We're sunsetting PodQuest on 2025-07-28. Thank you for your support!
Export Podcast Subscriptions
cover of episode Gabriel Weinberg: Practical Steps to Safeguard Your Data and Identity Online

Gabriel Weinberg: Practical Steps to Safeguard Your Data and Identity Online

2019/5/14
logo of podcast The Knowledge Project with Shane Parrish

The Knowledge Project with Shane Parrish

AI Deep Dive AI Chapters Transcript
People
G
Gabriel Weinberg
Topics
Gabriel Weinberg: 本期节目探讨了数据隐私、网络安全以及如何帮助孩子更好地思考等问题。Weinberg 分享了强大的思维模型,例如“灰色思维”和“预见性功能”,帮助人们做出更好的决策,避免认知偏差,并找到问题的核心。他还揭示了公司在线追踪用户数据的方式,并分享了一些简单的实用步骤,帮助用户在不牺牲便利性的情况下保护数字隐私。他认为搜索引擎无需追踪用户即可盈利,基于情境广告即可实现更好的用户体验。他还讨论了政府在数据隐私中的作用,以及如何打破数据垄断,促进公平竞争。他认为隐私是一项基本人权,应该得到保护,并建议人们使用 DuckDuckGo 等隐私工具来保护自己的数字足迹。他还分享了如何与孩子进行关于成人话题的讨论,培养他们的批判性思维能力。 Shane Parrish: Parrish 与 Weinberg 讨论了在线隐私问题,包括搜索引擎如何追踪用户数据以及由此带来的负面影响,例如过滤气泡效应和数据泄露。他们还探讨了政府在数据隐私中的作用,以及如何平衡个人隐私与国家安全之间的关系。Parrish 还对 Weinberg 的思维模型和育儿方法表示了兴趣,并就相关问题进行了深入探讨。

Deep Dive

Chapters
Gabriel Weinberg discusses the evolution of privacy awareness and the harms of corporate and government surveillance, emphasizing the importance of privacy as a fundamental right.

Shownotes Transcript

Translations:
中文

You don't need to track people to make money because it's all based on the contextual advertising. So the better user experience would be like not tracking anybody at all. Hello and welcome. I'm Shane Parrish, and this is The Knowledge Project, a podcast exploring the ideas, methods, and mental models that will help you learn from the best of what other people have already figured out. You can learn more and stay up to date at fs.blog.com.

We also have a newsletter. It's called Brain Food. It comes out every Sunday. It's short and sweet. It contains recommendations for articles we found online, books we're reading, quotes we find interesting and worth thinking about, and so much more. It's all signal, no noise, and it's one of the most popular things we've ever done, and it's free. You can learn more at fs.blog.com.

That's fs.blog slash newsletter. Most of the guests on this show are subscribers to the weekly email as well, so check it out. On the show today is Gabriel Weinberg, the founder and CEO of DuckDuckGo and author of Superthinking, a giant book of mental models.

We're going to explore data privacy, touch on a bit of cybersecurity, go deep on mental models, and explore how you can help your kids think better. You don't want to miss this. It's time to listen and learn. Before we get started, here's a quick word from our sponsor.

Farnam Street is sponsored by Metalab. For a decade, Metalab has helped some of the world's top companies and entrepreneurs build products that millions of people use every day. You probably didn't realize it at the time, but odds are you've used an app that they've helped design or build. Apps like Slack, Coinbase, Facebook Messenger, Oculus, Lonely Planet, and so many more. Metalab wants to bring their unique design philosophy to your project.

Let them take your brainstorm and turn it into the next billion dollar app from ideas sketched on the back of a napkin to a final ship product. Check them out at metalab.co. That's metalab.co. And when you get in touch, tell them Shane sent you. Gabriel, man, I'm so happy to be sitting here with you. I'm happy to be here. When did you know that you wanted to go to MIT?

So I grew up in Atlanta, which the closest thing I had to kind of the engineering world was Georgia Tech. I went to like a engineering summer camp there. And I thought at an early age, I wanted to be like a computer engineer. But I honestly had no

knowledge of the different schools. So I think I had an irrational want to go to MIT that didn't really understand the landscape of the schools. But around like in maybe freshman year of high school, I had my mind set for some reason on going to MIT. And that was the only school you applied to? So I applied early to MIT and got in early. And then I was going to apply to a bunch of other schools, but I just stopped applying because I figured I'd just go.

And you, you did a double degree. What was your degree soon? So I was an undergrad in physics and I met my wife there. She was in grad math. And then, um, I started a company right out of school, actually, um,

I graduated early. That's a different story. And did this company for a few years, and then went back to graduate school at MIT in technology and public policy, which is like this interdisciplinary economics, law, public policy degree. MARK MANDEL: You couldn't get enough. DAVID ABEL: Yeah, exactly. Well, my wife stayed there for nine years. She got a PhD in operations research.

So I was already hanging out at MIT. So even after I graduated, I was running my company, like they say, at a library. And I would go audit classes. And her department was in the interdisciplinary area, operations research is like statistics and logistics.

And so there happened to be this other cool interdisciplinary program in her building. So I was like, I'm going to apply here. And I actually ran my second company while I was doing this degree and just still working in the library, that kind of thing. What was the company that you were? So that company, the first company I started was a social, I mean, an educational software company right out of school. The idea was to

Help use the internet to increase student achievement by increasing parental involvement Good in theory 15 years too early That stuff still just happening now, I know you have kids and yeah still hard to get what they're getting in school. Yeah, chronically Oh totally So that's what we were trying to do it with this whole like portal and system for lesson plans and things like that So that ultimately didn't go anywhere

Second company was a really early, effectively social networking company that we eventually sold to Classmates.com. It was kind of the almost antithesis of Facebook and their privacy practices. Like we had a paid business model there.

We didn't collect much information. It was mainly about getting in touch with old friends. But we also just got destroyed by Facebook. And so eventually sold it and then moved on to the current company, which is DuckDuckGo.

And so let's dive into that. I mean, let's start with the privacy issues online in terms of not just DuckDuckGo, but in terms of Facebook. And what should people understand about this that they don't understand about online privacy and like just something as innocuous as a search and what happens behind the scenes?

Yeah, I mean, there's so many aspects of privacy and harms. And I think it's interesting. Awareness has really started to increase over the last two years. But it's all over the map in terms of what people really care about. So in 2013, when our company started to take off, that was a Snowden Revelations. That was DuckDuckGo. Yeah, DuckDuckGo. Do you want to explain that just for a second? Yeah, sure. So DuckDuckGo is a search engine that doesn't track you.

So it's an alternative to Google. And then in the last couple of years, we've expanded. And now we offer a suite of privacy tools. It's one download. It's called Privacy Essentials or Privacy Browser. And it includes private search, tracker blocking, so it blocks Google and Facebook trackers across the web, and encryption, which helps stop your ISP from snooping on you. But in any case...

It wasn't even though privacy like was a main thing for our search engine It wasn't the main attractor to it until 2013 when the Snowden revelations happened, right? And then that was all about I remember that government surveillance, right? So it's like is the government spying on you and

which is now becoming more an increasing thing. China's doing some crazy things. But in the intervening years, corporate surveillance has really spiked. And they're actually related because government surveillance actually uses the data you give to corporations. That's where they're getting most of the data from anyway. And so they're linked.

But corporate surveillance, namely like corporations tracking you all over the web and building big profiles about you, has additional harms. So we're talking about corporations tracking you as you go along the web, not corporations tracking employees. That's right. Effectively, as you browse the web, there are hidden trackers on all the pages that you visit. What's a tracker? A tracker is, so you go to, say, a random site, say the New York Times or something. You expect to interact with the New York Times.

But in reality, there are about 30 other companies sitting on that page that the New York Times embedded into it. Now, some of them are for analytics, like who's looking at this page. Some of them are for advertising, like they run Google ads on their page. Some of them are to help them run their site. But the privacy policy of these companies are generally they can use your data for whatever they want. And so they may help the New York Times in X, Y, or Z purpose, but then they're aggregating all that information

into big profiles that they can go sell right and ultimately this comes in large profiles about you that are being bought and sold kind of in the so something is like simple as google analytics can then be used to affect your search engine ranking because google would know that the the page on the site is popular and therefore it could bump it up yeah and so there's a few kind of negative effects of that one is all the creepy ads that follow you around

is one that people notice. A second one, the one that you're just mentioning, is called the filter bubble. So when you're on search results, say on Google, or the same thing happens on Facebook News Feed, or even things like Netflix and whatnot, they're using your search history or your profile information or your browsing history to show you certain things they think you're going to click on.

But by doing that, they have to hide things that they don't think you're going to click on. And that can end up being opposing viewpoints. And we've done several studies on political terms. And people in different areas are seeing completely different results when you search for things like gun control or political candidate name or something like that. And effectively, especially in search engines, your expectation is you're seeing the results, but you're not getting the results necessarily.

And it effectively biases you when you're doing research. That's one harm. The other harms that people are justifiably really upset about are things like data breaches, which relate to identity theft. And so people would just like to reduce their digital footprint online, kind of block all this extraneous information that's getting out there about them. And that's kind of what we allow them to do. Do you think that everybody should have the same search results?

So, yeah, I think by default. Because with DuckDuckGo, that's what you use. Yeah. So on DuckDuckGo, we don't have a filter bubble. And so if you search for the same topic at somebody else at the same time, you'll get the same results. I think that things can be made opt-in. So, like, we have a region setting where you can say, I want more Canadian results, if you want. Everybody wants more Canadian. Yeah, exactly. Yeah.

Or do they? You should opt into something like that, right? And so there are some notions like that, but I don't think it should be personalized in the way it is without you opting into it. Do you think we're moving to a world where the least of our worries is sort of like personalized search and we're moving into more personalized content where you have almost the same fundamental content, but now you can shift perspectives based on if you're

Democrat or Republican and you can sort of play into that or is that too sci-fi? Um, say that again. I mean, I think that's generally the danger of that. I, that I see coming is I, the correlation causation piece is hard here, but, um,

Polarization has just increased at least in the US and it's relatively, you know correlated with all of this social press social media and tracking and so the thought is at least they're interrelated in the sense that People are just in their own bubbles, you know, they're not really seeing these opposing viewpoints its exposure to opposing viewpoints enough probably not but it's like a Necessary but not sufficient condition right if you don't have any you're definitely not gonna sorry engage with the other side and

And so aside from the annoying ads that we see that follow us around, is that the Facebook tracking pixel sort of? Yeah. So on all these sites that are lurking behind web pages, right? If you go and look and analyze which companies are there, Google's on about 80% of the top million sites. And a lot of that's Google Analytics, but also the Google ads and stuff like that. Facebook's on 25%.

And that's this Facebook pixel where it could be the Facebook login or just sites run Facebook ads or they want to just track what audience and then serve ads on Facebook. And so Facebook and Google are the biggest trackers. I remember last year we took the Facebook tracking pixel off. Good for you. Because I was tired of people targeting our audience. Yeah. And I was like, oh, that's really weird when somebody pointed it out to me. And then I was like, okay, we'll just delete that. That'll be.

That'll be, that's the, that's the key thing you're getting at is like you put it on for your purposes, but it can be used for everybody. Yeah. That's the real problem. If it was, if these companies are just a service provider for you, that would be fine. But it's that the data is leaking beyond the use that you set up for it. Yeah. I remember the first time I went into my parents' Facebook account and tried to create an ad targeting Farnham street readers. And I was like, what? This is crazy. Like,

My mom is not connected to Farnham Street at all, so it was a little weird. What do you think, sort of like, how do you think this plays out over the, like this Snowden stuff came out, which is more about sort of like what governments were doing to protect themselves. Actually, maybe that's a good discussion. What do you feel the role of government should be in data privacy?

So we're having kind of a moment now in the US at least after. So what happened is Europe kind of took the lead on this.

And they made something called GDPR, the General Data Protection Regulation, which is actually a restatement of something else they made in 1995. They've been kind of advanced at this for decades. But they're really putting in some rights and saying, privacy is a fundamental right that should extend to online. And you should have the right to know what companies are tracking you and know what information they're collecting you and have a right to opt out of them sharing it.

Among other things, it's a long regulation. So then California, in the U.S., there hasn't been a general privacy regulation ever. There's some specific ones like there's HIPAA for...

medical data, and there's stuff for financial data. But there's nothing general like that. So California, through a ballot initiative last year, passed the first one, which goes into effect in 2020. And that set a clock for the federal government to either make their own and preempt California or have to deal with California's law. And so that's kicked off this process that's happening right now over the next nine months before that law goes into effect.

to think about what federal privacy legislation should look like. And I think, to answer your question, that yeah, I think the government needs to play a role here because, I mean, my general proposition I've been saying for years is if the data can be collected and if it can be used to make profits, it's going to be used for that by people unless the government stops it.

And it's an externality in general. Like it has a lot of data breaches, a good example. All these companies are collecting information about you. They generally have terrible security practices, Equifax, et cetera, et cetera. And eventually they get hacked. All the data gets out there. Tons of people have identity theft. And they didn't really pay to protect the data. It's an externality for everybody. And so the government definitely has a role to clean all this up.

Do you think that the data collected by these companies, and maybe I'll add some context to this, is like the way that I'm thinking about this problem is if you're a company like Google or Facebook or Amazon, you get a lot of data that other companies don't have access to. That data can be used to extend your competitive advantage. It can be used to provide better recommendations that the guy in their garage can no longer do.

Does the government have a role in equalizing the playing field? And what does that mean? And how does it look? Does it mean that data becomes a public good, like a public utility?

or does it mean like how do you see that yeah so you're asking a separate question so independent of data privacy legislation which gives individuals rights over their data there's a whole antitrust competition question going on i actually recently testified in front of the us senate on this notion about general privacy legislation but my message was exactly what you're saying is that

In the world, there's a duopoly in digital advertising. And it's Google and Facebook owns basically all the digital advertising on the web, which relates to all this data profiles. And the more data they have, the better advertising they have, the more targeting they can do. And so they're locking up all the growth in the industry. And that's what's putting out a business, things like media companies, as well as small business ad tech players and other advertisers. No one can really compete with it.

So, yes, I think if that that has a inherent network effect in it, because, you know, they're getting all the data and all the eyeballs and it just is a feedback loop that is getting more and more. And so unless the government breaks that feedback loop, I don't see another way out of it. What I suggested to the Senate was a few things. One is.

So they're collecting all this data from different kind of business lines like Facebook bought Instagram. They bought WhatsApp. Google runs Google Analytics. You know, they run DoubleClick. They run a bunch of other things. The government and Germany is starting to do this already can prohibit them from sharing data across their business units. Right. That's kind of like a lot of financial firms. Exactly. Kind of a Chinese wall situation. That's kind of a problem.

weaker argument, which I think would work of the gotta break them up kind of argument and spin off some of these different business units. And then there are other things that you can do too. But like if fundamentally you don't kind of break up these data monopolies in some way, then I don't think there's, I think it has reduced competition and there's no easy way for other people to compete. Do you think that the solution is

making the data, bringing up the data monopoly over sort of like the monopoly writ large in terms of the size of company and maybe the scope of their influence. It's funny because I think that there are, I think this is where

it gets complicated. I think there are actually different issues. Like, so, um, I think you have to do something about data monopoly, but independent of that, some of the companies have had bad practices within their core markets, right? Like we've had some trouble competing in search, um,

just because of search issues that are specific to search and nothing to do with the data monopoly. And Europe has come out and, you know, find, say, Google a bunch of times related to those practices. So I view them as independent. And generally, they've been independent agencies pursuing them in the government. The government has the reality is these companies are huge. They do a lot of different things. And so the government, different government agencies need to be kind of on top of it.

I want to push back just a little bit on like whether they're independent, because maybe I'm thinking about this wrong and you can correct me. But like one of the reasons that they get so big is because of the data. Yeah, no, no, they're related. OK, yeah. I just think that there are you can address them separately. Sort of. Well, also, if you clean up one, it won't necessarily clean up the other.

Because say you regulated privacy, the data monopoly piece, they're already so big that they can still use their weight to do other monopolistic practices, which would be bad for competition outside of the data monopoly piece. So you need someone else looking at those competitive issues. Do you think that Apple's strategy is to be more privacy focused? Do you think that that is...

a good business strategy? Yes, I do. I think that in general, so people have had this misnomer that like, you know, people say they care about privacy, but they don't really care or they won't do anything. We've been running our own kind of research data on this for years now, really trying to dig into what people are actually willing to do and what people do.

And there's kind of two realities in there. One is as awareness has increased, people's willingness and interest in doing something has greatly increased. And we've seen about a third of people take some actual significant action to reduce their digital footprint online in some shape or form.

Like a third of all people? A third of a subset of people. A third of all people. Our polls are in America, but we think it's related similarly to other major developed countries. Like across socioeconomic, across gender? Yeah. Well, that's a whole other interesting thing is when other people think it's like, oh, young people don't care about privacy and old people care or vice versa. And we didn't find that to be the case either. It's like across all demographics, a decently similar percentage have been taking action.

I don't know what exactly threads them. It's not education. It's not income. It's not age. It's not political affiliation. None of the main factors that we get on like the general survey stuff like spiked. But we consistently see this growing percentage of people taking, trying to take at least actual action. Some of the actions they take, they don't realize actually don't help them, but they did try to take an action. But they're trying to, but they're actually trying exactly. And Apple recognizes this and,

And I think they also view it on principle that, kind of like we do, that privacy is a fundamental right and people deserve it, therefore. And if there's no alternative, then people have no choice. And that's kind of what I was getting at. And so people had wanted to take action. There just was no action for them to take. Right. Right. If you have two options and both of them are bad, you still have to pick one. People want a smartphone, you know.

But now that Apple, say, offering a smartphone that has privacy, those group of people can now choose Apple's smartphone or they can choose DuckDuckGo as a search engine. I think that people need actual tools that work. So if our search engine was terrible, they're not willing to have too much sacrifice. But let's suppose it's equal. And at that point, it's somewhat a no-brainer, at least for this group. You get to switch. You get to reduce your digital footprint. Right.

And you still get all the tools and good search results. MARK MANDEL: OK. Do you think people pay for privacy? Is this the future where Apple, for example, and their integration and their solutions, you tend to pay a higher price for that? Do you think we-- like DuckDuckGo, do we end up paying for that in the future? Or is that-- MARK MIRCHANDANI: So there's kind of two ways to look at this. And so in the Apple case, yes, I think people are paying for additional services that are private.

In search, you're already paying for it in Google already because you're paying with your data. It just doesn't come out of your wallet. Yeah. And in effect, it does because there was a really interesting story the other day in the New York Times about one of the reporters quit Facebook for five months or has quit for five months. And they're kind of recounting their stories for doing so. And one of the things that happened is he started spending 50% less on

Oh, that's interesting. Yeah. Because he wasn't being exposed to the ads. Exactly. And so I think there's two components there. One, he wasn't being exposed to all the ads. Two, some of the ads are just purely manipulative. I mean, if you think about A-B testing, and it's hard to know where this line is, but they're manipulative in two ways. One, advertisers...

hyper target. So they find the exact group that is, you know, willing to buy this thing, whatever it is. And two, they test different messages over and over and over again, different images until they find what is the exact emotional trigger, all these different influence mental models, if you will, of like things that'll actually get you to buy. So some of those things he maybe didn't buy because he just didn't see the ad. Some of those things, um,

maybe he didn't buy because he didn't see the right manipulative message. And so that manipulation piece, which is probably, I don't know exactly what order, but let's call it 50%. That was money directly out of his pocket. Right. And so he did pay for that. He was paying for that. Just, he wasn't, didn't sign his credit card up to them. Right. But some decent portion was because of using that free product. Um,

So one answer is you're already paying. The second answer is for something like Google, if we charge for our search engine, I just think a lot less people would use it. There would be some people who would pay for it. And maybe we'd make the same amount of money, but we just expose less people to it because it is a big friction to take out your credit card, free versus not free.

And so pitting interest almost like wide influence versus like running a profitable business in a way and so yeah So I could see some markets the direct answer question is some markets definitely that already have a precedence of pain Pay a little extra for more privacy. That's fine on a market that is already free. It's hard to see a mainstream provider go paid on it because there's so much friction going from free to paid and

That said, just to be clear, then the question is like, how do we make money? There are advertising in and of itself, this is a good thing to say, is not bad. We make this distinction, this is what I'll say to the Senate, between contextual advertising and behavioral advertising. So contextual advertising is just advertising based on a context of the page. So like you search for something on DuckDuckGo, you search for car, you get a car ad. I don't have to have anything about you to serve that ad. That's different than the ads that follow you around based on all of your history.

That contextual advertising used to be the way the internet worked, and it could go back to that. And if you imagine, like, you're on New York Times again, you look at an article, it's an ad just based on that article. Right. Or an ad based on the video you're watching, the content of the video. Nothing to do with you. Right.

And so that contextual advertising doesn't have all these privacy problems of manipulation and filter bubble and all these things. And my argument is, is basically we've been up this feedback loop of trying to improve behavioral advertising for the last 10 years, but no one's been doing that for contextual advertising.

If we have been doing it for contextual advertising, it might be just as good right now. And in fact, there are some decent experiments of that that have just come out. Like after GDPR in Europe, the New York Times decided to ditch all the behavioral advertising and put up contextual advertising only. And they increased revenue.

This is one of the first data points out of this. And so I think this actually has legs in that you could return to a world of contextual advertising. And that's all DuckDuckGo does. And if you think about it, that's really still what Google does most of the time on Google search. Because all the search ads are still bid on keywords. They're just using your search history in addition to target ads that you on YouTube and Gmail and the rest of the internet. But they don't really have to for the search ads where they make most of their money. It's still all contextual. Why do you think revenue went up at the New York Times?

So I think it strikes me as surprising. Yeah, I think it went up probably because I don't know the whole story, but my guess is they their first party ads now they're selling it themselves. Right. And so instead of using a broker, they're cutting out a whole middleman at that rate. So if you even get near the same advertising rate, then you're just cutting out the 15 to 20 percent you're paying to a broker. Yeah. Sometimes the splits could be worse for, you know, for different advertisers, depending on the network.

I know like we get less money on Farnham street because I mean, we charge a sponsorship, which isn't based on page views and there's no tracking code. There's nothing embedded in there. Uh,

where companies get information on you. Whereas if we use Google AdWords based on page views, we'd probably get more money. But then you tie into this whole tracking contextualization, and it would be kind of creepy to go read an article on metal models and see this book that you searched for last week pop up in the sidebar. Exactly. I mean, for you in particular, there's a good Darren Fireball and John Gruber. He sells his own ads directly. Yeah.

I think if you have a niche audience like you do, you might actually make more money if you start selling them at a high price. Yeah, you should. We've bought his ad before. Oh, cool. I want to come back to the privacy and hardware, and I want to take the role and expand it. So thus far, we've sort of been talking about the individual in the role of the ecosystem and sort of like e-commerce. I want to talk about...

How that role shifts as you think we move up in the stack and by up in the stack, I'm talking like you have a local government, you have a state government, you have a country. And then globally, do you where do you where do you stand on privacy? Should the government be able to break into your phone? Do you think that those companies should make that physically impossible or should the government have a key like walk me through privacy?

Yeah, so there's a lot of nuances in here, but I'm of the view that

Privacy is kind of two fundamental principles, right? Privacy is a fundamental right. So you have a right to have kind of private communications. In the past, that was you could write a letter. You could kind of write things down on paper. Now it's in your phone. I still think you should have that fundamental right. So just to be clear, that's not like a U.S. right? You believe that's a fundamental worldwide right? Yeah. So you should be able to like...

Send an email to somebody anywhere in the world and have that. Yeah, it's not currently the case on email, right? Yeah, but yeah, I think so. I mean this this gets into You don't want to tell other countries how to run their government. So maybe it's too You know, maybe it's crossing some line but at least in the US The idea has always been like you have a private thought space, you know You have you should have private space your home is your home and without kind of

really good cause, no one should invade that privacy. I think if you start invading privacy in that way, it leads to all sorts of bad effects. Such as? Chilling effect is a big one, which is people change their behavior once they feel that they're being watched or they don't have privacy. There's all sorts of good examples of this. To give you, to really back to search,

Right after Snowden, there were two studies that were really interesting. One is People Stop Searching Terrorism-Related Terms.

because presumably they feel that they might have been investigated if they searched. So terms like Al-Qaeda and stuff like that went down. Wikipedia also saw lower amounts of their articles being looked at. That's interesting. I never knew that. Additionally, someone did another same-time analysis and found that health-related terms also went down on Google. So people stopped searching as much for their own health symptoms and things like that. And the thought was, now that they...

aware that governments or their data was generally being tracked. They're worried about it leaking either and coming back to haunt them either from maybe insurance companies or other things. And so it went down. But that is just immediately harmful for people because they're not searching about their health information.

Those are two minor examples, but when you get into things like China and stuff like that, you get into much more major ones and people are afraid to say anything bad about the government and things like that or have public discourse. So I think that you want a private space and able to be able to communicate with people completely privately.

And so I do think that end-to-end encryption, which is kind of what you're talking about like built into iMessage now That was the Apple FBI case You should be able I should be able to send you a text message and no one should be able to read it except the two of us Now the counter argument to that is well, what what if they're sending really bad things and stuff like that? I think that is a bit of a distracting argument for a couple reasons one is

they act there's actually more data value online than ever before so what does that mean like there is also you're putting out all sorts of data online it's much easier to surveil you than it has ever been in the past like your phone's setting out location information which isn't necessarily encrypted um you're even on the text that we send each other there's there's a record that it went through the internet at least right and so maybe you can't read the message but you can probably figure out the metadata that we talked

All this information didn't exist in the past. And so if you look at the overall information law enforcement and government has to surveil people, even with end-to-end encryption existing, it's way higher than it was 10 years ago and way, way higher than it was 20 years ago before. So you believe the government should be able to use metadata? Yeah.

I think we should restrict a lot more. I'm just saying that the general argument of that I need every last piece of information is not a good one in their own argument because they have more information and all the tools that they have exist are much better than they were 20 years ago. No, I think the metadata should be restricted too. I really think we should be able to communicate privately without anyone knowing. But the reality is of the current tools is

That is discoverable. Because when we send a message across each other to the internet, unless we're using even more sophisticated tools, it's figureoutable that we talked. So if the government had the capability-- I'm just trying to think through this. So if individuals could communicate completely privately, governments could communicate completely privately. Well, this is the other point about the key. If you enable a backdoor, which is what some governments are requesting,

It is really naive to think that that backdoor would only be accessed by that government who wants it. And that was one of Apple's major points is that if I build this backdoor for you, US, China is going to use it. And not only that, it's probably going to get leaked and all sorts of bad actors are going to have it. So like, for example, I don't know how familiar you are with any of this stuff, but the NSA made all sorts of hacking tools. Those eventually got leaked.

And they're all floating around the internet now. Tools that the NSA made for themselves are now available to all sorts of other people. And so all this stuff has a tendency of coming out and being used and exploited if it exists. And that's Apple's point. It's like the only way to protect against that is to have no backdoor.

So do you think that that will be fundamentally impossible? I mean, so I think in the San Bernardino case with the iPhone, Apple basically said it wasn't possible. Then they admitted it was possible, but they weren't going to do it. Do you think we move to a world where they actually want to be able to say it's impossible for us to do it?

Um, I think that would be ideal. I don't know how possible that world actually is because you have to manufacture the phones and things like that. Right. Um, but it's become harder and harder, like built into hardware and things like that on the internet wise. The closest thing we have to that is Tor. Yeah.

You know, you can basically go into... Can you explain Tor? Yeah. You can basically go into a special browser that kind of works like it did the movies where it bounces off things before it gets to the other person. And so it makes like three hops into random places before it gets to the other side. Right. And so if both of us are using Tor to communicate, it's extremely difficult to break that.

And as far as we can tell, Tor itself isn't really broken. There have been some exploits in the Tor browser, like using Flash. But things like the news organizations now, to give anonymous tips to them, they use something called SecureDrop, which works over Tor. And it's worked decently well.

And so that's what you need to do in the non-hardware world to communicate. You can do that on any device. How do you think about government's role, not with their citizens, but government's role in terms of the world? Walk me through your thinking. I don't want to use specific countries, but should one country be able to eavesdrop on another country that they feel might pose a significant threat to their citizens?

I mean, I don't think there's any stopping countries from doing that. They've done that from the beginning of time. And so they're going to have an incentive and desire to have espionage agencies. I don't think there's much possible to prohibit that.

Oh, but then the tools exist, right? The very tools that you would use against your citizens, you would be using against foreign entities. They're making all the tools they can in that process. What I'm suggesting is people like Apple should be the opposite of that, or DuckDuckGo, and enable citizens to do the best they can to avoid that kind of surveillance, either corporate or government. What got you so interested in privacy?

Interesting. I'm interested in general public policy of all kinds. So I, you know, you're talking about the mental models book. I'm interested in like everything and have, you know, this degree in general public policy and technology policy. I started this company and kind of backed into search privacy in particular. Um,

I had sold my other company and was trying to find something that I really wanted to do. And I got really interested in search in general and eventually started this search engine that wasn't initially private. I hadn't thought about it. It was more just trying to make better results. And then I got some initial messages after launching about like, well, what's your...

on search privacy and stuff. And so I, myself, out of just interest, took more of a deep dive on it and decided, oh, wow, this is really personal data, arguably the most personal, like what you search. You kind of give your financial problems, personal problems to your search engine. Um,

You don't need to track people to make money because it's all based on the contextual advertising. So the better user experience would be like not track anybody at all. And so just made that unilateral decision. It was just me at the company. It was a one person company. And that's basically how I fell into it.

Funniest side, do you remember when Gmail used to display ads based on what was in your email? So a couple of friends of mine, we figured this out. And what we would do is we put keywords at the bottom of the email, but put them in white text. So people would get ads that would have nothing to do with the email, but we could control what ads they were seeing based on the keywords we were putting there. Very funny. Yeah.

How does search work? I'm really curious on the back end. Like, how does like walk me through we can geek out on this and go into as much detail as you want. Like walk me through how DuckDuckGo builds a search database, how it determines relevance for a search term at a particular time, and then how that relevance changes over time.

So I started out originally doing all the build your own index kind of stuff and then quickly realized that

It is really expensive. And also building your own index is like going out, touching every website, keeping a copy of it, like somehow sort of creating metadata based on the site. I just want to make sure that we're like, everybody understands the terms. Okay. Yeah. Right. Um, I crawl in the whole internet basically. Um, and I realized a few things. One is there were a bunch of companies who raised a bunch of money, all trying to do that, basically compete with Google at their core game and really

realized it was somewhat unnecessary because there were a few different web indexes at the time that existed, and they had reached diminishing returns in quality. And so you could get good web index results from Google, from Bing, which wasn't Bing at the time. It was MSN. You could get it from Yahoo. E-index was at the time. AskG's was still crawling. There were a bunch of ones. And I realized that

It was unnecessary then to focus on that and you could treat it as a commodity and instead try to differentiate on other things I also realized that we were one of the first people to instant answers And so what's happened major in the last 10 years in search has been when you search About half the time you don't click or use the actual you just start web results. Yeah, well not the autocomplete stuff like the um

get weather or product instant answers or sports scores or Local restaurant listing or whatever it is. Those are not full web indexes. Those are vert or Wikipedia information You know, those are very vertical eyes indexes and so No one was doing that when we started. So what I originally figured out was I

I'll just try to not spend all my time on the web index and focus on instant answers. And so the first thing that I focused on was indexing Wikipedia and getting really good knowledge graph answers. And over time, so the other thing I ended up doing, this was way back in 2007, there was a lot of spam in web search back then. There was like all sorts of content forums and things like that. So we did crawl the web and looking for spam results and then remove them from the index. It was like a negative index.

MARK MANDEL: In any case, what's basically happened is we're a small organization still. We're like 65 people. And now it's gotten so much more expensive to crawl the web and make that deep index. It's probably hundreds of million dollars a year. MARK MIRCHANDANI: Just to maintain it. MARK MANDEL: Just to maintain it. Arguably, maybe a billion.

Most of the indexes went away. And so what we've decided to do, also the individual verticals have gotten more expensive. So like Yelp is like the best in class for restaurants and you have Maps. There's only a couple of Maps providers. So we've decided to do instead is maintain privacy at all costs, but try to work with the best partners in all these different verticals. And so then our basic technology therefore is

Getting a query, trying to figure out what is the best instant answer provider and who are all the providers we need to work with to get

to deliver you the set of results. And in any given search, it might be two or three different providers, you know, to like, if they get the web. So it's like Google give you access to their sort of like, uh, we currently don't work with Google at the moment, but there's still, um, Yahoo, Microsoft. We work with Google on YouTube because it's the only place where the videos are. Right. Uh, so video search is another big one. Um,

We work with Yelp and TripAdvisor, and there's literally like 400 different partners because it goes into all the different long-tail stuff. Some of the stuff we still have to build ourselves. So we still use Wikipedia and build it ourselves. You know, like Stack Exchange, Stack Overflow, that kind of stuff. We build our index of that. So we have a bunch of these we actually have to do ourselves. But we also work with all these other partners and then try to, on any given search, figure out, is this a local search? Is this a programming search? What kind of search is this? And then...

give you results back. And then we have to make decisions of like, how newsy is this search? Like where should we display news results? Should it be at the beginning? Should it be in the fourth result? That kind of thing. So you mentioned like indexing Wikipedia and then you mentioned knowledge graph. What does that mean?

So Wikipedia... Like you read this page programmatically on Wikipedia. Unfortunately, Wikipedia is somewhat unstructured. It would be better if it was way more structured. What's the difference? Structured would be like, here is the person's name and their age and everything like that. Wikipedia is...

you know, edited by anybody and it's got all these different rules, but it's basically just a lot of text with some markup on it. And so if you index it, you have to like figure out all the different edge cases for all the different markup and somehow extract, you know, your name and age and all that stuff. Right. How do you do that? A lot of, there's a lot of code. So do you literally just write like for one off exceptions or is it, there's a lot of edge case exceptions. Yeah. Uh,

They've tried to standardize it, but it's not very standardized as well as it should be. Yeah, it's a large code base. But you're effectively standardizing it for them.

in a in a weird way or am i misunderstanding like yeah we're kind of reverse engineering it if you will you know like turning it into a standardized format yeah right for our own purposes and then could you give that information back to wikipedia and like have them address the page at least in like metadata or something there have been some projects like that that haven't gone very well yeah other people have tried there's like a one thing called wiki data and such uh but yeah that's effectively what we do for wikipedia and

And so the graph then is like the author is this relevant dates are this, the, and then how do you pick out, like when you read text through the computers, like what is the, what is the algorithm considering in terms of trustworthiness of the site? What is the algorithm considering in terms of like, how do we find the relevant passage in the text we're reading for the query that you're

It really depends on the vertical. And so like-- Can you walk me through a couple of them? Yeah, so like local search, for example, like restaurants and whatnot.

really focuses on like when you do a local search, you usually have a what and a where clause. So it's like, I want pizza in New York or whatever. Right. That'd be too broad, but yeah. And so like for that one, once we hear those local search, we have to pull, figure out like the what and the where and like pull that out kind of semantically from the query and then do different things with them. Like figure out what the location is. And even that gets more complicated because if you like,

Paris Something in Paris. Are you talking about France? You're talking about Paris like Indiana and there's bunch of Paris's in the US I don't know if I'm right and and same thing with the what like is pizza category or is that a name of a restaurant and so on these verticals you have different indexes and different rules

And so for Stack Exchange, for example, the Stack Overflow as another example, which is programming instant answers, that has a lot of special characters that you have to do stuff with. People type in programming.

and less than, greater than signs. And in other contexts, you would ignore those. But in the programming case... They're super relevant. Yeah, they're super relevant. So what ends up happening is you really need to first decide what domain this thing is in and then apply different rules, different relevancy rules and search rules per that domain and kind of do different things with it. How should we think about...

search moving forward? Like, what is it like? I'm just trying to contextualize a couple of things here. One of which is like an argument that Google would probably make is like, if you're standing in New York and you search for pizza, like you probably want local contextualized results without your permission. Like we'll just give it to you because that's giving you a better answer. Well, one thing about local, which is interesting is you're,

you can do that without tracking people on DuckDuckGo because your location is generally sent over by your browser automatically. And we can also do that contextually and throw away the location after that query. So how does that work? Like your location is sent by your browser? Yeah, so there's kind of... So no matter what browser you're using, Chrome, Internet Explorer... Well, there's some location built into your IP address. So whenever you connect to anything, you're sending your IP address over. And so you can get a rough location based on that.

You also have something called the browser location and sites request, can you use your location with this app or this site? That'll send a more granular location based on GPS on your phone. And so on DuckDuckGo,

If it's a very hyperlocal query, we'll ask for that. Would you allow us to use the location? But we promise to throw it out immediately. We don't store it at all. We just use it to bring you back to local restaurants around where you are, and then it's gone. So there's no location history or anything like that.

And then where is search going in your mind in the next five years? Is it going to voice? Is it going more mobile? Is it going to use search based on images? Are any of these trends that we hear about a little bit in the media, but are they what you're witnessing? So what's interesting about the history of search is those trends are real, but they've never supplanted earlier search queries. So like

the rise of mobile search queries has been great in the last 10 years, but desktop search queries actually haven't decreased. And so it's just like people are searching more.

So that's one thing more of that new share is mobile. Yeah, definitely mobile voice has been slower than I think people predicted But it's definitely a new thing like if people aren't searching less on mobile and moving to voice. It's just like other queries So that's one thing the second thing is it's the the rise of instant answers and just like more instant and

more like not waiting through results has just continued and continued. And I imagine that'll keep continuing. Does that mean that like Google, Facebook, Amazon, to some extent duck, duck, go, you get to play gatekeeper. Like, do you get to pick winners? There's a big gatekeeping role. And I think that's, uh, that's been a bunch of the competitive complaints. Um, you know, um, that's an argument and search right now of like the user experience is arguably better if, uh,

You give somebody one result, but it's a gatekeeping role. How do you think that that should be handled? So what we've been trying to do, but we haven't had the luxury because we don't make our own ones all the time, is try to give people choice. And so when we have map results, for example, people can choose which provider they want to

do the directions when they go off of DuckDuckGo onto a directions provider. And so that's the model I think would be good is giving people more choice. But even that has its own problem because there are hundreds of providers and you can't give, you run into the paradox of choice model just like, or Hick's law, it's like,

You give people a billion choices. Yeah. What if I want to start up a company? Yeah, exactly. Well, then it's like you can't, you give a Dropbox of a hundred things because you'll just confuse people. And so at some point you do have to make choices.

Um, we've been trying to work with the biggest partners that have the most breadth and best results. And so the way that you think about making that choice is like, who's the biggest partner who's got the better sort of results for people. But even then, like if we're not here yet, because it's not like this case in every country, but you can imagine, um,

Like if you were really into Amazon versus Target or something and we give you a choice, would you rather prefer Target results or Amazon results or Yelp or TripAdvisor results? You had like a favorite. Right. You could choose and then be like, okay, see more of that on when you get those type of consent answers. But then we're still like sort of anti-competitive to new platforms.

Yeah. I mean, I think it's inherently a problem with instant answers. So do you think like in a weird way that we haven't had before, and I'm just like thinking out loud and throwing this out there, but do you think like bigger gets bigger? I mean, I think in general, when you have data network effects, you have winner take most markets. Because I'm thinking like I used to, like if I wanted to compete with Walmart, I could open a store and I can compete with them and people would see my store and they would find it and they would try to buy it.

But online, that's not necessarily the case, right? Just because you create a website or you have a better product doesn't mean that you get noticed. No, I mean, there there are there definitely are equivalent ways to get noticed. But what you're basically saying is, which I think is generally true, is your capitalism in general has scale effects in a number of ways, you know, just kind of as a scale, network effects, etc. And that is a feedback loop that makes things bigger unless there's some disruption that happens.

And I think that naturally happens. And there's, there's a good argument to show that that's happened a lot, at least in the U S markets in the past 20 years. So if you put it all, not even online, just even offline and sorts of industries, for example, what comes to mind? So, uh, oh man, I just read this book that was, uh, the myth of capitalism that had all sorts of examples. And now they're kind of like, Don, I'm, uh, I'm escaping me, but there was like from eyeglasses was one of them.

I think it was like oil refinery. Oh, because they're just control. They're effectively like the lenses are controlled by one. Yeah. It just ends up being scale. So like it's all the same effects. Like people get bigger, they buy up other companies. There's been a lot of acquisitions, you know? Right. And nothing nefarious necessarily. Right. But just the concentration. Yeah. Over time, the concentration becomes more and more and more and that creates less innovation. Yeah.

And that's kind of the general argument. That's interesting because as you're talking about that, I'm thinking like one of the big national debates is sort of like this wealth inequality. Yeah. But we don't necessarily talk about it. They directly tie this in that book to the same thing. Okay, yeah. Because I was going to say, we don't tend to talk a lot about like company inequality where you have this big disparity between the sort of

I don't want to call it winners and losers, but the bigger companies get bigger almost as a necessity of them being bigger. They can make more acquisitions. They have more preferable terms with people. They can be front and center. They can buy advertising more than other companies. They can generate revenue.

more data that they can use to make their products better. Yeah, there's a direct tie to inequality there in a couple respects. The biggest one is that a lot of the newer companies, especially online companies, have a lot less employees than the older line companies.

It's taken less there's just more leverage online because digital has no marginal cost generally, right? And so it takes less employees to reach the same market cap, right? And so there's less people benefiting from that right like comparing Berkshire Hathaway Which has like 400,000 employees to Google which I don't even know what they have but it's nowhere near that Yeah, exactly. And then even that the subset of the you know employees who are making lots of money is lower and

Right. And, and that's very different than say like a Walmart who employs like a million people or something in the U S you know? And so over time, the rise of online and bigger companies has created some of the wealth inequality, at least that's the theory. What are, before we move on out of this area, I want to talk about what are some of the things that people can do to reduce their footprint tangibly if they want to do it without downloading anything or like,

DuckDuckGo is obviously one of them. But what are the other things people can do to reduce or increase their privacy online? And let's start there. Yeah. So DuckDuckGo-wise, that's been our goal is to give the one download you need. And so on iOS, Android, Chrome, and Firefox, you can download either browser extension or our browser. And it has all the kind of essentials for search and browse. Right.

In addition, we have a blog at spreadprivacy.com that has a device tip section. And so for any major device you have, it could be laptop, desktop, or phone, there's a set of settings that you should probably change on your iPhone or your laptop that would generally help you.

That kind of helps you on the main search browse, use the internet side. Then there's like all the other services that you use that kind of aren't search browse, like email. And for all these services, they're generally,

like more private alternatives, sometimes really private, but also just getting off of Google and Facebook brand helps you not put all your eggs in one basket. Right. And so for email, there's like proton mail. We have something called fast mail. Proton mail is very private, fast mail.

is not just built for privacy. They're, they're just an alternative paid email provider. Right. You know, and so moving your email off of Gmail on something else, the company that provides search. Yeah. You want to separate your kind of data as best you can independent of all of that. You know, there are a bunch of all these like people sites, you know, that like you look up your name and you see like,

or whatever, age, whatever, you know, you can opt out of most of those. Okay. Um, there are some services that will do a paid, but there are some lists where you can just go through and just request your removal. I'm super happy to have Shane Parrish. Google that. There's this guy in Australia, Dieter Berman or something. It was like a TV star. And so the number one result for Shane Parrish is like this. It's great. Yeah.

That's awesome. Yeah, so, I mean, those are the things. I think if you want to get more extreme, there are other things. Like we talked about using Tor or something like that. But in terms of just like no sacrifice, very seamless, you know, I would use something like Tectico, like a tracker blocker, private search, more encryption, and then tweak your settings. Yeah.

Do you have Instagram and Facebook yourself? No. So I quit Facebook, I want to say, like seven years ago, maybe more. It might have been like 2010. And I don't use Instagram. I don't really use my social media except I use Twitter. Do you think that there's a correlation between unhappiness and the use of social media? So I do. I will say that I have it.

it's just the more mental models, but I really try to go with the thinking gray model of like, not totally making my mind on anything, but all the research that I've seen and haven't like in depth read all the studies, you know, show that, um,

Yeah. Quitting Facebook has had very positive effects for people. So not only the, the, your wallet effect, which we talked about earlier, but just people feel less isolated, you know, they feel less lonely in a counterintuitive way when they quit Facebook. And so I, what explains that like in your mind, like what are the probabilistic reasons that are the majority of like what would explain? Yeah. I mean, I think it would be interesting. So I'm not the expert I've done the research, but,

But my guess is twofold one is there's just an opportunity cost of your time So right like you were just spending a lot of time wasting time on Facebook that time is substitute with something else Which is probably better time spent that makes you happier whether that's talking to other people or you know reading something something else That's better

At least that was certainly the case with me. The second is the actual content that you are engaging with is sometimes pretty toxic. And so like there's kind of two effects of that that have been talked about a lot. One is like you're seeing this kind of Instagram, other people's awesome life effect. You're seeing other people being seemingly happier than you or something like that. But it's often a kind of a ruse because they're posting like the

fake photos are the best things that are coming out of their life. Um, the other is just like, you know, I remember just seeing endless political debates, um, that were very like you get emotionally charged. Yeah. Yeah. And it was just like, it wasn't constructive debates. It was like ad hominem attacks and things like that. Um, and that's just a lot of negativity. And so it's like, you're cutting out negativity. Um,

I think that's always a good idea. You know, it's, it's kind of like the, um, there's another metaphor, like the, uh, you know, you're the closest of the five friends you hang out with kind of thing. You are what you eat in a way. And so if you reduce negativity and replace it with positivity, you're just going to be happier. Do you think we can have, um, constructive online debates and what do they look like? I think so. I think it's possible. Um, do you ever listen to intelligence squared that podcast?

It's pretty good. I only listened to one, the knowledge. Yeah. Fair enough. It's a, they've done maybe 200 of these. It's a structured debate podcast that,

Uh, they have a topic. I've had a bunch of social media ones, like a social media, good for you, bad for you kind of thing. They have four really good experts on, and then they play both sides of it. Moderate debate. It's really good. And it's totally civil discourse. We actually listen to it with our kids a decent amount. How old are your kids? Um, 10 and seven. Uh, and yeah, so I think something like that would be good.

I've worked on a project with somebody else who's been working for a couple of years on this very topic and have been going back and forth with different ways you can kind of do that. I don't think he's really hit the nail on the head yet, but I do think it's possible to your point. Do you think it's possible to have a national conversation about a topic in a constructive way? I don't know. I don't know exactly what that means because it's just so many people, you know?

What we were kind of envisioning is part of the problem is, and there's kind of a bunch of mental models in here, but like people effectively just talk past each other in a lot of these kind of debates. And so take something like climate change.

There's actually like, you know, 10 different debates within climate change, probably more. And I made a flow chart one time to try to isolate, okay, if you're going to argue about climate change. The individual arguments? Yes. Which part of the argument are you in? Like, okay, do you believe it's man-made? Okay, do you believe it exists? You know, if you believe it's man-made, do you believe we can do anything about it? You know, do you believe there's harm? Do you believe it's intimate harm? You know? You're sort of bringing it down into its fundamental... And so like in any particular...

What you really want to do is, or what I try to do at least is, if you really want to have an engaging debate with somebody, you want to start with the premise that there's, suppose that you think you disagree on something. You want to find, break down the argument into premises and find which one you actually disagree on.

With something like climate change, there's a ton of different underlying premises. And so you have to go down and be like, "Okay, do we agree on that? Do we agree on that? Do we agree on that? Do we agree on that?" That's not generally what happens when you have a public debate. What happens is people throw out their fact and maybe someone else throws out their other fact, and they're arguing totally different things. And so you can't have a productive debate unless you're arguing on the same premise.

And so that's what I was hoping this kind of software or ultimate debate would do, would help people zoom in and figure out what do you really believe on this topic? Which premises do you believe or not? And then if you're going to engage in an argument about it, let's define the scope of what premise we're arguing over. Could a search engine do that? Like if I Google climate change, like why couldn't you take that? You had broken that down into like 11 different things. Why couldn't the result be sort of like,

On the left, you get sort of like one side. And in the middle, you get the argument. And on the right, you get the other side. And that is your result. And you can click through to... That would be kind of ideal. That would be like an embedded of this software that this gentleman's building. He was calling it context, you know? Okay. Because to really give you context on the argument. Right. And then once you get into a premise, then it's like, okay, well, what are all... Then you can have more reasoned discussion because you can be like...

OK, what are all the evidence pro and con or different aspects of this premise? And then you can cite where the sources come from. And people can comment individually on those sources and say, OK, well, the main fact that everyone's behind on this is here. And then people can be like, well, I disagree with the methodology of that study or whatever it is.

but you're having a much more reasoned debate. And if you, if you approach a topic, you can come in and say, okay, here's all the different parts of this. I can get up to speed on this topic. Right. Really easily. Right. And that's kind of the idea. Whereas some of these topics now, it's just, you don't even know where to start. Why don't, why don't newspapers do that? And sort of like, instead of publishing, like publishing sort of like simultaneously, uh,

Both sides of an issue with the middle it would be great any issue. There's a few sites online that I've tried There's something called like pro con when we were kind of researching for this The Vox does it interesting they have these explainers, you know Which is pretty good, but nothing really captures this the way I think it should be so to answer your question I think yes, I think it is possible. Okay, but nothing's really hit upon it yet. All right, if I wasn't doing this I might work on that. I

Problem. Maybe you should do both. Maybe. Let's talk mental models. You have a book coming out. It's super thinking. I've read it. It's great. What made you create the book? So in a way, it's been ruminating for like 20 years. Like I had wanted to and started writing different versions of this book for a while.

But it would literally lapse for like five years. I could get into them if you want. But then what happened was maybe four years ago, DuckDuckGo started to more take off. And we have not really hired a lot of people from outside the company. We've really tried to grow people within the company. And so we had this executive team that was, you know, needed to step up and learn more. And I was trying to understand, okay, how do I train our executives to

Like, what do they need to do to really actually make good strategic decisions more of the time? Because that's really what executives do. And I got to thinking, okay, what I really think it is, which is really counterintuitive, is I think they need to know like these 300 different topics, these 300 mental models, which I think you've come on to the same point.

notion is like if you knew all these things and internalized it and could pattern match relatively effectively like the situation applies to these three mental models then you can skip lower levels of thinking and have this whole like higher level conversation really quickly right whereas if you don't know these mental models you're like starting from scratch every time and you you may you kind of don't know the design patterns of strategic thinking and so i started making a list

And I was like, okay, I'm going to make a list. And I list, I initially made a list of like a hundred things. And I asked somebody and I was like, how many of these do you know? You know? And they were like, I know like 30 of them. Were these a hundred things that you applied or did you make this list? Like you went to university one-on-one and you're like, what are the big ideas? Out of my head. I was like, these are the a hundred things that keep coming up again. Okay. Uh,

mental models that I continually use. And so did you always think like that? Or was this something that evolved? Like, did you come into school going like, Oh, I'm going to think about this in mental models or did you come like, did you, this is the more of the longer term history. So I came out of school undergrad. I took a lot of courses in different disciplines. And so the original book I wanted to write was like this, uh,

interdisciplinary like guidebook. Right. You know, and I was like, I think a lot of people would benefit because they only have one major. Right. Like you specialize. Yeah. You specialize immediately, but then that totally kind of ruins your education in a way. And you really want to, you know, all these things across systems. So I was like, what if we just made this like interdisciplinary textbook, like a survey course of everything. Right. And,

I eventually gave up on that. Why? I don't know. It just wasn't coming together. And I started my company and other things. And then maybe like five years after that, I was like thinking about... I had come on to Munger at that point before though, and I hadn't come on to him. And then I was thinking, well...

Another formulation I had of this concept was you know, I keep coming back to problem-solve like I problems in front of me and when you solve them you want to like run down a list of mental models like is This a critical mass problem or does this have to do with power law? Is that how you solve them like you have a mental checklist that you iterate through? Well, I was doing that at the time and I was like Wouldn't it be cool if there was something like a problem solvers handbook and it was like I

you know, when you're faced with a problem, like here's like a checklist or help jog your memory to see if you can think of any solutions. So I started on that road and then that, I, that didn't, I couldn't get that to work either. So then five years after that was like more like this problem. And then I ended up with this list of a hundred things. And I asked the guy and he's, he knew like 30 and I was like, okay, well that to me is what people's core curriculum should be. They should learn the rest that they don't know.

So then I went and I tried to more systematically list all of them. So I went back to Munger, and he has that quote about 80 to 90 models, but no one ever lists them all. So I'm like, okay, well, I'm going to list them all. And then I more systematically...

tried to brainstorm my own. And then I went and started going down the rabbit hole of looking in depth at each one and like, what are the related articles on Wikipedia and all these different things, just jogging my memory of other things. Most of which I already knew some of which I learned from, from scratch. And then I put out a blog post, uh, mental models that I use repeatedly and ended up being about, uh, 250. And that did really well. Like people really took to it.

And for a while I was just using that as like in our reviews internally and saying like, you should go anything you don't know on this list, you should go read about it. And I linked all the Wikipedia articles. So it's like, here's the starting point, you know? Um, and then we internally, we use them, you know, consistently when we're faced with different problems. Um, and then our, my publisher from our, my first book reached out and was like, well, this post is really successful. Would you consider writing a book on this? Um,

And I was like, I just wrote a book and it was really difficult and took me five years to write. That was the previous book. And so I don't really have time for this, but my wife had just kind of left her job as a statistician and was interested in doing it. And so we decided to do it together. Ended up being extremely long and amount of work. And I somehow like forgot how much work it was. And it took two and a half years basically to turn that list into an actual book.

But now it's done. So I'm happy about that. I have so many questions here. Let's start with like, what are the big five sort of mental model? Like say a hundred carry the most weight. Well, what are the five out of those hundred that you think? Yeah. I'm curious what you think about that too. Cause you, you have also a big list. Right. And so, so we, what we ended up doing is,

The original list was a big list, and it organized it by discipline. And then when we started on the book, we realized that's actually not the best way to do it because there are...

even though they came out of one discipline, they're actually interrelated, often cross-discipline. Right. And so we set out organizing them. They're all connected. They're all connected, exactly. So we ended up trying to group them into, like, really, how are they connected? Right. And so we ended up with nine chapters that relate all about 300. And so there's, like, you know, 40 in each chapter, roughly. But they're in a narrative-related fashion. Right. And so those, like, general...

end up being a lot of the general formulations of things you should know. So the first chapter is on bias in general. The second chapter is on unintended consequences. The third chapter is on using your time wisely, that kind of stuff. So within all of those, though, if you're going to boil it down, and some of those are on business and management, and so some of that doesn't apply to everybody all the time. But the ones that really come up a lot to me are opportunity cost, which is where I started, which is that

I would just tell the executive people, and I still do, is that you shouldn't argue that something's important and therefore we should do it. It should always be it's important and it's more important than all these other things. And that just frame of mind changes almost everything because it's almost every decision you ever make about how to spend your time is like, am I spending it in the most optimal way relative to my other options?

So opportunity cost comes up. Another one that comes up a ton, which is more rare that I don't think people mention as much, is forcing function to me. So like forcing function is the idea that it often operationalizes something in your calendar where you've preset a time or mechanism where something's going to happen and it forces you to think critically about something. Right.

So in a company, that can be a one-on-one standing meeting that you have every week to talk about something. Or it could be board meetings often serve this function because it's a forcing function that gets you to compile all your metrics and stuff. Even if the meeting's not that useful, compiling all that stuff is that useful. And you can do this in all sorts of ways, like schedule a time to go to the gym or whatever it is. What's an example of a forcing function you use that you find super valuable?

Yeah. So another mental model that we kind of run through that is this one-on-one standing meeting. So in our company, everybody has a career advisor and they have to have a one-on-one with them every week. That's an unstructured meeting. That's the advisee. It's their meeting. But it's a forcing function to get them to think about

anything that is going on that week what's really on their mind another one we use is every project needs to have a summary update at the end of every week kind of like what happened what's next and it's a you know it's again it's a forcing function to be able to think critically about kind of what's going on we have kickoff calls and post-mortems for every project and

Also kind of are you doing like agile sprints in the background or like how the projects get run? No, that's a whole different subject I mean we've kind of because objectives and projects can be more exploratory and not or may have external deadlines or not We try to stay away from deadlines if I can yeah But we do have these like kickoff sometimes mid-mortem or pre-mortems where it's like a forcing function that keep people to think critically about what's going on and

So we had opportunity cost, forcing function. What would be next on your intuitive list that I put you on the spot for? Probably one I mentioned earlier, which is thinking gray. And so, yeah, no, I should have probably brought a list. Oh, no, no, no. Yeah.

But I actually did but maybe that's more realistic because it's the ones about it But I think that comes up a lot the idea there is like, you know You could view this as confirmation bias, too But once you commit to a decision It's really hard to receive confirmation dice and you're you're like committed to that decision your psychology wants You to look for things that confirm your decision if you force yourself to actually not make a decision or not

not commit to absolutely believing something. Say like, I'm leaning this way, but I'm willing to entertain other evidence. It doesn't become part of your identity. And then you're a freer thinker. And so that's this concept of thinking gray.

There's another... People sometimes mean something else slightly different with that, which I also agree with, which is this notion of... But I categorize something else in the book called black and white thinking. Right. Where it's like things are either, you know, zero sum, one way. There's only two options. Right. Usually that's completely nonsense. I mean, usually there's lots of options and there's...

There's a whole degree of things. Nothing's usually black and white. In the podcast we did with Annie Duke, she had a really interesting way of thinking about this, which was just by default, say you're 99% certain on everything.

Never a hundred percent. And the reason I love her stuff. Yeah. Right. It's like, well, you've signaled to the other person that you are open to changing your mind. So now the conversation becomes like, what would cause you to change? Yes. So you get a more productive debate out of it, but you're less likely to sort of like ignore things that are contrary to your beliefs. I love that. That quote from Munger one is my favorite one is like, you know, I always know the other side person, the argument. Yeah. You know? Yeah. Yeah. And, and,

And that's what I think the thing in gray is all about. It's like, you want to understand all sides perfectly for the argument. Right. And do you use these terms internally? Yeah. Yeah. You actually use the term and do you have like a, um, I wouldn't say a checklist, but do you have like, um, what is your decision making sort of methodology or is there one? Um, it really varies. So what we do is we have a,

We have a somewhat unique way of running the company because it's all remote. So we're not everyone's remote. It's a remote-first company. And so as a result, there's a lot of... And also people are in way different time zones. And so there's a lot of asynchronous communication. And the way we run effectively is to make everything really transparent so all the projects are all open to anyone to follow.

to make all that work effectively, that isn't chaos, every project has a standard scoping template. And so it has like a background, objective, impact and complexity assessment. And so when we have a project and have a kickoff call or postmortem, it's in that framework of like, what was the objective of this project? Like, what are the success criteria? That's another way we define the objective.

And we can talk about like what it should be. Did we meet it? You know, right. There's like a way to frame that. And you do that before you start the project. Yeah. Before we start the project. That's rare. Yeah. Well, the other thing is like we're, we're a small company. I mean, I think a lot of companies are like this, but we're competing against companies that have infinite resources relative to us. And so this notion of opportunity costs, there's, we don't have a lot of resources to waste. Right.

Right. And so one of our kind of core values that we have three articulated is question assumptions. Okay. And so... What are the other two? Validate direction and build trust. Okay. And so with question assumptions, like we're basically looking like, should we be doing this at all? Like, is there a simpler way to do this? You know? And so we're asking right from the beginning, almost trying to like say,

maybe we shouldn't do this project. And so the only way to do that is to write out all the reasoning and then discuss it and then really question, well,

maybe we can do this a different way that's cheaper or better or faster, you know? Okay. And so there's no list of like mental models in the office wall that you're like walking through. Well, it's funny cause I do have a, I, I, I got to be friends with a guy around school of thought. Okay. Have you heard that? So he's got, he's got some, um, uh, websites once called like logical fallacy is, and

it's like a list of all the biases kind of thing. And then he's got another list of logical fallacies. And so I have two posters of his on my door. Okay. And so those are lists of mental models, but they aren't the list like from the book or anything. Do you think awareness or maybe to what extent does awareness of,

Because I think people get this list of cognitive biases and their inclination or default is to be like, oh, okay, well, I'll have a list of them and I'll go through this checklist. But they're really good at explaining why our thinking leads us astray, but they seem less good at being able to

in the moment or with foresight, avoid those thinking errors. They're really good at explaining why we were stupid. They're really bad at sort of like preventing stupid. And my experience has been a lot of times, not always, but if you have a list of them,

The smarter you are, the more you just convince yourself that they don't apply. So like you're actually not even open them anymore because you go through this list and the smarter, more intelligent sort of that you are, the better the story you tell yourself about why you're not overconfident.

Yeah. I, so I think there's something real true to this. I mean, this is why we have the core value of question assumptions. Right. And so I think what you need often is somebody else to call you on this, your stuff. Yeah. And then be open to, and then the organization or you need to be open to, you know, thinking gray about it and,

and changing your mind. Right. And so what, that's what we've tried. That's the culture we've tried to create. And so I do think it's actually pretty hard to do on your own to your point. Yeah. You know, I think it's better if you have somebody, you know, to talk to. And so like, for example, like,

I am lucky in that I walk pretty much every morning with my wife for an hour. And that's also how this book got started is we talk about all sorts of things, so company things, but also just current events. And we apply these models and stuff. But she helps call me on things.

My bullshit. Dinner table conversations must be awesome. Yeah. And vice versa, you know, and so she's got a PhD, but she's got a double degree from MIT too. Yeah. She is. Well, there's a whole chapter in this book about statistics, which you read and she, she was, she was the main one on that. Um, and to make, she was worried that she gets them wrong. You know, we should say something wrong. Um,

Because what's funny about these mental models is you don't need to be the world's expert to apply them. But if you're going to write a book on it, you also don't want to get the statistics wrong. You have to be accurate. What is the validate direction? What is that and how does it relate to feedback loops? And what sort of feedback loops do you set up? Yeah, so validate direction is, this gets to your point about decision and is the project going in the right direction? And so what we basically try to set up is

Another mental model we use really heavily at DuckDuckGo, it could be in the top five, is directly responsible individual, DRI, which we really took from Apple, but really expanded it, which is that every task, every project, every objective has one person who owns it. And they are the directly responsible individual, and it avoids...

this diffusion of responsibility and the bystander effect, other mental models, which basically like if you're on a email with five people and no one responds, it's because no one felt they were like it was stressed to them. You know, same thing in a meeting. If you have action items and no one's assigned the task,

It often doesn't get done because people are like, well, I thought he was going to do it. Exactly, right? Those are the worst meetings. You spend an hour in a room, everybody walks away going, oh, we decided what, but nobody knows who's doing what. Exactly. So to avoid that, we assign a DRI for everything. And projects also have a DRI. So someone will go on in their task and we have the kickoff call and they're like, okay, I know what my objective is. I'm going to go off and do this thing.

But they could go off and spin their wheels. They go from the wrong direction They could do all sorts of things and so we try to say is there a lot of other smart people in the company who? May be able to do things faster better know the right answer And so generally you should be coming back and validating your direction, right? And so the easiest way to do that is to just write what you're doing Yeah, and say not stop yourself. Just say look I'm gonna go to this and

Does anyone have any ideas or object to what I'm going to do? If no one objects, I'm just going to keep moving forward. The default is this is going to happen, but you can-- Yeah, exactly. Chime in if you want. And that works really well, because it doesn't stop people in their tracks. But anyone can follow anything.

And sometimes people will call out certain people that they know might have information about this, you know, or knowledge. And if they do, they can chime in. And if not, they're not. And so that's really like the core operationalizing of the Validate Direction. I like that. How does this relate to being a parent? Like how do you teach your kids about not only mental models, but sort of like preparing for like general knowledge about the world in multiple disciplines?

Yeah, that was one after we, so the original abyss right in the book was train executives. But then when I got into it, that's what my wife, our abyss was more. We think that all kids, including our kids should learn all these things and they generally don't learn them in school. And so, yeah. I mean, they learn a lot of the concepts in school, but in a domain dependent. Yeah. They don't learn how to apply them. And then even half of them, they literally don't learn. Right. Like half these things, um,

I either had to come across randomly or I was really lucky to go to this interdisciplinary graduate program where I was exposed to a lot of these things. Right. Um, but had, had I not done that, I would have been stumbling along, not, you know, knowing these things for a long period of time. Um, so we've been, um, with parenting, we've been trying to be like really explain things to kids and our kids in an adult way. So like I mentioned, we were listening to this, um,

debate podcasts with them. And it's an adult show. It's nothing to do with kids, you know, and really complicated topics. And we just play it and pause it and ask the kids what they know and what they think about it. You know, um, this morning I've been, so I've been driving the kids to school every morning and

These are just like random examples, tactical examples. Yeah, the more examples, the better. We've been, I've been listening. We first came on a bunch of kid podcasts that we listened to, but then I realized like,

My kids like listen to the adult podcast as much. So we start, I usually listen to the daily, which is the New York times daily podcast about things. And they, they do a really good job of like having kind of deep dives on historical topics. This morning's was about Brexit and Teresa May and all the situation going on there. And so we just listened to daily on the way to school and talk about it.

What do you talk about? Like, what does that look like? It talk, it's like, um, do you understand what's going on is the first level, you know, then it's like, okay, why are they wanting to leave? Why do you think they want to leave the Airbnb union? You know, like what are their options? Why is she resigning? You know, all that kind of stuff. Just trying to like,

pick apart things that we know more intuitively as adults or, or we're told in the podcast in an adult way and trying to make sure they understand it and the logical chain into like, okay,

okay, well, how did they get to this? And then do you sort of like nudge them to multiple perspectives to you? Like how does the EU feel? Yeah. That's why I like the debates in particular, because you get really intelligent people talking on both sides. Right. And you get like, well, what do you like about their argument versus their argument? And it's like, well, what about what this person said versus that person said? Yeah. Agreed. And so are you ever like, are you trying to resolve things or are you just leaving it open? Open. Yeah. Generally open. Yeah. Yeah.

and do you follow up um not really okay what are what are the other things maybe we should follow up i don't know man what are the other things you do sort of like maybe around the dinner table or with your kids to get them um just thinking in different sort of ways so uh that schools don't teach yeah well i'm really lucky we just switched schools and i really love our kids new school so i feel like

happy that our school may actually start to do a lot of things that we want to do now that we're not going to stop. But so one thing I'm more encouraged about the kids school at the moment, but, um, so I, my kids are two very different. I don't know about your kids, but, um,

So there's different strategies for each. So my oldest is really into movies. And so one entry point to him-- I also like movies a lot-- is we've just been watching a lot of movies and then talking about them in the same way. Movies are great because the production value or thought put into every minute of a good movie is really, really high. It's a really well-produced thing. There's a lot going on there that you can talk about. And they're also very entertaining.

So we've been watching a lot of movies and then kind of talking about them, all different kinds. My youngest... Seven, right? Yeah, seven. He's really into math and really he's gotten into programming a bit and into watching like science podcasts and things like that. I mean, not science podcasts, science channels like on YouTube. Are you doing like scratch programming? Yeah, he took a...

They're in like this Johns Hopkins correspondence courses. So he took a scratch one. Yeah. This last semester. And now he's taking a cryptography one. Oh, cool. And my seven oldest is, yeah, he's going to come out with like a blockchain company. He's definitely more advanced than I was at this age. I'm super jealous of what the opportunities they have. Oh yeah. That's great. And yeah, my oldest is in, in a Python one. And, but he, he's more wants to talk about like,

the youtube videos he's watching right you know and so like i try to i guess what i'm trying to say is i try to engage with them at what they're interested in right so you try to bring it back to them and in terms of like whatever they're interested in but relate the concepts yeah exactly let's talk about that um the other thing we did relatively recently is i put all the watch crash course no it's so it's really good um it's a uh it's summer by pbs

But two brothers, they basically made online animated, but them talking heads to courses on economics and biology and chemistry and film and all these different things on YouTube. What was that called again? Crash Course. Crash Course, okay. It's really well done. I put them all on DVDs.

And so we drive to school every morning. Right. It's like 30 minutes. And so, you know, not every day we watch the podcast. Sometimes I watch craft course, you know, and do the same thing. I guess it's all the same kind of general concept is find really good content. Right. And then talk about the content with them. Right. In a way that reflects. Yeah. I guess it's all the same really technique, whether it's a video or a course or a movie or a podcast, you know, it's like,

find something really engaging you can talk about and then relate things back to it. What made you switch schools? You said you switched school. What was the criteria for the new school and what made you leave? We were at a public school and it was generally a good school. They both have, I don't know what it's like in Canada, but

In the US, it's state by state, but they have gifted programs. And they had something called GIAP, which was like an individualized education plan for their kind of gifted center. But things were moving really slow. They were bored a lot. Our youngest, who's really kind of good at math, he skipped kindergarten and then skipped another grade of math and then passed out of that grade. And they didn't really know what to do with him.

and it was just kind of bored, I guess. Yeah. And, um, so this new school is, is, is kind of brand new. It was only four years old. It is a, it's a, it's a school for gifted education totally, but it's not like you have to be super smart to get in. Like they're willing to take a lot of people who I think are willing to just engage with the idea and it's more project based learning. And so, um,

they, they do two things. I like one is they don't waste any time during the day. It's just kind of like, so they do way more. And so they have all the regular stuff, but they also have coding every day. They have a project class every day. Oh, that's so cool. And, um, and then on regular things like science or something, they're, they're doing a project related to that thing. And they also kind of force the kids to do it. And they have, they fail if they fail. Right. And they don't, they don't like huddle them with it. Right. And, um,

And then they had all these other kind of random things like, you know... Do they do things to like build resilience? Like as you're saying that, because a lot of gifted kids, like things come very natural to them. And then when they're faced with a struggle, like what do they do to build grit or resilience? The main thing they do is let them fail at things and just make them try again until they get it. And I think that goes a long way, seemingly. Although we're relatively new, so...

And are the kids liking it better? Kids love it. Yeah. Because they, um, they're both like super creative and love projects. And so it's like half the day is like doing different projects. Like Eli's in the fourth grade, they're building an escape room in their project class for the semester that they have like two, uh,

They split the fourth and fifth grade actually runs like a middle school. And they split them into two groups and they're like making escape rooms for each other. And they have to like build it and like do it all. And then test it. Yeah. And so they're just like doing that. And he loves that kind of stuff. Oh, that's so cool. And so I'm excited for them. I wish I would go to school. Oh, my God. Seriously. We didn't start programming in school until.

12 and it was like Pascal. Yeah. That was, I mean, yeah, I'm 39. So yeah, we're the same age. That was exactly my experience as well. But it was great. Cause I mean, I started much earlier than that. So like the coding class became like super easy. Yeah. Great way to get my grades up for university. I'm with you. That was my path as well.

Listen, this has been a phenomenal conversation. I want to thank you for taking the time and maybe we can follow up. Yeah, love to again. Hey guys, this is Shane again. Just a few more things before we wrap up. You can find show notes at farnamstreetblog.com slash podcast. That's F-A-R-N-A-M-S-T-R-E-E-T-B-L-O-G dot com slash podcast. You can also find information there on how to get a transcript.

And if you'd like to receive a weekly email from me filled with all sorts of brain food, go to farnhamstreetblog.com slash newsletter. This is all the good stuff I've found on the web that week that I've read and shared with close friends, books I'm reading, and so much more. Thank you for listening.