Capitalisn’t: You Can’t Buy Trust

Episode Transcript

Jimmy Wales: So many people come to me with their idea of how to make Wikipedia better by having a price system of bidding on, and you could stake your claim, and if you got reverted, you would lose your money. I was just like, "Yes, let's put the truth up for the highest bidder. Now, that doesn't make any sense."

Bethany McLean: Luigi, you have been lobbying me for months to interview Jimmy Wales, the founder of Wikipedia, about his new book, The Seven Rules of Trust. I do not understand why a not-for-profit encyclopedia is interesting for a podcast about capitalism.

Luigi Zingales: Let me explain. One of the key justifications for the superiority of a capitalist economy with respect to a centrally planned one comes from a 1945 paper by Friedrich Hayek. Hayek starts from the premise that information is highly dispersed across society's members, and that since no central body can gather all this widely dispersed information needed to allocate resources, we need to push down the decision power to where the information is. In other words, we need to decentralize decision-making. How do you coordinate the decentralized decision-making? Prices do the work. The problem of resource allocation should be assigned to markets, not to governments.

Bethany: I love when you play professor, but what does this have to do with Jimmy Wales? [chuckles]

Luigi: You're right. While a traditional encyclopedia is the product of central planning, Wikipedia represent a decentralized solution. Individuals contribute local knowledge, corrections emerge from distributed participation, coordination is decentralized, and all that emerge spontaneously. This resembles Hayek's idea of spontaneous order.

Bethany: All right. You've convinced me on this part, but let's move on now. How is this related to democracy?

Luigi: Think about the following problem. Socially dispersed information can be transmitted in three forms: market prices, talk, and votes. The voting mechanism, the one we generally use in democracy, has not been working very well recently. I don't know whether you noticed, but it didn't. Is there something we can learn from the Wikipedia mechanism to aggregate information?

Bethany: Interesting. You might really have to work here. What is the business insight, given that Wikipedia is a not-for-profit?

Luigi: This comes more from his book than from Wikipedia itself, but I think he emphasizes two things. Number one, he emphasizes why trust is so important in any business, and Wikipedia is not a section and to what extent Wikipedia cannot be run for profit because a for-profit entity will destroy that trust. Also, in the book, Jimmy Wales is very keen about the importance of purpose as a mechanism to coordinate firms' activities and to enhance trust in those firms.

Bethany: Jimmy, you stated that Hayek's article, The Use of Knowledge in Society, was "central" to your thinking about how to manage the Wikipedia project. Why?

Jimmy: There was a time when there was a big debate about central planning versus price system, market economy. Hayek put forward the notion, which I think has been proven largely correct, that one of the problems with central planning, or the problem, is the information problem. Your choice is either to send all the information inward to a central authority of so-called experts who will make decisions about the marginal costs and benefits of everything, or you move the decision-making out to the endpoints where information exists, like how much bread do I need tomorrow for my restaurant, and things like that.

That's only an analogy because Wikipedia is not a market system. There's no editor-in-chief. There's no central body. When articles are being edited, it isn't being supervised from up above. It's a conversation and a discourse that's happening right there. That was the meaning of that.

Luigi: Can you walk us through the process? Suppose Bethany and I are trying to edit a page on JFK. I write something, and Bethany disagrees, and edit out, and I edit back. My understanding is, after three rounds, something happened. Suppose that we go back and forth, how do we resolve this issue?

Jimmy: We have something, yes, three-revert rule. It's a custom and a standard. It's not a hard rule, and it's definitely not in the software. The three-revert rule isn't to say, "Well, now you're not allowed to edit that anymore. You've had your three chances." It's to say what you should be doing. What the three-revert rule encourages you to do is to say, "Okay, actually, going back and forth, back and forth, it's boring. It's really not helping. [chuckles] Neither of you is going to get anywhere. You could do this for the rest of your lives, and it's annoying the rest of us."

Instead, what we say is, "No, come on, let's talk about a compromise. What is the common ground you might have? What is the thing that we might step back from this and say, 'Okay, actually, what we do agree on is this. Okay, let's write that, and let's try to be productive in our conversation to try to reach some collaborative compromise.'"

Luigi: Suppose that I claim that JFK had sex with 25 women, and Bethany thinks they are actually 26. I cite Breitbart, and she is more educated, she cites The New York Times. How do we resolve this dispute?

Jimmy: This would be where we look into questions of consensus and questions of what does the more broader community think. We have standards on sourcing, which people will rely on to say, "Well, actually, for obvious reasons, not every source in the world is exactly the same as every other source in the world. Some are higher quality, some are lower quality." Probably just being empirical about what would likely happen is people would say, "Well, hold on, let's take a look at more sources."

Then some very nerdy person would go out and say, "Well, actually, here's the seven different books about JFK's life by prominent historians. They've all got the number as 25. The only claim of 26 was only surfaced in Breitbart last year, and they didn't actually give any evidence or reason. Probably, it's an error in Breitbart." I don't know. Maybe they've got an agenda. I don't know who they're claiming JFK slept with that got them excited.

In a case like that, the question is, how do we determine consensus as opposed to unanimity and as opposed to majority? One of the things that we try really hard not to do is to vote. Voting is very problematic because voting is gameable. People can go out on the Internet and say, "Oh, look what the terrible Wikipedians are doing. Come on, everybody, let's go in and we'll out-vote them."

Consensus and determining consensus, obviously, there's a bit of ambiguity there. It doesn't mean 60% plus one. It's the preponderance of the arguments and the debate. You are focusing on a part of Wikipedia where there is this kind of looseness to it in the sense that there is no rule. Part of the reason that there is no specific number like, "Oh, well, you need to get 60% plus one," is the context can change so widely. Sometimes you say, "Well, actually, we're about to make quite a controversial claim. We better have a really strong consensus for that," or "This actually isn't that important one way or the other. Actually, if it's a slight majority, maybe we'll go with that."

Other people are like, "Yes, it's not that important." Then some of these debates go on for a long time. If people are reasonable, it normally works out okay. If they're not reasonable, then it ends up with the arbitration committee or whatever it might be to say, "Actually, let's figure out exactly what we're going to do here."

Bethany: How has doing this made you think differently about truth? Do you think there is generally a truth, or do you think that truth is more complicated than you ever would have imagined before you went down this road? I started my career as a fact checker, so all of this is very interesting to me.

Jimmy: I believe in reality, and that true statements are recognized facts of reality. I think there is a truth, but I also think that doesn't get you very far in this kind of situation because obviously other people think something else is true, and other reasonable people may have reasons to believe that something is true that I think is false. If we're trying to write an encyclopedia, then you have to have a spirit of humility and a spirit of kindness to say, "Okay, actually, I know a lot about this, and this is definitely the case, but I also can recognize there is this other competing view and people are putting forward evidence for it. I'm not persuaded by the evidence, but it's actually not a completely insane thing to say."

Therefore, the article needs to reflect the state of the debate. Well, it turns out even if we disagree, we can fairly characterize the debate. That's quite common. If we think about, I don't know, an issue in economics, like what is the impact of minimum wage laws on employment? Maybe we're never going to agree about that, but we can probably do a good job of characterizing, "Here's the evidence. Here's the state of the debate amongst experts. Here's some of the public discourse around that."

That's what an encyclopedia should do. That doesn't mean that we are adopting an idea that, "Oh, none of these is true," or "Who knows?" My favorite example of false neutrality is, ah, the moon. Some say rocks, some say cheese. Who knows? That is a neutrality. Whereas there are plenty of issues where I may have a very strong view myself, but I also can plausibly recognize that, "Oh, okay, actually, in an encyclopedia article, you should be more broad than that."

Luigi: Wikipedia showed that large-scale knowledge production can occur in a decentralized, coordinated fashion, only in the presence of rules, norms, and some governance mechanism. Can this idea be applied in other areas?

Jimmy: Yes, I think it definitely does. I think we definitely see this in a lot of the processes and traditions and procedures of academic research. It's decentralized knowledge production because they're all independent researchers who have their own minds and have this. There are debates that rage on for years. In order to try and make that a productive thing and not just people yelling at each other, there's journals with standards, and you try and get published and so on and so forth.

Hopefully, those journals have a spirit of inquiry and curiosity and are occasionally willing to publish something that's unusual and surprising and dissenting because that's part of progress. Sometimes it turns out it was not correct, but that's okay. It's part of the process. I think this has always been how we as humans do progress knowledge. We progress through dialogue, discourse, through reason, reflection, and so forth.

That process can be well-managed or badly managed. It can work well or poorly. The highly politicized Soviet encyclopedias were full of nonsense and in parts because nobody was willing to take the political risk to challenge orthodoxy and so on and so forth. We need to preserve processes that actually tend to produce better result.

Bethany: In some ways, obviously, Wikipedia is very anti-capitalist. Yet, when you began talking about Hayek and the core idea of the market in sourcing information, in some ways, it's also very capitalist. Do you ultimately think of it as a capitalist project or a non-capitalist project?

Jimmy: I would go with non-capitalist rather than anti-capitalist. It certainly isn't anti-capitalist. The way Wikipedia works, as I say, the analogy to thinking about price system and the distribution of knowledge and society and information is really just an analogy. In fact, actually, maybe one of the reasons I bristle at the comparison to blockchain is during the boom of blockchain, so many people come to me with their idea of how to make Wikipedia better by having a price system of bidding on, and you could stake your claim, and if you got reverted, you would lose your money.

I was just like, "Yes, let's put the truth up for the highest bidder." No, that doesn't make any sense. That's not a process that leads to truth. It's a process that leads to a reasonable production of bread. It's outside of that system, but in the same way that Christmas dinner with your family is outside the system. It isn't anti-capitalist. It's not pro-capitalist. It's just dinner with your family. There's relationships.

If you, as many economists are slightly prone to do, really force yourself to view everything through the lens of exchange of value, and you think, "Well, no, actually, my mom is paying for her dinner. She comes to my house for Christmas dinner by saying nice things to the kids." No, that literally isn't how it works at all. That's way too homo-economist nonsense. A lot of Wikipedia is just a bunch of nerds who love writing and talking about their nerd specialty subjects.

There's no market element to it at all, and there's no money element at all. Then other aspects of it obviously do exist within the capitalist system. We need servers. We have to get some money and buy the servers and things like that. That's obviously part of the system. Also, a lot of people may turn to Wikipedia for educational reasons, and they want to get a college degree. Why do they want to get a college degree? Because they want to get a better job and all of that kind of stuff. Depending on the level you look at, obviously, you could say it's part of capitalism, sort of. Yes.

Luigi: I appreciate what you said, especially about the Christmas dinner.

Bethany: I appreciated the note on homo-economist. Sorry, Luigi. [laughs]

Luigi: No, I think he's absolutely right. This said, Wikipedia is embedded in a capitalist world. We know that companies spend billions on lobbying, and they're certainly ready to pay millions to have their own Wikipedia page beautified because it impacts their reputation. How can you prevent this from happening?

Jimmy: Yes. As it turns out, I don't think they're willing to pay millions for it. There are definitely people out there who try to hire PR people to help with their Wikipedia page. A lot of times in online discourse, when people are talking about YouTube or something like this, they think about the YouTube community. I would argue there isn't really a YouTube community. There are communities of YouTubers who happen to know each other and sections of YouTube where people have made friends and all that.

What I mean by the Wikipedia community is not just everybody that randomly shows up and makes one edit. The people who are discussing the policies, who are enforcing, who are thinking about it all the time, who are really active, those people are healthy, happy, empowered to make decisions and to do all of that. That's really important. Then what happens is somebody comes in, and if they are making a contribution that fits within the parameters, like a classic type of example is a lot of companies, for example, are actually very boring and not very interesting to the Wikipedia community.

People will come and sometimes update their entry with the latest news or whatever. We don't approve of that, but we don't consider it like a massive, huge problem. As long as they're not inserting fluffy language and things like that, probably nobody's going to really complain. If you come to Wikipedia and you start deleting criticism, you start fluffing up the language, then people are going to be like, "Hold on, what are you doing? Why are you doing this?"

There's little templates we put up. "Apparently, someone with a conflict of interest has been editing this article and things like that." It's part of the process. It's always been with us, but it is something that we really try to discourage. There are a handful of-- they're all small, but small companies who are just really annoying. We have to ban them over and over and over. Then they resort to deceptive tactics and so forth and so on. It usually doesn't work very well. Sometimes it can work for a little while, but it's part of the whole process, really.

Bethany: I was reading one way in which the rise of AI is maybe not an existential threat to Wikipedia, but obviously, by constantly deluging the site and harassing the site to scrape it. The other way I was thinking, it's something that worries me, because I find AI to be wrong so frequently, that as there's more and more AI out there, does that start changing the nature of truth?

Then people can point to 100 articles that say that XYZ happened because they were all written by AI, and they all say the same thing. Do we then start living in a world where truth has lost its meaning?

Jimmy: Yes, well, this is something that should concern us all, for sure. I think, for Wikipedia, we are reasonably robust to this because we have a lot of expertise, and we spend our time debating about the quality of sourcing. Where I worry the most is what if journalists fall for it? If journalists fall for it, then suddenly we are vulnerable because we listen to journalists and that sort of thing. Just recently, I got involved somewhere here on my desk. I don't see it at the moment. I bought a book because I got obsessed with a very, very, very unimportant and narrow issue.

I was on Reddit, and somebody posted this Today I Learned, and most of those are links to Wikipedia. The story was this guy who was a survivor from the Titanic, later became an extra in a movie about the Titanic sinking because he wanted to go down with the ship twice. It's a cute line. I was like, "Oh, that's amusing." Then I went, followed the link to read about this guy on Wikipedia. I read this story, and I was like, "This doesn't really have a good source."

I started digging in more and more and more. Eventually, after a lot of work, I found the source. I was very excited, but then I looked at it more carefully, and it was an excerpt from a novel. A novel had been written as if it were a news report. It was a biography, and it talked about this as a side note. What clearly happened is some Wikipedian had seen that, thought it was a source. Sorry, if you Google it, the story is repeated on hundreds of websites. It's just not true. Probably our fault.

I posted on something, I was like, "I'm still not sure. There's only one place left where the story might be." I bought that book, and it's not in that book. I'm like, "Okay, I'm convinced." If somebody can find a source, great. It's not like I proved it was false. I just say there's no source. It seems unlikely. It seems like the only source is this novel. Actually, it's on my agenda to try and find the author of the novel and say, "Did you just make it up, or is there actually a source?" Anyway, I don't know how I got on that whole rant. It's like my boring story of how I edit Wikipedia sometimes.

Luigi: A lot of the LLM model have been trained on Wikipedia text. First of all, I would like to know, are you proud or are you angry about this? Second, now they start to be trained on more product of LLM models, so it becomes more self-referential. This becomes really a problem. As you said, if the so-called legitimate sources accept some of that, then they spread misinformation everywhere.

Jimmy: I completely agree. We're not mad about it. There's certain beefs we have with overhauling our site and things like that, but it is freely licensed. It's a gift to the world. I'm actually happy on a certain level. I think we would be in big trouble if AI were only trained on Twitter. It would be a very stupid and angry AI. I do think that if AI is training on the output of AI, there's not really a corrective mechanism in a technology that already isn't that good at having a corrective mechanism anyway. It just predicts the next token and so on and so forth.

I suspect the companies are well aware of this, and they're thinking hard about how to be careful about that because it doesn't do them any good to have an AI that spits nonsense because it read nonsense that it wrote itself. I actually think what is interesting because there's this idea of synthetic content being used to train. One of the things that the large language models are really good at is coding. One of the reasons is that, with coding, the software either runs or it doesn't. That isn't necessarily true in lots of areas of life.

It's like, what is the correct moral argument for such and such? It's like what we've been talking about. There's a lot of things in life that are complex and debatable, and there is no simple one answer that we can just say, "Yes, the AI is doing well because it got that answer right." Also, in areas where you could argue there probably is a class of right answers, maybe no one single right answer.

I love to cook. If you ask a large language model to produce a recipe for a chocolate cake, well, there is a range of recipes that you would say, "Well, that's chocolate cake at the end." There's a range that are just inedible nonsense. Put glue on pizza or whatever that old joke is. The problem there is, unlike code, which AI can actually then run the code and check if it's right, AI has no way of cooking a cake and tasting it to see if it's good.

That feedback loop to say, "Oh, it can generate a recipe, but it actually has no way of testing it." If you're asking for a variation on a common recipe, you can probably do something plausible, but there's no way to really refine it because is it right, is it wrong? Only humans can tell you.

Bethany: I saw something or read something recently about the difference between kind and wicked systems. Kind systems are ones where you can test and can check the results and where they do progress in a linear fashion. Wicked systems are more human systems. I thought that was such an interesting way of thinking about where AI might be good and where it might not be good. A question for you.

Jimmy: Oh, perfect. Great. I've just Googled that. I've got some reading material. That's an idea that's been in my head for a while. I didn't know any words for it. Now I'm going to--

Bethany: Good. You'll have to let me know if it's helpful. There's such an interesting maybe feedback loop between journalists and Wikipedia, because you rely on what journalists have done to ascertain what truth is, and then journalists in turn rely on Wikipedia. Do you have any lessons or any thoughts from your process on how to make journalism more trusted by people, or where journalists have gone wrong, where Wikipedia is still mostly trusted?

Jimmy: There's research showing that when newspapers endorse a political candidate, it reduces trust not just from people who disagree with that, but people who agree with the endorsement tend to trust less. I think that's super interesting because they think, "Well, gee, I agree with this endorsement, but now is this paper reporting the news with a slant to try to support their candidate? Are they bringing me all the facts or only the selected facts because they're trying to campaign for someone?"

I think there's an important lesson in that about neutrality. It's important for Wikipedia, obviously, to be as neutral as we can. We've talked a lot about how we try to do that, and it's definitely not perfect and so forth. I do think that's quite important. I think it's a complicated question, though. I wouldn't advise newspapers to suddenly become absolutely, completely straight and neutral in every aspect of everything they do because that would be hard to survive in a market where people do like to buy a paper that's a bit opinionated.

I think you can do both if you're careful and just have the right balance. I live in London. Two of the newspapers here in London, we have The Guardian and The Telegraph. They're both generally quality papers. The Guardian is a center-left paper, Telegraph center-right. That's a simple explanation. I love electric cars, and I own an electric car, not a Tesla, as you might not be surprised to know. I find both papers a little bit disappointing on the question of electric cars because, even in the straight news stories, I always say, "Just give me the headline. I can probably tell you which paper it came from because The Telegraph hates electric cars and The Guardian loves them."

I like electric cars, but I don't want to only be fed the positive things. I am actually interested in questions like, the electric cars are heavier, so the tires wear down faster, and then that generates more little rubber particles in the air, and so forth. That's interesting. I'm not so offended by someone claiming that or talking about that, that I don't even want to know or I want to hear a total whitewash of the pollution story, even though I think that's probably a fraction of the problem of burning fossil fuel all the time, but whatever.

I just don't trust either paper to give me that full story. It's a little disappointing that I'm not sure that I can trust The Guardian, which I prefer, but I definitely am not sure I can trust The Telegraph either when they say we're all going to die from air pollution, or The Guardian says, "Oh, it's nothing at all." I'm exaggerating both of them. I think there is a lesson there.

Luigi: Speaking of institutional neutrality, you're probably the only known academic who know and cites the University of Chicago carbon report on institutional neutrality. For the listener, in his book, it says that Wikipedia should follow the carbon report on not taking a position like the University of Chicago does not take a position on the fact of the day, in order also to protect the researchers who don't feel they have to take a stand, but they do independent research.

However, as you know, there are a few exception, and one that is very clear is when the core mission of the university is at risk. In that particular case, the carbon report says not only you can't take a position, you have to take a position. What is the corresponding exception for Wikipedia?

Jimmy: For the encyclopedia, there really isn't an exception. The encyclopedia is independent. It's written by the community. For the Wikimedia Foundation, the charity that owns and operates Wikipedia that I set up, it's probably very simple. The Wikimedia Foundation doesn't put out a statement about Obamacare, for example. It's not existential for us. Privately, we'll have our opinions, but it's not really for the Wikimedia Foundation to have an opinion about every single issue in the world. For one, why do we care? Not we, why would anybody else care what the Wikimedia Foundation, a nonprofit devoted to a technological project for an encyclopedia, think about healthcare policy? We're not experts.

Then the second is, if it is existential, we do have policy views on freedom of expression, access to knowledge, internet regulation, and so forth. In the last couple of years, we've had this concept, we call it the Wikipedia rule. If you're thinking of some regulation of social media, pause and think for a second what it's going to do to Wikipedia, because you're probably not trying to cause problems for Wikipedia. You're trying to solve a problem with social media.

We've had some success. A lot of legislators do go, yes, actually, we should stop and think a little bit about that. What is a social media website? How do you distinguish social media from other things? Well, Wikipedia is social. Wikipedia is media. You've got to think hard about that. We have such a different model. Our whole moderation model isn't that feudal model that I talk about.

It would be possible to impose obligations on the Wikimedia Foundation, like weakening Section 230, for example, that would make the Wikimedia Foundation somehow responsible for content and in a new way, which would require them to then become the moderators, thus undermining the only model that we know of on the internet that actually works pretty well, which is the Wikipedia model of this community discourse and reflection.

You wouldn't want to break Wikipedia. You wouldn't want to require us to become like Facebook. There's a lot of these kinds of examples. In fighting piracy in Europe, actually, the European Copyright Convention is still a mess. One of the things that is often put forward is, "Well, the big platforms should scan every upload to make sure it isn't a copyright violation and then block the upload if it's copyright violation." That's a really bad idea for lots of reasons.

For Wikipedia, it's a really, really, really stupid idea because we aren't where piracy happens. In fact, a lot of work that's copyright, copyright doesn't give the owner an absolute right over everything about it. You can use copyrighted materials for commentary, all these kinds of things. There's a lot of fair use and all of that. We go out, and we remind people. We're like, "Yes, okay, you're trying to stop people pirating movies. Don't pass a law that's so badly written that it's going to force the Wikimedia Foundation to somehow figure out how we're going to filter uploads to keep out copyrighted movies, which ain't a thing for us anyway."

Luigi: Bethany, did I convince you at the end of this interview?

Bethany: I don't know. I'm still mulling over this idea of where Wikipedia sits in the spectrum of capitalism because, in some ways, per Hayek analogy, it has capitalist mechanisms in it. I'm not saying that very well because obviously there's no profit motive, but it is based on the idea that central planning isn't the best way to run information. I think that's really fascinating. I'm not completely convinced of this idea that it couldn't be for profit without destroying trust. I think that might be the case with Wikipedia. I don't know that that has any broader relevance for business as a whole.

Luigi: Let me push back on a second. We didn't have time to ask him this, but there is an iconic tweet that, at some point, somebody is saying how much Wikipedia would cost, and he said not for sale. Honestly, the questions I did ask him, and I think he underplayed this, and I'm actually surprised he underplayed so much, because I think that Wikipedia pages are quite important for companies. Companies would be very happy to pay money to have their stuff better, and countries would be very happy to pay money to have stuff removed. If you add for-profit organization, really, you wouldn't be tempted to sell out?

Bethany: No, I agree with that. I understand that with Wikipedia in spades. I'm just not sure how relevant that is to the broader business world. That's where I'm at. That's where I was quibbling. Not that it isn't incredibly important, because I see that obviously a for-profit Wikipedia would have no choice, especially if it were owned by a private equity firm, would have no choice.

[laughter]

Luigi: You couldn't miss to mention private equity.

[laughter]

Bethany: No, I couldn't. Sorry.

Luigi: No, but to some extent is right. I think that it's in a spectrum in which not-for-profit is at one extreme and private equity is at the other extreme. Once we're debating private equity, and I offer to my friend and our former guest, Steve Kaplan, we're all getting older. Now I want you to write down for your children when they have to put you in a retirement home. Do you want a retirement home that is run by a private equity firm, a normal for-profit firm, or a not-for-profit firm? Which one would you choose?

Bethany: What did he say?

Luigi: He did not answer the question. I think that to me, the answer is clear. I want a not-for-profit firm.

Bethany: Yes. Anyway, I'm mulling over what he said about what Wikipedia's success in being trusted can teach journalism. I don't know that he's right, but I'm thinking about it and I think it definitely has more relevance for news and information. I'm not sure how much relevance it has for business more broadly.

Luigi: To some extent, he's the living proof that spontaneous order was wonder because what is great about Wikipedia is this constant checking and transparency, feedback mechanism, repeat. He adopted early on in his career this approach and I think produced something. I don't think he's fully understanding how it works to some extent. He doesn't have a theory. In academia, we first have a theory and then we develop. I think he developed and still is developing the theory, which is perfectly fine.

I think it's very interesting because I think there is a wisdom so far, is a wisdom in what he has produced that is quite deep. I think that part of the story, the way I understood it, is you want to leave things a little bit unspecified because the moment they are too specified, then you create incentive for somebody to play around them. In a sense, there is a saying in Italian, but I think applies mostly to civil law, which is coded law, not necessarily common law, that the moment you create a law, you create the trick to go around it.

Bethany: Keep going. Sorry. I was going to say it's the famous principles versus rules debate in accounting. When accounting is so rules-based, it actually creates an end run because in the structure of the rules themselves or in the detail of the rules themselves is the roadmap for getting around the rules and for finding loopholes in the rules. Whereas if you don't have that roadmap, then you have to work a little harder.

Luigi: The tricky part that I don't fully understand, maybe you understood better, is that this system works with a lot of social norms, et cetera., of a small community. Did you get a sense of the details of how the system works?

Bethany: I think I did because I fact-checked so many of my own stories that I think I understand and fact-checked other people's stories too that I think I understand that debate or that process. I started my career in journalism as a fact-checker where I'd have to go sit back in those old days. The fact-checker was responsible for not just every fact in the story, but for the overall accuracy of the writer's point of view. You would have to go sit there with the editor and the writer and explain the changes you wanted to make and why and what you were going to say instead. I do think just from having lived through that, I think I have an appreciation for how that process would work without being able to put it into words.

Luigi: Maybe we should teach more fact-checking to people in general. I think it's becoming much more important for everybody today, especially when we rely so much on AI, and then you need to fact-check the facts of AI.

Bethany: Yes. I've often said that if I had all the money in the world and time, I would start fact-checking not-for-profit. What it would do would be to create course materials so fact-checking could be taught in schools. Probably at the high school level, but the old school method of fact-checking as practiced by The New Yorker and as magazines used to practice it where you literally have to put a check mark over things as you verify them. Then you have to figure out how to verify every fact and then how to verify the overall point of view in the story, whether it's supportable or not.

I think it's such a useful exercise because, if you've been a fact-checker, you never read the same way again because you see everything through that lens. I've enlisted friends on occasion to help me with fact-checking and they always think it's going to be much easier once I give them things. One of my friends was like, "Oh my God, this is so difficult," because it really is. Facts are slippery little beasts.

I also think just the fact that it exists and I had not heard the phrase before—Wikipedians—the fact that there are all these Wikipedians, I find very reassuring about human nature, that there are all these people out there who care about information. They're about making it accurate, too. It really is, in many ways, the antidote to an AI world. It's very, very human. I like that he keeps reinforcing that. Maybe he's doing so deliberately in this age of AI to make it clear that when you have Peter Thiel supposedly coming up with this idea to have a AI-driven arbiter of truth that will rule on whether journalists have been fair or not in reporting a story, the fact that Wikipedia exists, I am just profoundly grateful for that.

Luigi: AI arbiter of truth, train on Grokipedia.

[laughter]

Bethany: Right. An AI arbiter of truth when AI can't even summarize an article correctly. I'm just, geez. If I were to write a dystopian fiction, I would write the dystopian fiction to be written, I think, is this parallel world of reality that if-- I read something somewhere recently that all writing is beginning to sound the same because people are using AI to write. It all sounds the same. The idea of a voice has gone away. I thought, "Wow, if there's this parallel world of writing, then there's this parallel world of truth, too, where it almost doesn't matter anymore if AI is accurate or not or if it's hallucinating, because if it's hallucinating and everybody repeats the hallucinations as if they're truth, then does the hallucination become the reality?"

Luigi: Yes. If you keep repeating enough, it does become. I saw a fantastic play in New York that was called Marjorie Prime, in which people start to develop a prime that repeat the thoughts of somebody who died. They use it as entertainment. The old mother starts to entertain herself after the husband died, but then they start modifying the truth to make it more pleasant. Then the play ends with three primes talking to each other and reinventing the truth.

Bethany: Is the play still playing? Is it still playing? Do you know?

Luigi: No, unfortunately,-

Bethany: That's such a bummer.

Luigi: -it ended in early February, I think, but it was really, really good.

Bethany: I wish you had told me about it. My new favorite thing is going to plays alone. Once you go by yourself, you'll never go back to going with other people. It's really delightful.