothing in the same way that I did before. What about odor? Like I just... I know that you're saying like pee is okay. Even blood is okay. But I just feel like the warm water will take away that smell. Is that totally wrong? No, that's not wrong. And that's why we recommend what we do. Because the liquid detergents we recommend removed odor even in cold water. I see. How do you do the odor testing?
Well, first of all, my informal odor removal testing is having a teenage son. Oh yeah. That would do it. But formally, what we did is we burned bacon and used burned bacon grease, which really smells like really smells terrible. And we stained t-shirts with it and then had a panel of people decide which ones were removed most and which ones smelled the worst. And in testing, the picks we made all removed that odor really well, if not entirely.
Okay, now that we have fully graduated, we almost are at PhD level here with steam fighting. Andrea, I'm curious. If somebody is just really concerned about the environmental impacts of what they're doing in their home and they want to make better choices about their laundry, what are the things that they should be doing? That's great. The number one thing I would say to anyone is to wash and cold water when you can. Cold water is just a better choice in terms of energy efficiency.
So washing machines have internal heaters that warm the water. And when you wash and hot water, almost all of the energy that's being used in that wash cycle is to heat the water. So when you wash and cold water, you don't use that energy. So that's probably the best thing you can do. I would say air dry when you can. So get a drying rack. Pre-treat stains so that clothing lasts longer. And I would use less detergent when you say less detergent. Yeah, how much do you mean?
Yeah. You know, this is interesting. This is tough question because now so many laundry detergent companies are coming out with hyper concentrated detergent. But I would say two tablespoons for like a big load. If it's not heavily stained, yes. You know how they give you those caps with the little measuring thing on top? And it's like literally a cup. In college, I think I was putting two and a half cups. Why is my clothing slimy? Oh, yeah, that is what it would be. Yeah, you'll get residue.
And you just don't need that much. It's just unnecessary. And that is how you get allergies. Because there's it'll be right up against your skin. Londry detergent sheets. No, okay. They didn't clean well. You might as well just wash with water. Cut it, waste. Okay, but what I'm taking from this is that essentially the best things that you can do from a sustainability lens are also kind of just what's good for your laundry in general.
It's going to be the stuff that let you know get stains out the best. It's the stuff that will keep your clothing nice for longer. And so it's all kind of it kind of like what's good for your laundry is good for the environment too in that way. Yes, I would agree with that. And the detergent is this was army knife detergent for everything. Use it for everything. Cold water. Andrea, this has been great. Thank you so much for coming. Before we go, we've got one more question.
And we're going to ask this of all of our guests. What's one thing that you've recently bought that you really love? It has nothing to do with laundry. But I... She has other interests guys. No, you're breaking my heart. I've bought my husband that the girl rescue girl brush that's recommended by our kitchen team. And he really loves this girl brush. And because he loved it, I would say it's the coolest thing I've bought recently. Aww, that's really nice. If you like it, I love it.
Andrea, thank you so much. This has been so great. Thank you so much for having me. Guys, I feel like there are so many takeaways from this segment. I for one am going to go on the hunt for some powdered detergent. Oh, yeah. I actually ordered some on Amazon.com because that's where... Can you repeat that? Amazon.com. I think it's the only place that I could find it. I thought they only sold books. No, that's great. Okay, cool, cool.
I think my takeaway was washing cold because apparently I've been baking my cat's puke directly into my bedding. Yeah, that's cool. You're going to win that argument with my husband because he's been doing that. So that's a win. But to that point, I also think like you don't need to hunt around for like eco-friendly detergents necessarily. If you're trying to be more, you know, earth friendly, it's really just like energy is kind of the biggest environmental impact.
So washing in cold also solves that. Totally. And then like you don't have to go out and hunt for a fancy pre-treatment, like you're a laundry detergent that you already have will do just fine. Use what you got. That's right. That's right. That's right. I think that like you really can just get like two detergents essentially, a liquid and a powder and you're probably like 95% of the way there. Perfect. Andrea was a star. She was amazing.
I really hope that I can apply this, you know, when I have a machine in my apartment again. I really hope I can get the dirt stains out of my kids clothes finally. God speed my friends. And that's it for us this week. If you want to find out more about Wirecutter's coverage on laundry detergents or snack the products we recommended today, go to nytimes.com, Backslash Wirecutter or find a link in the show notes.
So if you like the Wirecutter who, which we all hope you do, right, I think people are actually going to listen to this. I hope so. I hope people are listening right now. And if they're listening, maybe they'll follow. Yeah. Or leave a review. I hope the nice one. We'll always read the reviews. Even if it's me, I guess. It also helps other people find the show. Yeah, for sure. Yeah. Thank you for listening either way.
The Wirecutter show is executive produced by Rosie Garen and produced by Abigail Keel, editing by Abigail Keel. Engineering support from Maddie Messiello and Nick Pittman. Today's episode was mixed by Daniel Ramirez. Original music by Dan Powell, Marion Luzano, Alicia by Etoupe and Diane Wong. Wirecutter's deputy publisher and interim general manager is Cliff Levy. Ben Fruman is Wirecutter's editor in chief.
Special thanks to Annelle Chetherpoo, Paula Schumann, Nina Lassam, Somi Hubbard, Jen Poyant, Jeffrey Miranda, Sam Dalnick, Julia Bush, and Katie Quinn. Can we rotate who scats each week? Oh my god. Damn, damn, damn, that's what you think. Oh dear. Do do do do do do do do do. Oh, sabotada. You won.