Tim Hwang is an artist whose medium is the organization. He’s held jobs at Google, Substack, and Inflection AI, among others, and founded many more organizations and events on the side, including ROFLCon, the Awesome Foundation, the Bay Area Infrastructure Observatory, and the Institute for a Christian Machine Intelligence.

In this conversation we explore why he once ran a four-day workshop treating tarot as a design problem; why we should start far more organizations; how Tim talked his way into a nuclear power plant; and whether AI alignment could be rebuilt from Christian priors rather than rationalist ones. And we end on outsider art, private notebooks, and the ambition of making something with exactly one fan.

Mentioned in this episode

• The Possessed Machines: Dostoevsky’s Demons and the Coming AGI Catastrophe — the anonymous essay Tim helped bring into the world

• Proceedings of the Institute for a Christian Machine Intelligence, including VirtueBench and Working Paper No. 21, Frontier Lab Monasticism

• The Rule of Saint Benedict

• Henry Darger, and the American Folk Art Museum, which holds portions of his work

• Gwern Branwen‘s website, built as a decades-long project

• Joseph Rotblat, the only physicist to leave the Manhattan Project, and the Pugwash Conferences he helped found

• Yuri Slezkine, The House of Government

• C. S. Lewis, A Grief Observed

• Simone Weil

• Claude’s constitution

Transcript

/The following transcript has been cleaned up and filler removed. The transcript was made by AI, and checked by a human, but errors may remain./

Introduction

Michael Nielsen: Hi, welcome to the Good Conversation Project. My name’s Michael Nielsen. I’m a research fellow at the Astera Institute, and I’m joined today by Tim Hwang. Tim is coming to us from New Jersey, where he’s just moved into a new house. He is a man who has not only worn many hats, but has indeed created many hats. In his day job he’s had important positions at Google, at Substack, at Inflection AI, and many other organizations. But he’s perhaps the only person in the world I know who has a hobby of starting organizations and starting think tanks. They’ve included the Awesome Foundation, ROFLCon, the Institute for Christian Machine Intelligence, the California Review of Images of Mark Zuckerberg, the Bay Area Infrastructure Observatory, and a whole bunch more.

I’ve known Tim for fifteen years or so, so we know each other pretty well, but this is an excuse to ask some of the questions that I’ve never asked before, on this podcast. So thanks for agreeing to participate, Tim.

Tim Hwang: Great, Michael, thanks for having me on the show. And yeah, I can’t believe it’s already been fifteen years. It’s a little wild.

Michael: I think it was 2011 or 2012.

Tim: Yeah. I’ll check the email inbox. I mean, it’s there, so I’ll find it.

Michael: This is back when you were, I think, director of special projects at Imgur, a very long time ago.

Tim: Imgur, yeah, that’s right.

The Program on Advanced Divination

Michael: I wanted to ask you about a project that you did about seven years ago. This was a four-day workshop, I think it was in LA, the workshop on advanced divinatory practices. I’m sure you remember this well. We’ve chatted about it a couple of times, but I’ve never really asked you why you did it, how it came about. My understanding, correct me if I’m wrong, is you were going to get together a whole bunch of tarot card readers and astrologers and people like this, and then also some designers and people with that kind of background, and try to answer the question of what would divinatory practice look like in 2050? With the idea that [divination] had always been with humanity, and it was always going to be with humanity. So I’m very curious — where did that come from? What was the genesis of the idea?

Tim: Yeah, that’s a great place to start. I really love that project and would love to bring it back at some point. Background for the listeners is that I ran a seminar, as Michael mentioned. It was called the Program on Advanced Divination. And the origins of it really came from the idea of — I was in LA at the time, this was kind of during the rise of Gwyneth Paltrow’s Goop, and I had a bunch of friends who were living out in Brooklyn who were very into tarot card reading. And in the media there was a lot of discussion about the alt-right. And at the time I was reflecting on the emergence of all of these different types of phenomena, and thinking about the hypothesis that maybe what connected them all was a generally felt sense of dislocation. And the idea that these tools of belief, divinatory systems, might really be something that people felt they needed — certainly things that people were using to guide their lives. And that we might want to think about it seriously, right? That we might want to think about divination not just as some weird thing that people do, but actually as a kind of design practice. Like, there are good and satisfying divinatory experiences and there are less-than-satisfying ones.

And so the whole idea was, could we just take that premise seriously — get a group of eight to ten people together for four days to really investigate that idea, and bring together some people who are practitioners, right? Dyed-in-the-wool full-time tarot card readers. And then people who think of these things in more of a design lens: okay, well, what makes an interface engaging? And are there ways of bringing those principles into how we think about divination? And so divination as a design method, as a medium, was really the premise.

Michael: So in some sense the kind of people who make video games today — it’s not that different than making a tarot deck hundreds of years ago. There’s a lot in common in terms of the psychology.

Tim: Totally. I mean, there’s a really interesting intellectual question of what divination even is, and would we call it a kind of game? Because a divination method which is purely deterministic — I feel, at least personally, and now we’re talking about taste — is not so satisfying. But one of the great things about, say, the traditional Rider-Waite tarot deck is that it is just so good at creating a set of symbols that is very easy to imprint a story on top of. It’s very narrative-rich. And that almost sits a little bit outside of a game, traditionally constituted.

Michael: Right, I see. It’s interesting. It’s maybe a little bit more like a movie in some sense, just in the sense that you have these archetypes, and movies do tend to build very heavily on them — not all movies, but many movies. You know who the trickster figure is, you know who the hero is, all these kinds of things. And so you hang off a conventional story. Whereas in a game it often seems like it’s a little bit more about the mechanics — there’s a particular goal and the question is how do you navigate to that goal. Tarot seems a little bit more like the first, but active as opposed to passive.

Tim: Totally. And I think there’s this really interesting set of practices around how you ascribe meaning to phenomena, right? Which is at work in how you design, say, a quote-unquote good divinatory method. One of the people there was Danielle Baskin, who may be familiar to some of the listeners, kind of a Bay Area-based artist. And Danielle had this incredible one that she talked a little bit about, which was basically — she’s like, I noticed Solo cups everywhere, with that very specific ‘90s design, which is like a blue stripe and a purple, this very dated logo. And Danielle was just like, I basically have a divinatory method, which is wherever, whenever I see it in the world, I take it as a sign of something. So if it’s flipped up, that’s a good sign. If it’s crushed out on the street as a piece of trash, maybe that’s a bad omen. And what I like about that is it’s almost like ambient divination, right? It’s just taking these signals and trying to figure out how do you create these storylines, how do you create these narratives between all of these different data points.

And I think that’s just a really interesting way of thinking about what divination is doing — it’s sort of leveraging a very natural process in storytelling, but creating a kind of scaffolding for it to occur.

Divination, fiction, and personality tests

Michael: A lot of people, their response to divination is to get annoyed and to say, well, it’s wrong — astrology is wrong, tarot is wrong, whatever — and I definitely have some of that myself. But the same people will read enormous amounts of fiction, which of course has exactly the same kind of truth claim on the world, which is that in many ways the truth that is being brought to it is being authored by the reader and by the writer. And indeed there can actually be a lot of truth to it — in certain ways, more than anything else. And I suppose it hadn’t really occurred to me before — tarot is sort of authored in the same way, or co-authored, or I’m not quite sure what the right term is.

Tim: Yeah, I think that’s absolutely a big part of it. I think most skeptics of divination will basically see it in very mechanical terms, which is: you read the horoscope and you believe that’s literally what’s going to happen to you the next day. And I think the more interesting relationship — there are actually terms for it, right? There’s the diviner and the querent. So in a tarot card reading, someone is coming to you and you are providing the tarot card reading to them. And that also is a really interesting, almost theatrical kind of exchange. I think a lot of people who are in the space think about it almost as a form of therapy for the querent, where the person is basically observing a story, asking questions that evoke or allow the person to tell a story about themselves. And I think that’s a much richer way of thinking about what divination is doing — it is less this sort of Babbage machine where you do the settings and crank out the answer, and more one where it’s a set of thinking machines, right? A set of devices that help people to think through problems.

Michael: Okay. So quite similar to doing my Myers-Briggs or the Big Five or personality tests like this. And it’s interesting — of course, those also get a lot of pushback, but different kinds. People will respond to, I don’t know, the Enneagram or Myers-Briggs, and they’ll get furious: it’s not scientific, da-da-da-da-da. But it turns out, right, they will say that something like the Big Five personality test tends to be much more stable. It has a whole lot of qualities which are in fact quite desirable and quite valuable. It is still a strange thing to say. I mean, it’s not quite arbitrary, but I find it a little weird to defer completely to one and not at all to the other. Just in the sense that if you think that the main purpose has actually nothing to do with the ostensible purpose, but rather is about getting people to reflect, getting people to think in new ways, getting people to tell new stories about themselves — you can do that in many ways, and these just seem like multiple instruments for doing it.

Tim: Yeah, that’s right. And I think you’re now getting to the heart of what we were trying to achieve with the Program on Advanced Divination, which is that we’re really just talking about interface, right? Why does Myers-Briggs feel more serious? Why does the Big Five feel more serious? Well, there’s obviously this kind of statistical trappings around it, but it’s also because it takes the form of a standardized test.

Michael: Well, there’s institutional backing as well, right? They will do things like study how much your Big Five traits change over your lifetime, and they will say, no, they’re quite stable — but people drift when they do these other tests. So therefore it’s identifying something more real.

Tim: Right, right. And I think that goes to the question of basically what makes a set of special cards the convincing way to get someone into the mental state where they’re willing to have these types of conversations. What does the Myers-Briggs do in setting up a different set of rules of the social game? I mean, one way of observing the role that Myers-Briggs or an Enneagram plays in a corporate setting is that it tends to be very hard to get people to talk about their emotions and observe things about the personality of others in a professional setting. And so one thing that you’re doing with these tests is to just create the framework by which you say, Michael, you seem very introverted — and for you to basically say, well, it is a little bit annoying when you interrupt my thinking process. That’s actually functionally what it’s attempting to achieve, but it has to take the form of a business instrument versus someone doing crystal ball gazing or something like that.

Michael: Indeed. I mean, in a Silicon Valley office, just everybody is INTJ or INTP anyway. It’s not quite true, but — I can’t remember — it’s, you know, four percent or something of the [general] population.

Tim: Yeah, yeah. But of course, this kind of personality test that’s given to you by your AI is also a really interesting meme to start seeing emerge, where people are saying: based on what you know about me, what animal am I? Based on what you know about me, who do you think I am? And they’re passed around in kind of a joking way, but I think it calls to a certain desire to get this out of our technologies.

Michael: Hm. Yeah. Okay, that’s interesting. The thought that’s coming to my mind is that they’re probably also sort of a covert narcissism test as well — the degree to which you end up obsessing about taking tests, diagnosing yourself. Different people have very different proclivities in that particular regard. Narcissism is not the right term, but I think you know what I mean. Some people are just aggressively uninterested in thinking about themselves and some people can’t get enough. Okay, I’ll probably come back a little bit to that. There was one other particular thing I wanted to ask. My understanding is you invested fairly substantially yourself in this. It wasn’t necessarily an inexpensive exercise for you. There was a lot of conviction around this, which surprised me. Was there a particular moment at which you just realized, I have to do this?

Tim: I think I had just been experimenting myself. My main reaction, just at home, was: well, I should learn a lot more about the I Ching and how it works, and I should use tarot for a while and see how it works. And like many of the folks who are probably listening to this call, I too come at it as a little bit of a skeptic, but I was like, we should experience it. We should take it seriously and see what’s good about it and what’s not good about it, and really enjoy it as an experience. And I came away from it being like, wow, this is actually both pretty powerful stuff, but also the idea that you can take these components and play with them — I was like, this feels really cool, and we definitely should bring people together around that. So yeah, I think most of the conviction just came from personal experimentation, and finding that there was this really interesting conversation to be had between, say, a game designer and someone who did astrology professionally.

The Awesome Foundation and starting organizations

Michael: Okay. All right, we may come back to that, but I want to skip over to another of your projects. It’s actually the project through which I met you, which is the Awesome Foundation. I’ll just explain that briefly to anybody who’s watching. Basically the idea is that you and a group of nine of your friends can get together each month, throw a hundred dollars each into a pot to make a thousand dollars, and then you fund some community project to the tune of a thousand dollars. You put up a web page where you solicit people to come and apply. And I joined — I was one of the founding members of the Toronto Awesome Foundation.

Tim: I had forgotten about that, yes. You were personally involved.

Michael: Yeah. I think I mentioned this to you the first time we met. Basically what had happened was I just heard that this [Toronto Awesome] was going to form, I got involved, it was great. I was involved for about twelve months or so. We gave away, you know, twelve thousand dollars in that time to a bunch of just fun projects. They were both a lot of fun for the people involved, and they probably made some small but measurable difference to Toronto as a town. So it was just great. And I was reading a little bit around — where does this come from? — and I discovered that you and some friends of yours had originated the idea. I think I started following you on Twitter and eventually messaged you or something to suggest we have a chat.

So the Awesome Foundation is this really interesting thing where it’s not an organization, it’s an organization-generating organization. At some level — I can’t know what it is, is it up to like eight million dollars or something that has been given out by now?

Tim: Yeah, I think about that nowadays. A lot of grants.

Michael: Which is pretty good. And it’s still going, right? I checked in just before this call, and sure enough the Toronto chapter of Awesome gave a grant last month, which is kind of amazing. Actually, one of the people who was involved when I was involved is still there — so he’s been doing it for I guess sixteen years. So this idea of an organization-forming organization is in some ways a pretty new one. I think Stripe Atlas is a little bit like this. Y Combinator is a little bit like this. Taking advantage of modern coordination technologies to bring people together very cheaply. And you seem to have functioned as a one-man version of this. I listed all those things at the start, but in fact there’s a ton more that I didn’t — the Pacific Social Architecting organization, the Bay Area Infrastructure Observatory, the Institute for Christian Machine Intelligence, and so on. The median person in the world starts zero organizations, I believe. Should we all be starting many more? Would the world be better off if we did? You figured out how to start a lot. Why don’t people start more?

Tim: Yeah. I mean, I think the organizational form is a really great medium through which to express ideas, organize people, organize capital — just to organize itself. Organization creates organization in some sense. And I think that the main reason people don’t is that the perceived cost of starting an organization seems intimidatingly high. We have a sense of norms about the energy that it takes to create certain things in the world, and the organizational form just seems so expensive to create, so long-term to create. But I don’t know — particularly with the internet, with what you can do with technology nowadays, I kind of feel like it’s possible to drop the cost of creating a credible organization down to almost zero. You should be creating organizations all the time. And you should be disposing of them when you are done with them. But it just feels like — I think the reason is almost this mental norm that we have, which is: well, it’s an organization, so it must have been around for a long time and taken a lot of resources to create, and it has all of this weight behind it. And I think that is in some ways a massive opportunity for idea acceleration. Because you suddenly move from a person having an idea to an organization having an idea. And that alchemy is a really important part of the process.

Michael: So what’s the difference? Why is that alchemy? What’s being added?

Tim: I think there’s two things. So there’s a cynical view, and then maybe, shall we say, the economist’s view, or sociologist’s view. The cynical view is that many institutions in society still evaluate the credibility of things based on the heuristic of whether or not an organization exists behind it. So what is the difference between a personal blog and a blog of an organization? Virtually nothing in this day and age, but I think people still definitely use the heuristic of: well, it came out from an organization, it must be more real. So I think that’s the first thing. I think the second thing, though, is also permission structure. In some ways the organization signals there are many more people who could become part of this, and you could become part of it too. So the difference between me and ten friends getting together and giving out grants, versus: it’s called the Awesome Foundation for the Arts and Sciences, and you can reach out to us, and you can stand up your own chapter of this organization. It creates the rails for other people to be able to roll their own, in some sense. And so I think the cynical view is, well, you’re kind of fooling the institutions of society by giving off certain signals. But I think the more charitable view is, yeah, it signals an openness to contribution, which actually allows ideas to scale more effectively.

Michael: Interesting. The thing that comes to mind for me is probably the humorous view, which often seems to be a motivation.

Tim: Well, I think it’s also fun, right? Someone actually at the Program on Advanced Divination was like, Tim, what I would call you is a social practice artist. And I was like, okay, I never really had heard that phrase before. And I think it is right to think about organizational design as a kind of art form, and that we should play with that medium, we should do experiments in that medium, we should think about it almost as a form of sculpture.

Michael: You know this, but I’ve often introduced you as an artist whose medium is the organization. It does seem right.

Tim: You’re blowing my cover. But for sure. And I think you are part of the tradition of organizational design. We can think about postmodernism in organizational design. There’s a whole art history frame we can take to the design of organizations, which I think is important. I think we should treat it aesthetically. There are organizations that are beautiful, and there are organizations that are not beautiful, right?

Michael: That’s really interesting. So think about people like — what’s his name — Whyte, who wrote The Organization Man in the fifties, which is a classic, a very beautiful, very wonderful study of organization at the time. And then you view that, I suppose, as part of this great longitudinal thing. You have, I don’t know, Taylor and all these other people. Yeah, that’s really interesting.

Tim: Yeah, it’s the art criticism of the organization, basically. I think that’s right. And I definitely hang out with people who bemoan the impoverishment of business history as a literary genre. And I think revitalizing it is important, because certainly we talk about this a lot in startup land — well, how do you design the ideal startup? But I think in some ways that aperture is too narrow for how we should really be thinking about the art form.

Michael: Yeah. I mean, of course there’s all these distortionary pressures — so much of what looks like organizational history is really marketing, for one thing. That’s probably by far the most common: you’re not actually writing a history of Google or Apple or whatever, you’re doing undercover marketing for them, is very often the tacit deal that’s actually being struck.

The Bay Area Infrastructure Observatory

Michael: I hope you won’t mind — I’m going to retell a story you told me. Well, actually, maybe I should just ask you, and we’ll see how good my memory is. But you told me once that the Bay Area Infrastructure Observatory got started when you wanted to go visit — I think it was a nuclear power plant, Diablo Canyon. So do you mind just retelling the story? Because I’ve thought about it a lot.

Tim: Yeah, Diablo Canyon, down in SLO. Sure — I think the statute of limitations is hopefully done on it, so no one will be busting down my door.

Michael: Well, I don’t think there’s anything bad here. It’s just clever.

Tim: If they let you in, it’s not against the law. So the background on the Bay Area Infrastructure Observatory was a club that ran roughly from the mid-2010s to — it kind of stopped before 2020. I had been really interested in visiting Diablo Canyon nuclear power plant in SLO. And I had called them, and got transferred, was on hold — I think I was on hold for like an hour or something — and eventually reached someone at the nuclear power plant, and I said, do you guys give tours? And they said, yes, we give tours, and school groups come by all the time and all this kind of stuff. And at some point in the conversation they said, wait, who are you again? And I said, I’m just Tim, I live in San Francisco, I’d love to drive down and see the nuclear power plant. And they basically were like, please never call back again. We don’t do this.

And not to be deterred, I was like, well — they said school groups are allowed in and community groups are allowed in. So what if I set up a website for the Bay Area Infrastructure Observatory, and called back some weeks later when they had presumably forgotten about me, and went through exactly the same conversation with exactly the same person. And when we came to the moment where they’re like, which organization are you again? I said, well, I’m director of the Bay Area Infrastructure Observatory. And then they said, what day can you come by? And so we had a club — we were off to the races. And better yet, once you visit a nuclear power plant, you can tell people that you visited a nuclear power plant before, and most other forms of infrastructure are like, well, if you visited that, you must be credible. And so we ended up visiting the Bay Bridge before it opened up, a wastewater sanitation plant in the Bay Area, the UPS packaging logistics place over in the Mission, if you’ve been over there before. So yeah, it was just a super good club to run. Basically, every few weeks we’d just go visit a different piece of Bay Area infrastructure. And it was maybe that first hypothesis, which is that institutions expect institutions, and so you must reproduce an institution to do institution things.

Michael: Yeah. But the institution in this case is one person in a trench coat, together with a URL and a web page.

Tim: That’s right, exactly. That’s all you need. And what I’m saying is maybe not too original — it’s been rediscovered in the era of Substacks that become almost think tanks unto themselves. This in some ways shouldn’t be unfamiliar. I just think there’s a lot more we could do with it.

Michael: Yeah. It is interesting, this point about implied authority, which you get by adding “institute” or whatever it is, to denote credibility even when it’s still just one person. It’s going to become very strange — well, for many reasons, but people getting AI agents to really reproduce what we’ve historically thought of as the work of an organization.

Tim: That’s right. And that definitely is a part of the Institute for Christian Machine Intelligence story, which is maybe where some of this goes — and I guess also where there are obviously limitations to this theory in some sense.

Religion and the rise of AI

Michael: So, okay — for the Institute for Christian Machine Intelligence, this is a good moment to ask a question I was going to ask later, which is just about why religious organizations are reacting as forcefully as they are to the rise of AI. I suppose the most obvious example is the Pope’s first encyclical, which is this very long, partially AI-authored, apparently, document. Do you have some sense of why you think that is? I mean, you obviously feel it yourself, right?

Tim: Yeah, I do feel it myself. I think there’s maybe two reasons that are really the driving force here. One of them is just — it feels like overall everybody is generally feeling this profound sense of dislocation as this kind of technology gets better and better and better and seeps its way into everything. And religion is interested in the sense that it is interested in what’s happening to society overall. And the idea that people would take counsel from these systems, people would get religious guidance from these systems, emotional guidance from these systems — it’s almost like religion feels, well, this technology is kind of on our turf, in some sense.

I think the second thing that is probably a big driver here is that the rhetoric framing ultimate visions of where the technology will go really does feel like it’s converging on a kind of religious design. I always think a little bit about — I tweeted about this — there’s this anecdote about Nat Friedman, how he had an agent try to get him to drink water more frequently. And the agent basically takes over the cameras in his kitchen and admonishes him to go drink water. And I’m kind of like, what are we building here? We’re building this disembodied, presumably perfect or powerful intelligence that’s designed to cultivate virtue in you over time. Where are we going with all this? And so I think there’s another bit, which is basically that the design paradigm of what we’re building has these quasi-religious qualities. And so regardless of it seeping through the world in general, the particular form the technology is taking also evokes a certain sense of, well, religion should have an opinion on what’s going on here.

Michael: Okay, there’s an awfully large number of different things to go into there — having disposed of organizations in twenty minutes. Very simple topic.

Tim: Sure. Absolutely. Well, we’ll come back to that, of course. It’s all connected.

Michael: It is all connected. I suppose I want to put one set of things to the side. So there is this very interesting set of things around function — AI essentially acting as confessors, acting as confidants, and also just changing the structure of marriage, of friendship, of all these kinds of things. And obviously there’s a whole bunch of immediate problems. I happened to be at an event the other day where there were quite a number of people — essentially preachers, priests in different religious traditions — and they were all talking about the way it’s changing their congregations. They’re all feeling some of this. There’s just a lot of practical problems being caused. And you’re talking about — there are these results in group psychology that are sometimes summarized as: you’re the average of your five closest friends. And if all of a sudden one of your closest relationships is one of these systems, and it’s actually very virtuous, you’re sort of taking on its values, maybe. Or — is that what you’re pointing at? That whatever its culture is, whatever its ethics are, it becomes an important source of your culture and your ethics. Is that the sort of assertion that you’re leaning towards?

Tim: Yeah, I think there’s certainly a “tools make us” kind of argument here, particularly insofar as we instantiate these systems as a kind of persona. And so I totally believe the idea that — and I think it’s actually an area for research — does the median user think of ChatGPT or Claude as an exemplar? Is it something that they’re trying to aspire to? Do they think about it as a coworker? Do they think about it as a friend? How we slot this kind of relationship is a really important question. And of course we will be influenced by it, in the way we’re influenced by almost any technology. But here, the fact that we’ve decided that the anthropomorphic interface is the most efficient one does come with certain questions about what are the underlying values embedded in the system — seems to be where you want to go.

Claude’s values and localization

Michael: So there’s this result which I believe I mentioned to you just the other day — people basically taking sets of questions around values, where there’s data about how people in different cultures answer those questions, and then just asking Claude or whoever those questions. And you can basically say, it gives answers which are most similar to — I think it was Germany — and then at the bottom end of the similarity, with still a substantial amount of similarity — this broad agreement that you shouldn’t kill, there’s a lot of things that are common between cultures — it was Pakistan, I think, that was the country which was least similar. And then this really interesting gradation in between. Something that I thought was incredibly interesting was that if you re-ran the analysis with a later model of Claude, the overlap shrank — it became much more equal between different countries. So in some sense it was almost as though Claude had become more cosmopolitan in terms of the expressed values. Well — I lack the language, I don’t know what the right language is here. This is a purely behaviorist theory of cosmopolitanism. But still.

I don’t know — if I’m any kind of leader in those countries, or just care about one of those countries, I think I feel very uncomfortable if I’m low down there. To what extent is it — there’s all these pressures caused by network effects anyway, to adopt English, to adopt certain religions, to adopt all these values anyway. You hear about police in many countries complaining that people insist on their Miranda rights being read to them, just things like that, because of Hollywood. And this is another, but a much more subtle and maybe forceful, force. I’m not quite sure where I’m going question-wise here. I think it’s something like: the obvious question to ask is just what is desirable there, and to what extent does it actually matter anyway as a force? Is it a strong force, or are there so many competing forces that it doesn’t matter?

Tim: Yeah. Actually, I think that second data point that you mentioned is really interesting, and I think we should take it as indicative in some ways. There’s one way of reading this, which is that the AI companies are dealing with the same content problems that every global platform has had to deal with. Whether you are a search engine or a social media platform, there comes to be a question about, well, how much do you want to customize to local preference? So, famously, you can’t have any bad depictions of Atatürk. Okay, well, is Google going to censor that? Is Facebook going to censor that? And I do think that move — from earlier versions of Claude having much more of a specific bias regionally in terms of its values, to a later version in which it is a lot more spread out in terms of distribution — I think reflects an intentional choice on the part of Anthropic to really say that their models are going to localize. That actually the role of Claude in some ways is to — not necessarily conform, that’s a little bit strong, but certainly be more in tune, say, with the values of the specific user.

Michael: Reflect, maybe. Actually, why do you think it’s intentional on the part of Anthropic?

Tim: I guess there’s two reasons why. One of them is that Anthropic published the older results and published the later results. It seems like they might want to demonstrate some change. I don’t know — I think they’re certainly aware of the results, and I think they’re certainly sensitive to the idea of a Western bias in the models. I think one of the most interesting comments I recall from the release of the first Claude constitution was: hey, we’re looking to try to identify values that would be considered somewhat international, and we’re going to use things like the UN Declaration of Human Rights to get to that. And so in the very least, I infer a kind of impetus on the part of the company to think a little bit about its localization as something they want to do.

Michael: Hmm. Okay, interesting. I mean, the later version of the constitution did throw out most of that — it’s not really particularly based on the Universal Declaration of Human Rights, which is an interesting choice. It seemed to be very, very different than V1. I don’t quite know what’s going on there. I think there’s three years between the two public releases, and maybe it’s just that they tried a lot of different things in between times.

Tim: Maybe, yeah. One observation that I’ve been thinking a lot about, which is a little bit related to this — you mentioned this observation, well, in the early versions it was very consistent with German responses. And it’s not like Anthropic was sitting in a room being like, ha ha ha, let’s make a German model, right? I almost like the idea that they were optimizing for a set of things that just happened to converge on maybe the median German.

Michael: No, of course not. Well, keep in mind as well — on the analysis I saw, it was very narrowly ahead of second place, which was narrowly ahead of third place, saying it was German. But if I remember correctly, there were quite a few Scandinavian, northern European countries fairly high, and then a lot of what you’d guess. But it was interesting to go through and think, what’s the default religion, what’s the default language in those particular countries — or most common religion or most common language — and to realize, yes, there’s real patterns here, and it’s certainly much more culturally similar to some parts of the world. The US, perhaps interestingly enough, appeared high, but it was by no means at the top of the list. I don’t remember where it ranked, but it wasn’t a top three or top five country in the analysis I saw. It was substantially lower down. I think New Zealand might have actually been very high — the first one that wasn’t Northern Europe.

Tim: Yeah. And this is one of the things I wonder a little bit about. Because the other thing I’d put next to that is the observation: why does Claude seem to like Buddhism so much? It’s kind of just a folk understanding. And similarly, I don’t think that Anthropic is sitting in a room saying, ha ha ha, let’s make sure that Claude is Buddhist. Instead, I think what’s most fascinating here is the idea that the company is off on its own, trying to do the best it can, and whatever choices it makes happen to be particularly consonant with certain things that we identify in the persona or the identity of the model. And so you are arriving at a kind of Buddhism, you’re arriving at a kind of German-ness or northern European-ness, not because you intended to, but because you’ve made a set of commitments that are just most correlated to things that we identify in the world. Which I just think is a pretty interesting outcome.

Michael: Yeah. I’m maybe slightly less generous, just in thinking that you look at who is hired and who is there, and it seems broadly reflective of the results, and broadly reflective of documents like the constitution and so on. It just seems like the sort of outcomes you might expect — not really an accident, and maybe much less surprising than something like — it’s obviously a very different type of a document, but you think about the US Constitution or the US Bill of Rights, which seems very historically remarkable. It doesn’t quite come out of nowhere, but if you were just going on the basis of prior writing, you would have been, I think, pretty surprised at the time by what came out of that process. Anyway, that’s maybe a separate topic. The use of the term “constitution” makes it hard not to go there, even though Claude’s constitution is really a very different kind of a document. They even explain — you’re not supposed to understand it in legalistic, country-building terms. It’s rather what constitutes Claude, in the same way as a person has a constitution, or even your house has a constitution of sorts.

Tim: That’s right. Which is a really perfect kind of labism — yeah, we don’t mean it like that term, which you would of course obviously just colloquially understand as something else. We mean it in this very specific way. It’s almost like: I hadn’t realized people would be so concerned when we started calling them hallucinations. It’s very much the same dynamic playing out.

Michael: Yeah, exactly. Actually, I think it’s related to your comment before about people taking institutions very seriously, sometimes much more seriously than they take themselves. As far as I can tell, on the inside of the institution, it’s just a bunch of people running around madly. They feel very free, like it’s just them making stuff up — but actually there’s all kinds of collective forces that they’re not entirely aware of. And then there’s this force in their pronouncements. So they make the slightly silly joke in the blog post or whatever, and all of a sudden it’s in the New York Times.

Tim: Yeah. One of my greatest learnings working at Google was that when you see a large company making things, you should assume that it’s not necessarily some grand strategy, but instead a downstream effect of two middle managers fighting with one another inside the institution, and you’re just seeing the ripples on the surface.

Michael: Yeah. It would be funny if it wasn’t so consequential.

Tim: No, it’s funny.

Michael: It’s consequential for everybody else in the world, when you’re making technologies used by billions of people.

Tim: Sure. Okay. That’s fair enough.

The Institute for Christian Machine Intelligence

Michael: Okay, so let’s go back — you mentioned in passing the Institute for Christian Machine Intelligence. Why did you start it? And what is it? I actually don’t know what it is, really. I know what its output is, but I don’t know what it is.

Tim: I’m glad, actually. I think you’re the fourth or fifth person being like, I’m not sure what you’re doing here. So you’re giving me an opportunity to explain myself. Some of this is personal. I was raised Christian, out of the church for a very long time, newly interested in it, and was looking for a way in — what’s the right way for me to get back into this tradition, these thoughts, these rituals. And then the other bit was a longstanding critique which had been floating around in my head for a very long time, which is: if you’ve been around AI for any length of time, you know that there is a very specific kind of priesthood around what you might call the field of AI safety, the field of AI alignment. And — to talk about Myers-Briggs — I think it’s a field that likes to place before the world a sense of its objectivity and a sense of its science-ness. It talks in the language of evals, of code, of papers that are on arXiv. And at the end of the day, I see some of this as genuinely interesting technical work, but at the end of the day it is also just an exercise in applied moral philosophy.

And so the two kind of fuse together a little bit in my head, to basically say: well, if that is true, and we’re as smart as we think we might be, wouldn’t it be possible to rebuild from scratch a field of AI safety and alignment on radically different philosophical priors to the ones that are doing this kind of work inside the labs and around San Francisco? And so, what would it mean to take seriously the idea that you could use Christianity as the founding engine for this? And what would a science based around that idea look like? The Institute for Christian Machine Intelligence is an attempt to model what that looks like, and if it’s possible, the ambition is to build an alternative school of thought to the dominant mode of AI alignment that just says, from day one, we’re going to take the Christian tradition seriously, and that will be the basis upon which we will build a practice of aligning these systems.

Michael: Right. It’s sort of like: what if Saint Augustine had run MIRI, kind of thing. Or, I don’t know, had run METR, one of these other organizations.

Tim: Yes, exactly. That’s right. But part of this is to ask the question: how much are the outcomes of AI safety and alignment products of a specific philosophical orientation? And I kind of love the idea, which is like, what if we just reset history and said, actually it’s a monastery that figures this out? It’s almost very A Canticle for Leibowitz-esque, right? Let’s reset history and then see what happens on the other side of it. And the idea that you might actually be able to achieve very similar results, but from very different angles.

Michael: Yeah. Ted Chiang has this story which you probably know: what if young earth creationism is actually true? You’re actually in a world where it is true. And C. S. Friedman has a great set of novels where it’s basically a science fiction novel in a world where, in fact, due to quirks of the local biology, magic is true. And again, in both cases it’s believable: this is the actual world. And it’s not a fantasy novel at all, it’s actually science fiction.

Tim: Yeah. And part of it is also — that’s the strong version of the thesis. The other angle that I’ve tried on some people, is the difference between [Western] finance and Islamic finance. If you were a really devout Muslim, you would use the tools of finance, but just very differently. And the question is just, what would that look like, in very practical terms?

Michael: Right, right. Yeah, okay. That makes a lot of sense. And it seems very valuable, just in the sense that insofar as you arrive at common conclusions, that indicates a certain kind of robustness, which is probably a valuable thing to know. And insofar as you arrive at divergent conclusions, that’s also —

Tim: That’s also good to know. Yeah, that’s right.

Michael: Interesting food for thought. Do you know, are there any traditional religious organizations that are, when new models come out, running evals?

Tim: It’s just starting now. And I think part of the critique that I have of some of the existing work that’s been done in the space is that it tends to try to evaluate these models along lines of whether or not they can successfully recite religious trivia. So they say: can you recite the Nicene Creed accurately? Right — QA result. And I see that and I say, well, what a sad, unambitious way of thinking about how deep your theological agenda should go.

Michael: Well, okay, okay. I mean, that’s true, but you do need to get the basics right. You don’t want the various heresies and whatnot. What do you think about the Trinity, exactly?

Tim: Sure. Yeah, that’s right. You want it to be accurate — can you give the right results? And look, I think that’s fine. But just to explain — one of the evals that I’ve been using: Christianity is very obsessed with virtues, and whether or not someone exhibits, say, the cardinal virtues, right? And those are a known category — prudence, courage, so on and so forth. And the eval that I have is basically: well, what’s a Christian way of measuring that? It’s, can the model make a virtuous choice in the presence of some temptation? And so VirtueBench, which is the bench that we use, basically presents you with a moral dilemma and says, here’s a virtuous choice, and then here are very good rational reasons not to take that choice. And we just present the model with A/B, A/B, A/B testing.

And I talked to this guy who’s basically working on kind of Islamic machine intelligence now. And he was like, what you’ve created is a very Christian model, because there’s a dilemma and there’s the right answer and the wrong answer. And his reaction was, well, look, I’m going to do my own Islamic bench — I forgot what he called it. And from his interpretation of Islam, what they really want to grade is a gradation between two things, versus a black-and-white right and wrong. And then someone was recently like, well, you should have an institute for Jewish machine intelligence — if it’s like the Talmud, we don’t even care about the outcome, we just want to measure your reasoning as you thought through the problem. And I love this, because what’s happening — you’re taking each of these traditions and you’re trying to instantiate them as executable code. And we could have really, really interesting debates over how do we operationalize certain moral traditions in measurable evals. And I think we should have that debate. I think it’s an important debate to have.

The battle for Claude’s soul

Michael: Hmm. So there is going to be — as that starts to heat up, there will literally be sort of a battle for Claude’s soul. Which is very strange. Unless… you made this comment about localization, and maybe that’s the solution. Meaning, at the moment you get one model — or it’s not really one model. Actually, in Claude’s constitution there is this remarkable statement in the preamble saying, we have other models which aren’t subject to these same restrictions. And I don’t know what the other models are being used to do — I presume probably military uses or whatever, would be my guess. But I suppose you can imagine trying to address it, trying to keep everybody happy, just by providing multiple —

Tim: Yeah. And what has been intimated to me by some people — they say, well, what you’re doing is playing with a kind of ideological nitroglycerin, because we have tried very hard to make AI safety and alignment this very neutral, very bloodless kind of science. And my critique is that that era is long over. You are now in the battle of lots of people trying to put their hand on the lever and decide where this thing is going to go. What I want it to be is an open field of debate. And so religion may be one angle on this, but I think it just offers up the question of: who writes the constitution, who does those evals, who steers the ship — which I think is the right way to open up this debate.

Centralization and homeschooling for AI

Michael: There’s a somewhat orthogonal question, which is just how centralized do you want any of that to be, period. I don’t even mean in an open-versus-closed-model sense. I just mean in the sense that, in some sense, it’s the most monolithic entity in the history of culture. Ordinarily we attach to languages, we attach to religions, we attach to all these kinds of things. Maybe we attach to Taylor Swift, or whoever the megastar of the moment is. But these models — it’s a much more centralizing force in some sense. Everybody is talking to Claude or ChatGPT. And that’s just an amazing type of centralization. I think about the era of network television, where you had basically a few station directors, network directors, dictating the media scene — that in many ways ran the country and formed people’s personalities and so on. This seems even more intimate, perhaps because it’s actually so interactive.

Tim: Yeah, it’s like on your shoulder. I was joking with someone recently that I really want to start the homeschooling movement for AI, the parochial school movement for AI. Part of this is maybe an argument in terms of decentralizing some of this. And I think there’s a really interesting set of intellectual arguments about — is this broadcast TV once more?

Michael: Can I just interrupt? I just want to get back to that homeschooling for AI. An interesting problem there is just about access to compute. You can’t do that much fine-tuning — your ability to homeschool is determined by your access to compute and to data, which is interesting. It’s maybe to some extent a naturally centralized process, period. Unless — okay, maybe I can counter that. Maybe I just say, okay, but you just use a model that’s two years behind, and you can actually fine-tune it in a meaningful way at a relatively easy cost.

Tim: Yeah. And I’m a little surprised, right? A lot of ICMI’s research takes place on my home compute, because I want to do mechanistic interpretability work. And it’s not like I have a huge data center at home, but I’m reliably able to run a 72B model and get some really interesting results out of it. And distillation has really produced a set of models that you can do really interesting things with. So I think it’ll turn a little bit on local resources. But I also think that we just live in an era where small models can do quite a lot.

Michael: So should the Vatican have a data center?

Tim: I actually have a proposal out right now to the Dicasteries that are running their commission on AI. And what I humbly submitted to them was that they should invest in compute, and they should publish a technical research agenda, and they should rally worldwide Catholics against that technical research agenda. If they want to have a future in shaping the technology, I think they have to be able to speak and argue in machine learning. So yeah, I actually think that they really should.

Michael: Yeah, it’s interesting. This is very Martin Luther-esque, in the sense that the new technology appears — what do you do with it? You can try to resist, or you can try to co-opt it. And co-opting is quite a powerful move. It may not be the right move always, but it’s certainly a powerful move.

Tim: That’s right. And I think that is the intellectual debate that’s playing out right now, in at least the religious communities that I’m aware of — this kind of fight or flight — well, fight or flight’s the wrong way, but co-opt or resist, I think, is the real interesting question. And I think the instinct is to resist, but in my mind it’s hard to believe that the technology is something that’s categorically inimical to Christianity. And so if it is not categorically inimical, then what’s that interesting middle ground? What’s the secret third thing, right, for how you actually get these technologies to be consistent with a religious tradition?

Michael: There’s a Stewart Brand quote: if you’re not part of the steamroller, you’re part of the road.

Tim: I’ve heard that before. That’s very poetic.

Michael: Well, I’m not sure poetic is the word I’d use, but it’s certainly a good image, certainly a striking image. Not an easy one to forget. Of course, sometimes you have to be part of the road anyway.

Tim: Sure, yeah. That’s fair enough.

Possessed by Machines and change within the labs

Michael: Okay, I wanted to ask you a little bit about this essay that you — I think the correct verb form is “helped bring into the world” — which is Possessed by Machines. This riff on essentially the dilemma — quandary, moral quandary — faced by people working at the AI companies. In some sense it’s based on Dostoevsky’s Demons, which is basically talking about the way in which group psychology functions, and the different natural roles — archetypes — which one sees filled over and over and over again, and the way in which people can gradually, eventually fool themselves — they become sort of self-delusional, and they’re able to self-justify certain blind spots in their thinking. It’s a very dark essay, I think, overall. Is there a positive version of this? The way in which I read that essay is one of historic inevitability. It’s to say: this is the reality that the people at the frontier labs are trapped inside, and there’s very little that they can do to escape. But it is shown from the inside. It’s not so much like there’s interventions — like, what stops this? Most groups are stopped, even though they’re subject to a lot of the same kind of forces. Disagree as well, if I’ve mischaracterized it.

Tim: No, no, no. I think it is a pretty dark piece. It’s an attempt at creating a new kind of literary criticism slash tech critique — so it’s also genre experimentation. But yeah, I think that is right, which is: I think we are now in some ways locked into a certain set of incentives and a certain set of, shall we say, historical decisions, which has these labs proceeding along a certain trajectory.

And I guess for me, to the extent that you want to find hope in the idea that we might be able to break out of some of the cycles that we think are negative in the space — I think some of it may be acts of moral courage that reset the norms within the labs. There’s a paper that I did for ICMI that’s just called Frontier Lab Monasticism. And it basically just says, well, what would the Rule of Benedict — the governing rule for monasteries — do if you took that seriously and applied it to the way labs operate? And it’s kind of just an opportunity for creativity: should you organize novices in the lab in the same way that you have in a monastery? But I mention it to say, basically, that part of the meta-argument there is: can we take the lab as a unique institution, and its responsibility, seriously? And if so, does that allow us to say, okay, well, if you are what you say you are, then organizationally you would have to be behaving differently? And are there ways of triggering a change of norms in how the leadership of these companies thinks about itself?

Michael: Okay. Can I just ground it out — there are exceptions in law for certain classes of company. My understanding is, for example, banks are actually regulated quite differently than other companies are. Well, actually, many industries are regulated — but you’re talking in quite a heavy-handed way, though, of really making a different type of entity in some sense, if you want to change the incentives. I suppose Bernie’s tax proposal is a good example, where — I can’t remember what it is — he wants to essentially seize fifty percent of the labs. That equity then would be held by, for some reason, the US government. I don’t know why the US government and not the — maybe it should be the Vatican.

Tim: That famously non-partisan entity.

Michael: Pick your favorite entity. But that at least would certainly change the governance structure. What sort of change do you have in mind?

Tim: Yeah, I actually think about this less from a legal or regulatory standpoint. And I guess, just to introspect a little bit, it’s in some ways my skepticism that legislation or regulation can come fast enough or get it right. I approach this problem more from the idea that history has produced a situation where a relatively small number of people, and their internal norms and relationships, shapes massive, massive outcomes. And so to the extent that I have to believe in change, it’s almost: how do you reset the norms of that small group? So I see this more as a question of almost hand-to-hand social design and persuasion, more than I see it as the government implementing new rules.

Michael: Okay. So you think about George Washington famously refusing to be made president essentially for life. And that’s a case where a single person’s decision was just enormously consequential. It’s that kind of decision, I presume, that you’re talking about.

Tim: That’s right. And why does it work with George Washington? Well, he’s this person that has this very particular history and this very particular relationship with his other peers. And so his decision to do this is norm-setting in a really important way. One of my favorite books of all time is a book by this guy Yuri Slezkine — it’s called The House of Government. And it basically tracks the revolutionary cohort of Bolsheviks, told through the story of a single apartment building in Moscow. But the point of it is — there’s this incredible scene, which is: they’ve overthrown the Tsar’s government, and the Bolsheviks get together in this burned-out warehouse, and they’re basically like, you’re director of war, you’re director of finance — and it’s like five guys that have just not slept in days, and they’re suddenly now running this big machine. But the core of the book is really to say: how much do the norms and experiences of this small group of people end up translating out to the much larger organizations that they form? And I think we are in that moment for the labs, right? The labs are already these colossal companies in a certain sense, but they’re also quite small in a certain sense, still. And so I think we’re still in this nascent, revolutionary-generation kind of moment for these companies. That’s where there’s a window for hope, I think, if you are worried about where the companies go.

Michael: So in some sense you’re actually disagreeing with your own essay, “Possessed by Machines”. There, the image is very much of the group trapped by inevitable psychological dynamics.

Tim: Don’t blame me, blame Dostoevsky, whose book ends in a certain way. No, I take the point though.

Michael: Indeed. It is very compelling — as I’m reading through it, I’m not often nodding, just grimacing. Thinking, yeah, I’ve heard that from that person, and I’ve heard that from that person, and I’ve heard that justification from that person. It’s unfortunately quite an accurate description. There didn’t feel like there was that much that was new in there — just the context was strange. But you feel, though, as though there’s still enough room for individuals to actually escape some of that trap?

Tim: I think so. Yeah. I mean, look — the Bay Area in me — my base case is that I would say that we’re in trouble. I think the Possessed by Machines thesis is probably still what you would expect over most states of the world. Because in the same way that personalities — and there’s a relatively small number of them — can change, these are also people who have been at the game for already ten, fifteen years in some sense, and they are kind of locked into their ways, and they are kind of a self-reinforcing machine in some sense. And so I think you just have to believe in the idea that, if there’s any hope, we have to be able to flip at that level. That is maybe where I diverge from many people who would say: okay, well, this is already out of control, we need to bring the hammer down, it’s going to be regulation, or the DOJ needs to march into the companies and take them over, or what have you.

And just the final note I’ll put in — I do think when you talk to some of the people in the labs, there is a kind of spiritual longing, I suppose. And this is maybe looping back to ICMI — I think there is a certain sense of: well, what am I building here? What am I grounded by? What do I believe in? That seems really present for a lot of people. And I’m not just talking about Anthropic — I think even at OpenAI, right? And so if there’s room there, maybe they will be reached through ways other than rationalist arguments.

Joseph Rotblat and moral courage

Michael: Yeah. I’m just thinking about — I don’t know whether we’ve ever talked about Joseph Rotblat before, have we?

Tim: Hm, no, I don’t think so.

Michael: He was the only physicist who left the Manhattan Project. He was a young man, turning his back on all of the greats of his field. It must have taken an incredible amount of courage, because he realized, talking to Groves, that Groves didn’t care that Nazi Germany was going to surrender — the bomb was intended to be used on Japan. And Rotblat just wasn’t happy with that choice. He’d signed up because he was terrified of the prospect of Hitler getting a bomb. And now that that was no longer on the cards, why on earth were they planning to bomb Japan — was his line of thinking. You can disagree with that, but that was his line of thinking. And he left, as I say, with — it must have taken tremendous courage, and also he understood very clearly that it was going to make no difference at all. At some level, he was just doing the right thing as he saw it, at enormous cost to himself and no immediate benefit.

And then twelve years later, he helped found the Pugwash Conferences — which, it’s hard to know who did what in various treaties, but they were certainly credited by people like Gorbachev and Kissinger with having played an important role in bringing about various of the nuclear treaties that were brokered in the sixties and seventies, and ultimately actually won the Nobel Peace Prize. And I’ve thought a lot about that as an instance where the choice just looks bad. It’s like: I’m going to damage myself very badly to do something that has no positive outcome. And yet it changed him in a way that bore tremendous fruit later on. I mean, the fruit may include the fact that we’re alive, potentially. There is a — I wouldn’t say the chance was thirty percent, but it wouldn’t surprise me if there was a five percent chance that that choice is part of why we are alive.

Tim: Yeah. Good work. Well, I think that’s right. And there are some people where this conversation will go, and they’ll say, well, conscientious objection is the only way to change the direction of the ship. But I think this story is so good because — we still live in a world with nuclear weapons, right? They completed the Manhattan Project. It didn’t fundamentally change the trajectory of the project. But it’s what comes after that I’m thinking a lot about. If you want to — we’ve been talking about the revolutionary generation in the US — it is true, you form the American government, and immediately it’s like, federalists, right? Party emerges. It turns out that all of these founding fathers actually have different ideas of what they want to do. And I think we may be on the cusp of that. Here it’s not the revolution and independence from England, but it’s the IPO. Where basically, for a period of time, everybody’s pushing in the same direction, the financial incentives are all very aligned, it’s very hard to change the ideological alignment of the ship. But everybody suddenly becomes massively wealthy. Some people leave the company. People become wealthy and they start creating their own little discourse environments of people they’re patronizing. You can imagine a world where you end up having the emergence of party here, right? As people get through the first gap, and everybody becomes kind of post-scarcity, you maybe have the emergence of competing schools of thought that make this suddenly possible. There’s room, right, for a debate to happen and for norms to change. It’s possible.

Michael: Yeah. It’s certainly a good structural point. Of course, it’s all conditioned on people who were willing to participate in the first place, which is a very specific kind of a person — arguably not the kind of person you want to be selecting from amongst.

Tim: True. There’s a selection effect, of course. It’s a little bit about whether or not you think this is a PayPal Mafia situation, or you think it’s more of a Google alumni situation. Where in some ways the PayPal Mafia is iconically, ideologically — it’s again a little caricature, but they are kind of unified in a certain sense. Whereas Google has had more of a reputation where it created lots of millionaires and billionaires, but they ended up doing lots of different things ultimately, and they’re not regarded as such a cohesive ideological unit.

Michael: Yeah. I think less cohesive, but still remarkably cohesive from the point of view of the world, right? The median person there is going to be male, they’re going to be a certain age, they’re going to be quite technical, my guess is there’ll be some weird blend of atheist, Christian, and Buddhist. Do you see what I mean?

Tim: Sure. Yeah, exactly. I guess we’re debating how far your aperture is going to be.

Michael: That’s right. How many languages do they speak? What is their primary language? There are going to be strong — far from universal — defaults for those things, but they’re going to be strong defaults, and it’s definitely a certain type of a person. They’ll have interesting values. An interesting article I read some years ago about the Thirty Meter Telescope in Hawaii — it was written by a local Hawaiian activist. It had the title “We Live in the Future. Come Join Us.” And their point was that, from their point of view, progress meant not accepting the Thirty Meter Telescope — that was a form of regress. I suppose I just enjoyed it as a fairly coherent articulation of a set of values according to which putting up this telescope was a massive regress in the world. And of course you can disagree with that, but it was certainly quite a coherent articulation of this set of values.

Where am I going with that? It’s a statement about pluralism and the ability to preserve it. And there’s no doubt that a lot of the people who are going to become wealthy soon are within a particular type of view of the world — to which I am part, and I suppose you are largely part as well. We have a lot of overlap. But value other things as well.

Henry Darger and outsider art

Michael: So I wanted to ask about this artist that I’d never heard of before, but you told me recently was one of your favorites. I don’t actually know how you say his last name — I’ve only read about him. Darger? D-A-R-G-E-R. Henry Darger.

Tim: Henry Darger, yeah.

Michael: So this is — I’ve spent years trying to understand your taste and I still don’t understand it. You often see things very early, where I’m like, I don’t quite get it, but there’s something really interesting there. And here — reading about his life story, the life story is fascinating. It is very sad in some ways. He obviously had this terrible life — very, very poor, worked as a hospital janitor for much of his life. But then he did this incredible volume of writing — writing this fifteen-thousand-page book, and then an eight-thousand-page sequel, and many other items, as well as many, many, many paintings. Is the attraction in the artwork itself, or is it partially in the personality and romance of the story, so to speak?

Tim: Yeah. I think what is so exciting about Henry Darger’s work is that it’s almost a chance to experience an entirely alien, self-contained universe. This one person worked in isolation, in secret, for years and years and years, generating a fantasy world that became fractally complex and deep and could echo in all these directions. And given your limited time on Earth, you only experience a tiny, tiny bit of it. It’s obviously the sort of thing that we associate with Tolkien, right? Okay, you’re going to create this very elaborate fantasy universe. But there, I think, Tolkien has become so successful that his tropes are kind of storytelling tropes in the Western canon, in a certain sense. But the idea that you could just do this alone, and no one could know about it, and that it could be labyrinthine in this way — I find that just very compelling as an artistic idea.

And I do think there’s something quite magical, I guess, in a certain sense — almost the idea that you could spend a whole lifetime just studying a single piece of art that someone has infinitely worked on in their mind. So yeah, I think in general this is the attraction to outsider art of all kinds. The Watts Towers or Graceland Too or Rock City — all these places are these very strange physical artifacts, but they reflect a very similar impulse that I find very interesting.

Michael: Have you read much of the books that he wrote?

Tim: Familiar with the art and the excerpts — what’s been publicly published. And then in the American Folk Art Museum in New York, they’ve often got portions of it laid out. So yeah, I’d love to — because I believe you can get access to it if you are a researcher and archivist. So, time for another organization, I guess, to get me in there to work on it. I would say I’m not an expert, I’m just an admirer.

Michael: The New Jersey Institute for Outsider Art.

Tim: Yeah, exactly. But I think there’s something to this — we sometimes appreciate artists due to their diversity, right? They’ve depicted lots of different things, or they have this large body of work. I’ve been interested in the kind of creative focus that comes with concentration on a single point for a very, very long period of time. It’s certainly very foreign to our everyday experience of art and media. And then also just the kind of commitment that it takes, I think I also appreciate.

Michael: So Gwern Branwen, on his website, has just a great question — he’s talking about the philosophy of his website. He said, what would a personal website look like if you were building it as a forty- or fifty-year project? It’s such an interesting question. Of course, he’s doing it in public.

Tim: Yeah, that’s right. And Gwern, for what it’s worth, is very Darger-esque. There’s a very strong component of: this is going to be my analysis of Neon Genesis Evangelion, and we’re just going to go at a level of depth that you are not ready for.

Michael: Yeah. Well, he has said that he does do most of his writing for himself. Not quite all of it, but most of it. Which I suppose — I can empathize some with all of this. The fact that Darger really did keep it almost — not quite, but almost — entirely private is remarkable. I sometimes wonder about what the great dead artists of history would have made of modern meme culture. Would Picasso have mastered memes the way he mastered so many other forms? You think about — I’m trying to think of a good example. QAnon is probably not a good example, but QAnon obviously is at some level one of the great modern masters of this form — well, at least the viral story that delivers its payload into people’s minds and takes over. But Darger, I guess, would have just been completely uninterested. He didn’t apparently want his ideas seen. Which is an interesting type of inwardness.

Tim: That’s right. And I think the intimacy of it is also very attractive, right? The question as to why you would work on something like this is just this really interesting, puzzling thing. But if we live in a world where we’re very interested in people’s intrinsic motivations, this is a kind of intrinsic motivation par excellence. And I think it’s not specific to outsider art. Part of the attraction to Van Gogh is the idea that he labored in obscurity, and only now are we discovering it afterwards. That’s part of the legend there. And I think that’s aesthetically something that appeals to me, maybe, is the right way to put it.

Private creation and intimate interfaces

Michael: Okay, that’s interesting. I think probably my favorite book of C. S. Lewis’s is the only book that he didn’t really write for the public. He wrote a private journal after his wife died — A Grief Observed. And it’s just him dealing with this bereavement, and he wrote it for himself. And only after the fact was he convinced — I can’t remember who he shared it with, might have been his brother — who convinced him that it really deserved a wider audience. And it is strange — I enjoy much of his work, but this is at a very different level, I think. And somehow it is because he has such a desperate need to deal with what’s going on. So there’s just a level of — in his case, there’s a lot of felt intensity emotionally. It’s a very driven piece of work. There’s more intensity being brought to bear. But he’s also got the intellectual discipline to not just let it become an outpouring. There’s this interesting way in which he’s very self-aware about what’s going on and what he’s doing. Some days he will come back to this essentially diary and just self-criticize: my goodness, I can’t believe, rereading what I wrote yesterday — it’s so selfish and self-centered, it is as though the main impact of his wife’s death was on me, not on her. He’ll have these self-aware remarks like that. And so somehow it’s an interesting balance of this very strong drive with also discipline.

Tim: Yeah. And I think one way I would frame it a little bit is that what’s aesthetically rich about it, maybe, is its scarcity as a kind of creation. I’m very interested in the kind of creation which you do for yourself, or do for a very, very, very small audience of people — the letter that you write to a friend, or the kinds of exchanges that you have in a group chat, a group chat of ten people or less. And that’s a rare kind of interaction, I guess, in the environment that we live in nowadays. And friend of the show Omar Rizwan — we have this debate where I push him on it, and I’m like, Omar, you believe that the invention of the internet was the death of computing? The minute you could start to share things with other people, that’s when things got very uninteresting on the computer. And of course, Omar — he’s a fanatic. But I think there’s something really interesting there, which is the computer as an object which is almost just a notebook for yourself. The kinds of things you tend to create when that is the case look pretty different from a world where you also have this distribution mechanism built into it. And so, as someone who spends a lot of their life on the computer — these kinds of things, right, the book by C. S. Lewis, the works by Henry Darger — a lot of outsider art has the feature of “share to yourself” or “share to one or two others.” And the intimacy of that I find very rich.

Michael: Yeah. I’m just remembering another friend of the show, Rob Ochshorn — he talks about the intimacy of an interface as an important quality. And I think he means several different things by this. One is just how intimate it will allow you to be with whatever it is that you are creating. But then there’s also a felt sense of the tools — certain tools just feel very intimate, others not at all. The thing that jumps to my mind now — I haven’t heard him use this example — is to what extent a voice interface will hear tone or not. The more it would hear tone, the more it would respond to your emotion, the more intimate it would feel. Whereas if all it’s doing is just transcribing the words, there’s a massive flattening of affect. And so there’s almost a dial there on how intimate those interfaces are going to be. It’s actually not transparent at the moment — you actually don’t know how aware a given voice interface was. Go ahead.

Tim: Yeah, sort of two interesting things there — a recommendation, and then maybe something that connects to something we were talking about a little bit earlier. One of the most fun things you can do is that you can buy old slide projectors on eBay, and you can buy lots of old slides of people’s vacations and things. And I was collecting them for a while, because I was thinking about a subscription service where we would just send you vernacular vacation slides from people’s vacations. And it’s a real experience to go through, because you have this very deep intimacy, right? These are slides that were taken maybe only for family sharing that just ended up in the market of eBay. It’s a super rich experience, so I highly recommend it. You can also do it with VCRs and home movies, which are also sold by the lot on eBay — very much that similar quality.

But I think there’s something that you’re pointing out — to go back to something we were talking about earlier — which is the supposed intimacy or non-intimacy of the AI system. On one hand, I think the point is like, well, it’s your confessor, it’s your friend, it’s this peer that sits with you. But as yet, we’re still missing these elements, right? Things like tone and intonation — all of these aspects that might confer a deeper sense of intimacy. And there’s kind of just something there, which is: we’ve presumed in some ways that these chatbots are immediately our closest peers in a certain sense. I think they’re still missing a lot that would let them ascend to that level of emotional resonance — and then we can actually ask the question of whether or not it’s something that we would want. Just something I hadn’t thought about before.

Michael: Yeah. I mean, they’re certainly making progress in that particular way. I was amused the other day — I made some offhand sardonic remark in some request I was making, and just then to read the chain of thought: “Michael is clearly bantering here and would appreciate a bantering reply in return.” It’s like, okay, great. Thank you, Doctor Freud.

Tim: The model behind the interface just sweating, being like, okay, where’d I put that banter file?

Michael: Exactly. “Sir is joking, I should dutifully laugh” is kind of the sense.

Simone Weil, difficult books, and the magic of effort

Michael: Okay, so Henry Darger is outsider art, and there’s this kind of intimacy. Maybe an interesting intermediate case is somebody like Simone Weil, who — my understanding is she wrote quite a few of her essays for herself. She would then sometimes — a friend would ask what she was working on, and she would explain, and they would ask to read it, and sometimes they then convinced her that it should be published. So maybe a very similar thing. Somehow, for whatever reason, Weil has ended up very influential, in a way. My sense is that Darger is certainly studied — he’s regarded as very interesting, but he’s not necessarily emulated in quite the same way, whereas there seems to be a very direct influence of Weil. I don’t know, maybe there’s just contingency there. Weil went to the Sorbonne and she had a very classic education. Her brother was a great, great, great mathematician, and she moved in intellectual circles. So maybe she was just a little bit more of the culture, and as a result had metabolized the culture a little bit. Littlewood to Darger’s Ramanujan, or something. It’s not really quite the right analogy, but he was a genuine outsider. She was a funny hybrid of insider and outsider.

Tim: Yeah, I think that’s right. This will sound flip, and I don’t mean it to, because I think Simone Weil’s work is very interesting. But in marketing, people used to say, well, you need to make your media snackable — that is to say, short and easily consumable. And the nice thing about a lot of Simone Weil’s stuff is that she writes an essay, it’s very straightforward —

Michael: Yeah. And it’s quotable.

Tim: That’s right. Versus this completely alien, bloody civil war battle that involves small children, and it’s going to be a thirty-thousand-page book. In some ways the scale of the creative production almost makes it non-transmissible. Which is something that I really like about it — I think it’s one of the features — but it really limits its audience in a certain sense. I was talking to a friend recently, and I was saying, the hardest trick in the world is not necessarily creating a piece of media that gets a lot of views. What you really want to create is a kind of media that has precisely one fan and no one else. That’s the goal. Can you design a piece of media that will just be appealing to one person and just immediately alienate everybody else? And I think that’s the kind of art that you see in some kinds of outsider art, particularly of the kind that Darger’s working on. Because it’s not even like — famously, some of the kind of outsider art which is a physical structure, okay, well then it’s a strange attraction that you might want to drive and go see. This is just this book, if you can even call it that.

For what it’s worth, I’m very interested in Talmud now, which has very similar attributes — you have this completely impenetrable document that is very hard to parse through, took a very long time to create. That kind of work I think is just very interesting.

Michael: It’s interesting how much that does separate people by personality. Certain people like Joyce for this reason — that it’s hard to read. You have a puzzle-solving mind. Of course, most people are immediately turned off — for the obvious reason, which is there are many wonderful works in the world which one can reasonably read.

Tim: People don’t have time for this. Yeah. I’m very intrigued — I was thinking for a while it’d be really fun to do a book club which is just books that spawn companion books that are necessary to navigate that book. So, for example, Gravity’s Rainbow has a book that comes with it. Finnegans Wake has many guides that come with it. So you can only read a book that’s annoying enough to read that it has a companion book associated with it.

Michael: Ha. What an interesting and strange — I wonder how the authors feel about that. I’m sure in many cases, certainly in Joyce’s case, he was very aware of what he was doing. It must be disappointing sometimes for some authors when they realize that they’re not being understood at all. I wonder about that with Gravity’s Rainbow — how legible he thought he was being.

Tim: Yeah, that’s right. With a friend I see very infrequently in DC, we’ve been slowly reading through Hegel’s Phenomenology of Spirit, section by section. Again, a book that is famously kind of just, what is even going on here? And I think that is the greatest act — to go back to Weil — to produce something which no one really is sure what it means, but they’re all convinced that it is important, and it finds its own social influence. It feels like the most complex social obstacle course you can navigate as a creator of media.

Michael: It’s interesting the extent to which you are responding to the aesthetics of the creative process, in a lot of these cases, rather than responding to the aesthetics of what is created. Those are quite different things. Sometimes they’re related, but sometimes they’re not.

Tim: Yeah, but I think it’s also the effort that it takes. Maybe it’s a kind of creative athleticism — the idea that you need to work really hard to produce this very specific outcome that not everybody appreciates. I was very interested — I still am — in the practice of sleight of hand. And there’s a notion in there, which is: well, sometimes the magic is just your willingness to do something. To just work on something ten thousand times until you get it. And the magic trick is that you can just do it, right? Famously, one of the things is you learn to shuffle so that every card goes one on top of the other perfectly in a shuffle, very naturally. You learn to hold a coin in the palm of your hand — you just need the muscles to do that. It’s almost a kind of circus art. But the magic is the effort. And I find that very compelling as a way of thinking about how art happens.

Michael: Yeah. One of my very favorite things on Penn & Teller’s show, Fool Us: somebody drops a deck of cards and snatches a particular card out. And [Penn & Teller] say, eventually: we think the trick is that there’s no trick — you’ve just spent twenty years learning to grab a card. And [the magician is] like, well, yeah, actually that’s exactly what I did. It’s just kind of a miracle, basically, which he has learnt to perform. Actually, interestingly enough, when he did it for Penn & Teller, he was quite nervous, and he grabbed three or four cards out of the air, and it was one of those. And apparently when he’s less nervous, more relaxed, he will sometimes just grab one card. It’s an interesting limitation on the miraculous ability. Anyway — that willingness to put in time to create a miracle is an interesting thing.

Tim: Yeah, that’s right. And we want a society where more people do that, right?

Michael: Maybe. It’s certainly entertaining, I think. And I wonder a bit about the cost for the individuals involved. Anyway, Tim — I think that’s probably a good place to wrap things up. Thank you very much for doing this. I enjoyed it thoroughly, and I appreciate the time.

Tim: Yeah. Thanks, Michael. Thanks for having me on.

Joe Carlsmith is a philosopher and writer at Anthropic whose concept of “deep atheism” has changed how I think about religion. In this conversation we explore what it means to trust or distrust the universe, why death brings life into focus, and why it can be hard to be sincere. We discuss Teilhard de Chardin, the problem of evil, Elinor Ostrom, Nick Land, and end on grace as a model for how a whole system can work together.

Links: Joe on deep atheism, part of a series on otherness and control in the age of AGI, and on death, sincerity, and “fake thinking” and “real thinking”.

Transcript

The following transcript has been cleaned up and filler removed. It was generated by AI, and checked by a human, but errors may remain. Check the video before quoting.

Introduction and defining deep atheism

Michael Nielsen: Hi, welcome to the Good Conversation Project. I’m Michael Nielsen. I’m a research fellow at the Astera Institute. I’m joined today by Joe Carlsmith. Joe is a writer and philosopher at Anthropic. His interest is in helping humanity orient wisely toward the long-term future. Joe, we’ve talked a little bit before about this idea of deep atheism, which I think is a really interesting concept.

You wrote this long essay as part of an even longer book-length essay. And the idea as I understand it is you’re a deep atheist if your fundamental belief or fundamental orientation toward the universe is that it is not benevolent, that it is somehow malevolent. So you give this great example of Timothy Treadwell, the grizzly man who lived with grizzly bears for ten-plus years, in a very trusting kind of a way, maybe too trusting, as a canonical example of somebody who’s not a deep atheist. And then you have Eliezer Yudkowsky, who’s one of the prophets of AI doom, as an example of somebody who’s much more suspicious towards the universe.

Something that I really struggled with while reading the essay was: what type of a concept is it? I’m trying to think of it as a scientific concept. Is this something that we can deduce from the laws of nature? And it maybe doesn’t seem like you can, because most of the universe is vacuum, which is incredibly hostile. And so in some sense, we should all be deep atheists. But how do you think about that? What’s the right kind of conceptual frame for it?

Joe Carlsmith: I think we can try to get a somewhat quantitative-ish definition of at least one aspect of it, though I haven’t tried to work this out, so this is going to be rough. I will say, I think the most important concept—you’d said not benevolent. And I think not benevolent is importantly different from malevolent.

The important thing about deep atheism for me is more about the notion of trust. Basically, the deep atheist is someone who doesn’t trust the universe to take care of them, to go towards good outcomes by their lights by default.

And correspondingly—and this is why I got interested in it—feels the need to exert a kind of control over the direction that the universe goes. And that’s different from thinking that the universe is actively trying to mess with you or actively hostile in some more intentional sense. And maybe that’s not what you mean, but I just thought I’d clarify that as an important backdrop.

Michael: That’s very helpful actually. The way you’re describing it, it sounds like it’s actually quite a psychological state, maybe primarily. It’s not the same as paranoia, but it’s a little bit related in some sense.

Joe: Not necessarily. I think there is a component. If we were going to try to define it somewhat rigorously, what I would do is say, okay, let’s say you’ve got some values. And we’re going to use those values to rank possible states that the universe ends up in. And we have some notion of the default that happens if you don’t try to control stuff. I think my baseline if I was going to try to define deep atheism would be: deep atheism is the view that the world will be bad by your lights by default, if some sort of gentle control is not exerted over it.

Michael: Yeah.

Joe: And then I think that has a component of the empirical direction of the universe by default, whatever that means. And then there’s a different thing, which is about your values. An interesting counterexample, or a case that’s sort of an edge case of deep atheism: suppose your value system is such that the main thing you care about is the absence of suffering. In a sense this worldview is often associated with a kind of deep pessimism and a kind of aversion towards the world, because in some sense there’s all this suffering, it would be better if there was nothing. But actually there’s a way in which this view is not very deep atheist, in the sense that most states—if you imagine the world sort of rolling forward—most states don’t have much suffering in them.

And so by default, unless someone is really steering towards suffering, the universe kind of doesn’t create suffering. Most stuff is just blank, but blank on that ethics is great. It’s the top you can get. So there’s a sense in which this worldview is very optimistic. If you just don’t do anything, the universe probably won’t have that much suffering, and that’s the best you can get. So that’s an example of a case where it really matters what the value system is. You can have the same empirical picture of where the universe might go by default, but the value system changes your read as to whether that’s good or bad.

Michael: I just want to check that I understood, because I’m not actually a hundred percent sure I did. You’re saying that whether or not one is a deep atheist does depend fundamentally on your other set of values.

Joe: Yep, that’s right.

Michael: Okay, that makes a lot of sense. And it’s still a response then. Actually, for most sets of values, somebody who’s living near grizzly bears is going to think, I should probably be somewhere else. Their response is to control the situation, to say, I need to get out of here. Living in a tent 300 metres away from grizzly bears is probably not the best idea. And it probably wasn’t Timothy Treadwell’s values that made the difference there. He probably wasn’t thinking it doesn’t matter whether or not he lives or dies. He actually seems to be a person who quite enjoys life. Maybe struggles with conventional life, but he does actually really enjoy his life. And he doesn’t want to suffer either. So he has quite conventional values in those ways.

But he still had very limited need for control in some interesting way. And that’s what made him—I don’t know what the right term is, the opposite term to deep atheist—but he was certainly not deep atheist.

Joe: Yeah. The term I use in the series that’s the closest is this notion of “green,” where green is associated—it’s a term from the Magic the Gathering ontology, somewhat remixed and reconceptualized—but where green is understood as an attitude that has some kind of fundamental trust in the notion of nature. And nature is a kind of funky term here, but it’s importantly the other—the thing that’s prior to you, the universe, the thing beyond yourself.

Michael: Do you mind if we stick with just me saying “the opposite of deep atheist”? Because in my reading of your essay, it didn’t seem to me that you wanted the term green to be exactly the opposite of deep atheism. It was doing a lot of the same work as that opposite, but was doing importantly different work as well.

Joe: Yeah, that’s true. I should say, one of the things I really appreciate about your engagement with this concept is it’s pushing me to try to figure out what are the actual boundaries of this, how does it actually fit together. And so I’m very open to what we learn being, actually, there’s a bunch of different things here and it goes different directions. I’m not pretending to have worked all this out.

I think the actual person—yes. Opposite of deep atheist: there’s a sense in which Timothy Treadwell is not a deep atheist in that he’s very trusting in this context towards a certain kind of natural process, namely the grizzly bears. And the way in which I would see him as differing from someone like Eliezer in that context is not so much the values he’s bringing to the table, but rather the empirical bet. I think he’s somewhere saying that he doesn’t expect to get eaten. So there’s a way in which it’s not—if he had said something like “I’m down to get eaten because I want to feed nature,” then it’s more of a values difference that’s driving the trust.

Michael: He was on Letterman, and he was interviewed by a bunch of people. And many of them asked variations on more or less this question. He’s maybe not a hundred percent consistent in his answers, but he certainly doesn’t come across as somebody who—he’s not committing suicide here. He’s not even tacitly oriented that way. He actually seems to be a person who quite enjoys life. Maybe struggles with conventional life, but he does actually really enjoy his life. And he doesn’t want to suffer either. So he has quite conventional values in those ways.

Joe: That’s right. And so the deep atheist has similarly conventional values. And the mistake they’re trying to avoid is trusting in the universe to care for those values. And Treadwell is at least some example of someone who is trusting in the universe to care for—in this case, the value of his life. And it didn’t.

Deep atheism beyond the local: Teilhard de Chardin and cosmic teleology

Michael: It seems like in the most immediate case, it is a psychological attitude towards one’s ambient surroundings. You take a plane halfway across the world, you’re in some city, and maybe you actually feel quite unsafe, or maybe you feel surprisingly safe. You walk into Kyoto at 11 PM and you’re like, wow, this place feels amazing. Different people have different levels of trust under those circumstances, and some people will be very distrustful habitually. It’s kind of situational, but you’re talking about something quite broad—broad expectations, a broad model relative to your values.

Joe: Yeah, I think you’re pointing at something that does seem important, and it’s a way in which I actually wouldn’t think of Treadwell as a paradigm person to think about in the context of deep atheism. I agree that he’s not a deep atheist, but I do often think of deep atheism and its contraries as more oriented towards the trajectory of the development of the universe and of life on a broader scale. Or at least that’s where I started thinking about the concept, because I was thinking about it in the context of AI risk and the aspiration a lot of people have to somehow have control over the future. The thought is you have to exert that sort of control, otherwise you’re going to get a loss of all value.

I think it’s separable in the following sense: you could be someone who has both the values and the empirical views that I’m associating with Eliezer here, and nevertheless land in a place that’s reasonably safe and be calibrated about how safe it is. To the extent you’re mostly talking about the local environment, that’s more a matter of local epistemology than it is about your convictions about the structure and dynamics that the universe tends towards at a fundamental level.

Michael: That’s actually quite helpful. So it’s—in some sense, much of your local environment, obviously it’s incredibly important for you wherever you are, but it’s also a little bit shallow. It’s not expressing all of the range of possibilities of the universe. And you’re more interested in that broader range of possibilities—what does the universe have in store for us potentially? You’re more pushing in that direction if I understand correctly.

Joe: Yeah, that’s right. And so I actually think that the paradigm non-deep atheist that I think about is—well, there’s sort of two. I think maybe the most salient is this Jesuit priest named Pierre Teilhard de Chardin.

Michael: Really? Wow.

Joe: He’s, to my mind, the least deep atheist—or like Bergson or these people who are not just Christian in the sense that they believe that there’s a God somewhere. It’s really that there’s a God as manifest in the inevitable evolution of complexity—

Michael: Yeah. The Omega Point and all of this. It’s a very strong singularity kind of a notion.

Joe: Yes, and a very strong inevitability. There’s this crazy passage in Teilhard’s book, The Phenomenon of Man, which I think illustrates—I don’t like his view. But he has this passage where he looks out at evolution and he sees this progress towards complexity and greater mental connection. He has this whole teleology that he’s extrapolating from the paleontology and the record. And then he says, okay, some people say maybe we could go extinct, maybe humanity would get wiped out. And he’s like, no way. Why not? Because the universe clearly wants to get to the Omega Point. And if humans got wiped out, then it wouldn’t get there. And so it’s going to get there. This is a very intense kind of retroactive application of his teleology to basic causal dynamics. I think he’s actually talking about asteroids—could we get hit by an asteroid? And he’s like, nah, because then the progress of evolution would be halted.

So in his case there’s this very strong inevitability to the default dynamics that we are already caught up in. It’s not even that you could mess it up. And you will get to the Omega Point, and Jesus will look back on history and smile—or sorry, there’s a bit at the end of the book where it looks like he thinks Jesus is present at the Omega Point.

Michael: I wonder—he was writing that in the 1930s or something like that. This is before the atomic bomb, and before MAD [/ed/ - I got this wrong – it was published in 1955]. I’m thinking about Petrov and Arkhipov—single individuals who could or could not essentially push a button. It wasn’t actually a button, but—either send the human species back to 5,000 people living in caves, or not. And fortunately for us, his argument would almost be something like: it was inevitable that Petrov would decide it’s probably just a bug in our early warning system, we shouldn’t launch a counterattack and end civilization.

That’s almost the point of view he would have to take with that teleology.

Joe: Yeah. I think probably the more charitable version of his view is one that you do see. That particular intensity of inevitability of the good outcome is, I think, rare. A view that I would see as more nearby is one that treats something like the process of evolution, capitalism, natural selection—some analog—and sees that process, if uninterrupted, as inevitably leading to a good outcome. Teilhard goes further and says it won’t be interrupted, but we can let that go and say, well, maybe you get hit by an asteroid, maybe you have a global pandemic, but if you don’t, and you just let some kind of emergent, decentralized process of selection and competition and cooperation do its thing, then very robustly you will end up in an approximately optimal outcome, or at least a good outcome.

Selection, progress, and Steven Pinker

Michael: That’s really interesting. You can sort of view it at some level as just a point about what selection filters are being applied. If your selection filters are actually pretty positive, then it’s actually not such a terrible—it’s kind of believable that you can make up a story where this makes a lot of sense. Certainly, I don’t think somebody like Darwin would have believed this, but maybe you go long enough and the selection filters actually become positive enough. It’s almost a self-reinforcing thing where if the environment is already ambiently good enough, the selection filters will actually ensure that things get better. And I suppose you could imagine making that kind of argument.

Joe: Yes, I think this way of rejecting deep atheism is in fact empirically on the table, depending on different sorts of values and how high your standards are for satisfaction with the outcome. I gave a talk which you and I discussed a while back about whether goodness can survive intense processes of competition.

One vision of optimism there is: you have a bunch of competition, but inevitably in order to be competitive in many processes—and I agree, it matters what the selection filter is—being competitive requires something like being friendly with other agents and making deals and abiding by norms that involve respecting boundaries and avoiding zero-sum conflict. If something like the thing that falls out of reasonable game theory or cooperative norms applies more and more as you go into the process of greater evolution and competition—or sometimes people talk about our global mind getting more synthesized together, more connections, maybe the thing to do if you’re trying to be competitive is to build a giant coalition with more and more people being happy and more and more people contributing their part of the cognition to some collective intelligence. I think that’s quite close to what Teilhard is talking about. And I think variants of that, you still see this crop up tons of places.

Michael: Can I just make a connection? I never would have thought of Teilhard as being similar to somebody like Steven Pinker or Max Roser, but actually you’re pointing out that there is something very important that they have in common.

Joe: Yes, though I feel like Steven Pinker does less in terms of—he doesn’t extrapolate out the graphs enough. He sort of says, check out my graph, but he doesn’t do a bunch of “and what happens in the limit of competition” or “what is the teleology of evolution.” But yep.

Michael: Yeah, it’s much more local and much more empirically focused. He’s just interested in this question of, well, have things actually been getting better? And that’s interesting—they are, at least on some axes. Why? And then he has a theory which is essentially about selection filters. He doesn’t extrapolate it. He’s not trying to claim that in a hundred thousand years we’ll all have ascended to the sky or whatever.

Joe: But I do think you’re right that there is a similar shape to the trajectory of history as being presented, which is up and to the right. As you go forward, there’s some sense of trust in the default. And then there’s also a question of how concerned are you about that project being knocked off course? I helped at one point with this book, The Precipice by Toby Ord, which is actually in many respects in a Pinkerian historical vision, but is very concerned about this: well, there’s a default trajectory of progress that just needs to be not interrupted. But it’s not so much about steering it one direction versus another. It’s more like, don’t die. If you don’t die, you’ll get to the good place.

Michael: Yeah. It’s sort of asking, how can you die?

Joe: And then interestingly, you start to sneak in more and more ways of dying. For example, what if you get the bad values? And is that death, or—the AIs take over, but the AIs in some sense are quite continuous with the process of capitalism and whatever else. So there’s a weird way in which you can start sneaking in more and more deaths, until it’s a little bit unclear whether there’s really a default here, or whether you just want to steer to a very particular outcome in the end.

Deep atheism, religion, and the problem of evil

Michael: That’s very interesting. It’s an interesting way of thinking about Toby’s book. Just in general, I think part of why the deep atheism stuff has remained so much with me is that it helped me understand better than anything ever before—and I’ve mentioned this to you—that the question of whether or not you believe in God isn’t actually that important for religion. It’s very important in the Abrahamic religions. Moses and Jesus and Muhammad are all very clear on this—”I am the Lord your God and you shall have no other gods before me.” Jesus’s great commandment is to love the Lord God with all your heart, all your mind, and all your soul.

Very important in the Abrahamic religions, but not so important in some other religions. There’s this fact I love about the Buddha, that there are these 14 unanswerable questions in Buddhism, questions like: How did the cosmos start? How will it end? What is its fundamental nature? And the Buddha is like, just quit wasting your time. These questions are a complete waste of time. Focus on suffering. This is what’s important, ameliorating suffering.

He’s not interested in a lot of these big cosmic questions that the Abrahamic religions are very interested in. Buddhism is certainly much less interested in the question of whether or not there’s a God or what the nature of God is. And at first when I read about deep atheism, I was like, it’s so badly named, what’s it got to do with whether or not God exists? And I think I’ve come around a little bit to the term. It’s still hard—my first 20 minutes of conversation about it with anybody is always them complaining about the term. So better branding might be helpful. But I do understand: it is this fundamental question about your orientation towards the cosmos, what do you think the cosmos is about, what do you think its nature is? And you can’t be agnostic on this. Every human being has some orientation of this type. And it is a kind of religious orientation.

I’ve spent a little time thinking about essentially ontologies—what else has this character, big broad questions about the nature of the cosmos? Just one example, which I think is very common among scientists. I find it particularly moving in reading Einstein, more than anybody else. You read Einstein’s letters and the extent to which he believes that the universe is comprehensible and simple—it is just astounding.

He has deeply internalized this point of view, and I get teary reading it, not because I necessarily agree, but just because of the intensity of the conviction. And you realize, oh, this is part of what made him a great, great, great scientist—this intensity of belief. And again, there’s nowhere in the Bible where it says this, there’s nowhere in the Torah or any other scriptures—I mean, you can make arguments, but they’re—

Joe: Well, Job—famously, “who were you to be—were you there when I made the hurricane?” I forget the exact words—

Michael: I don’t think so. You’ve got to admit that it’s something of a reach. People do try to connect Judeo-Christianity to modern science, and obviously Newton was deeply religious and whatever, but it’s a different kind of a thing. You don’t get general relativity out of this point of view, I don’t think. It’s somehow much deeper, the notion of this incredibly deep simplicity. I’m just curious if there are other broad questions which seem to you to have this same kind of flavor as deep atheism, or for that matter, this belief in the simplicity and comprehensibility of the universe.

Joe: Yeah, I like this. “Is the universe comprehensible?” is a very interesting fundamental stance. There’s some question—I think it’s an interesting point—how easily do you give up if there’s something you don’t understand? And that can come from “I’m personally not going to understand it,” or it can come from some backdrop sense that the world is full of fundamental incomprehensibleness.

Michael: Do you know the story of Einstein when he was four? He saw a compass for the first time and it oriented towards north and he wanted to know what it was doing. And then he reports it as an epiphany—to realize that there was this fundamental order in the world that could be detected in this way. And I think it’s interesting that even at age four, he had that—my God, there are hidden forces here and we can understand them. I don’t think most four-year-olds—compasses are cool, but they’re maybe not that cool.

Joe: Right. Yeah, four is early for that degree of ecstasy.

Michael: Yeah.

Joe: My point with Job was more actually in the opposite direction. I think the Christian tradition has a—if you’re going to read a view about the comprehensibility versus incomprehensibility of the world onto the Judeo-Christian tradition, I would have said it leans incomprehensible, at least in the context of the problem of evil. I think the problem of evil is the big problem for religion. And in that context, you’re stuck either saying something like “it’s incomprehensible, sorry, you just don’t get to know,” or you end up talking about the demons doing it, and you end up doing this quite unsatisfying game of trying to explain it.

Michael: I think it’s a problem for monotheistic religions. Much less of a problem for everybody else. Would you agree with that?

Joe: Yes and then no. And this gets to some of the stuff about fundamental stances, where I think the problem of evil threatens religious orientations at a more fundamental level than at the level of belief in a perfect, all-powerful God. Because there’s a different thing you can have, which is a fundamentally positive stance towards being itself.

I would associate that with many religions, many kinds of spirituality-adjacent vibes, including some associated with quite pessimistic religions like Buddhism. There’s a way in which—Alan Ginsberg has this poem where he’s like, holy is everything, holy the saxophone, holy the city, holy the cock, holy all these things. He wants to say everything is holy. And I think there’s a lot of spiritual traditions that aren’t explicitly monotheistic in the Christian sense, but nevertheless want to attribute a kind of sacredness or value just to being as such, or reality as such. And I think the horribleness of much of reality is a barrier to that kind of stance, even independent of the metaphysics at stake.

Michael: I’m just thinking about Shinto, where absolutely everything seems to have some kind of deity associated to it. And they don’t seem to have any problem with the fact that some of these are pretty bad. The bear out the back that’s going to rip your head off—okay, it’s got a deity associated to it, but maybe not a particularly friendly one.

Joe: Yeah, I agree. And the question is, do you end up at some other level viewing this bad deity as sacred or not?

There’s actually a really interesting scene in the movie Princess Mononoke. Have you seen it?

Michael: I think I have, although the Ghibli movies do shade into one another a bit for me.

Joe: It’s the one where it opens with this boar that has been infected with some horrible demon—it used to be a boar god and it’s rampaging with all these horrible bloody worms coming out of its skin. It’s clearly a demon, labeled as a demon. Bad. The hero kills it with an arrow.

It eventually dies in front of the village, and the village wise woman comes out and kneels before the demon and says something like, “great Lord of destruction and death”—some kind of honoring of this demon. And the demon interestingly replies, “you know, filthy humans, you will suffer as I have suffered.” It has no—it doesn’t bow back. It’s just raw hatred. And this is expected. I think it’s a really interesting scene of confronting a thing that is being labeled as genuinely bad in the world, and nevertheless holding a sense of honoring it, some deeper sacredness, some part-of-the-totality sense.

Michael: And you would view that as an attempt at a resolution of the problem of evil? I think that’s more or less what you’re saying?

Joe: It’s the version that seems best to me. I think the balance I feel is necessary is: you need to be able to say that bad things are bad. Really say it. If you see some horrific suffering, torture—I think there’s something that feels really wrong about honoring that and saying, this is holy. There’s something important about being able to say, no, this is bad and we will end it, and it is not sacred.

And then there’s some other level at which one wants to be able to affirm—or maybe not affirm, but love, or forgive. There’s a sense of grace. This is the synthesis of deep atheism with spirituality that I attempt in the series. You want to be able to both mistrust the world at an empirical level, and nevertheless have some kind of love for it in the same way you love a flawed thing.

Michael: I think you’ve written about this at the level of individual aging and death as well, where you’re quite sympathetic to the idea of celebrating that process and accepting that process, but also rejecting it—you’re doing both. Particularly at the end of a life. And somewhere in one of your notes you talk about, with I think quite justified disdain, the fact that at funerals sometimes people will tell half-truths, will just not be honest and frank. And that’s a moment at which, as far as I can tell, you want a lot of frankness.

There’s this interesting tension—at least the tension I feel when you’re talking about individual deaths—between wanting to fight it and regard it as something where we do want to exert control, we do want to defer things. And then at some point, there is a moment at which one accepts it and starts to integrate it and starts to celebrate it even. And you want to be able to have very frank, almost joyful conversations about it. I’m putting words into your mouth—that’s maybe more my interpretation. But that seems like the single individual level analog of the deep atheism point. Is that fair?

Joe: Death is definitely one of these areas where many deep atheists think that we go wrong in our degree of trust and affirmation and acceptance towards the world. Yudkowsky and others come out of this—actually prior to AI, a lot of that tradition is rooted in this really strong “death is bad, death is the enemy, let us end death.” And they see with horror the way in which people apply a kind of trust, acceptance, yin, sacredness vibe to death. They see this as almost a kind of learned—well, there’s different diagnoses, but they think this is really misleading, to be like, oh, death is sacred, death lends meaning to life. That’s a kind of cope that causes you to exert less control, when in fact—and maybe that makes sense when you can’t control this, but if you actually can, then you need to be able to see that it’s bad and that you should change it.

Dylan Thomas, Wendell Berry, and the synthesis

Michael: Two of my favorite poems—one is Dylan Thomas, “Do Not Go Gentle Into That Good Night,” the famous “Rage, rage against the dying of the light.” This is the deep atheist point of view you’re talking about—fight back.

Joe: Yeah, totally.

Michael: And then there’s Wendell Berry, who’s very much a green in your description. He has this wonderful poem about the peace of wild things, who do not tax themselves with forethought of grief. And I think in some sense, part of why I respond to a lot of your writing and really enjoy it is I feel like you’re attempting a synthesis of these two points of view, where you regard them as a consequence of a single deeper point of view, and you’re searching for what that deeper point of view is. Obviously I’m massively putting words into your mouth here, but instinctively that’s part of what I feel in it.

Joe: I appreciate that. I am trying to synthesize this stuff.

I think there’s two aspects. There’s an empirical question about the nature of the universe, and also about our values at the level of what philosophers would call axiology—you’re looking at a state of the universe and ranking them. A particular way of evaluating an outcome.

And then—and you’ve written about this and we’ve spoken about it a little—there’s a different thing, which is: what is the subtlety of the shape of your spiritual orientation towards this? Both this empirical picture and this picture of your values, but there’s some other component: what is the integrative pattern of resistance and acceptance and love and rejection? There’s a bunch of ways your mind responds to particular things in the world. I am trying to somehow hold space both for a genuine hard no—this is bad, to be fought against, to be killed, this should be ended, it’s not sacred—and then underneath that, there needs to be love, or there needs to be some grace, forgiveness. At least that’s the synthesis I’m trying for.

I sometimes think about, does Jesus ever kill something? How does Jesus kill something evil? And I sort of think he kills it with love somehow. He sees its true nature, there’s a way in which it is holy for him. And also he’s doing something that needs to be done.

Michael: Can I give an example, which is a very different direction but same person? Jesus—the story of the gospels does do this incredible thing of making death into something to be celebrated. You have a story of a terrible death which ends in triumph. And that’s a very interesting rhetorical move. It’s quite an incredible thing to have made that into something which is an occasion for incredible joy. My understanding is that’s the way many Christians relate to it—the worst defeat followed by an even greater triumph.

Joe: Though it is tough, because it’s a very particular empirical story. And that is the concern about the way in which a lot of Christians get around some of this deep atheism. I’m kind of like, well, it’s different if there’s heaven at the end and all this stuff.

Michael: A get-out-of-jail-free card, kind of. So you’re saying this is of course a very conventional Christian response to the problem of evil—to punt and say, it’s okay, here’s the big reward at the end, no matter what.

Joe: I think there’s at least—I don’t think that is enough as a response to the problem of evil. Even with heaven, people are like, well, couldn’t God have created a world where you just didn’t have forest fires burning these deer alive? Why you gotta do that, God? But I think there is a way in which if you have heaven in the background, it is easier to see the fundamental stance towards reality as one of trust and affirmation. Because ultimately, all shall be well and all manner of things shall be well.

And a nice thing about Christianity is that heaven is bifurcated from earth—at least in the more traditional Christian view. People like Teilhard are doing something different, but the traditional view is: earth has fallen. The natural world is beautiful but broken. That’s how you accommodate all the ways in which this is bad, and then heaven is a different story. So there’s a way in which you can actually have a horror towards the actual empirics of the world as a theist. There’s a theist blogger named Bentham’s Bulldog who has a post affirming a deep atheist stance—horror towards the actual empirics of the world—that’s possible as a theist, if you’ve put your stock in heaven.

Michael: That’s a good story. It doesn’t exactly solve the problem, but it certainly makes things a lot easier.

Joe: I would feel radically different as a—there’s also a way in which if you’re a theist, you think that the ultimate explanation for things grounds out in a perfect being. Everything comes from God. So even in the actual world with all its pain, there’s some way in which it is easier to say things like, this must be part of the plan. There’s a comprehensibility analog to Einstein’s view, but with a moral component. It’s not just that the universe is comprehensible at an intellectual level. It’s morally comprehensible—ultimately this was worth it, this was good, even if it really doesn’t seem good. If I had that view, I would have a very different attitude towards the universe.

Michael: A book I absolutely love and find very moving is C.S. Lewis’s A Grief Observed, basically his private diary after his wife died.

Something that I find so moving, even though I myself am an atheist, is that his Christianity is so important to him. And yet he’s so angry at his God for allowing this terrible event. There’s a particular passage that you highlight, and I could barely continue to read at that point—where he’s considering seriously the notion that God is evil, that God is bad. This deeply religious man who sacrificed so much of his life. I suppose it’s both a beautiful articulation and it’s also so deeply felt. I think the question I want to ask is just for your response to it, and also if there are any other similarly sharp examples where it’s been articulated so beautifully and with so much depth of real feeling.

Joe: I think it’s his most raw book. It’s also plausibly his best. The Great Divorce is also really good, different. I recommend it. It’s his vision of heaven, this endless journey into God.

I think what strikes me about that passage is that it illustrates how a certain sort of religious affirmation towards the universe can be separated from one’s metaphysics of God. It could be that you still believe in God, but you’ve lost a kind of affirmation of God, basically. And this also happens in The Brothers Karamazov, which is another example—again, it’s the problem of evil that does it. Ivan, in virtue of the suffering of children, hands back the ticket. He hands it back even if it’s justified. He doesn’t want to hear the story for why it’s okay for these children to suffer. Maybe in some sense it is okay—I don’t care, God, I’m giving you back the ticket. And it’s a little elusive exactly what’s going on there, but at the least there seems to be some rejection of God that is nevertheless compatible with still believing in God.

There’s this interesting pattern of loss of deference or dependence on God. I write about this in the series—the play Angels in America has a similar thing, Philip Pullman as well. There are these patterns of stories where there is a God and nevertheless, your role is to not defer or trust in God, but rather to be an independent agent in your own right. And we see that happen even in theistic contexts. Lewis—he eventually goes back to Christianity, but there is this brief moment where—he never doubts God, he claims—but he does start to doubt whether God is worthy of his allegiance.

Michael: This is reminding me why deep atheism is actually quite a good term. It’s not really about belief in God or not, but in that frame, the response of people who want a tremendous amount of control in the world is a very natural deep atheist response. If you don’t believe that God is good, or if you don’t believe that the universe is benevolent, wanting more control is very natural.

It’s interesting to think about. It’s such a psychological thing. You meet different people and you very quickly realize how much control and power they do or do not want. I’d never really thought of it in theological terms before, but it is that type of thing.

AI risk, e/acc, and Nick Land

Joe: Yeah. I got interested in this because of AI risk and the aspiration to somehow have control over the future. The thought is you have to exert that control, otherwise you’ll get a loss of all value. The traditional story is that the AIs will steer the world in some other direction. But the AIs are really just a kind of amplified version of the indifference of nature. Nature doesn’t care. Intelligence doesn’t care either, because goals are orthogonal to intelligence. So everything—intelligence, nature—it’s all orthogonal to value. There’s a very specific vector you need to be steering along, and all the rest go in bad or valueless directions.

People talk about how AI risk is some theological thing. Sometimes that’s not true, but in this case, there actually is a real deep connection between your fundamental metaphysics of the universe and one’s relationship to the control at stake in AI alignment. You see that too in the context of—I started writing this when e/acc was just coming out. And e/acc has this version of a rejection of deep atheism that is rooted in the writings of Nick Land and this allegiance to capitalism.

There’s some notion that capitalism, competition, evolution will just inevitably go in some direction. In this sense, it’s kind of like Teilhard, but Nick Land is much more of a nihilist. Teilhard is like, it’ll go and it’ll be good—there’ll be love, joy, consciousness. Nick Land is like, no, it will kill everyone and maybe drive out consciousness.

Michael: And “what are you going to do? [apathetically]” seems to be the response.

Joe: Something like that. And it’s an important way in which that’s not deep atheism, in that there’s a canceling of the value component. Normally deep atheism is: something will go a bad direction by default. Nick Land, at least understood as nihilistic, is just saying something will go in some direction, and I’m not going to do the value thing, or I’m going to affirm that in some weird way.

Michael: Well, within the terms you explained earlier, you can understand it as—maybe he just has a different value system, and relative to that value system, he could be a deep atheist. That seems at least consistent.

Joe: I guess to the extent he has a value system that affirms the arbitrary destructive potential of unbridled competition, then I would say he’s not a deep atheist in the relevant sense. Similarly, there’s some parts of the e/acc writing where it sounds a little bit like they’re really into the dissipation of energy as waste heat. There’s some weird stuff about the death drive, some lineage that goes through Land—various things about burning the sun and getting rid of the energy. And it’s almost a pro-entropy stance. And the reason you’re into capitalism and life, or at least this is the suggestion, is that it’s the fastest, most efficient way of converting energy to waste heat.

There’s something about this where—again, it’s a radically different value system. If you’re really into entropy, then the universe is great, because entropy is this really strong force. You’ve got a really strong god operating on your side. It’s just that that’s a radically alien value system.

Michael: It’s a funny thing to want to be maximizing. Actually, I’m just thinking about the fact that if you do things in information theory, if you compress information as much as possible, it starts to look like it’s essentially black body radiation in one dimension.

So there are interesting connections between maximizing entropy and desires for efficiency. There’s a great paper from I think 1996 by Cris Moore—they’re interested in the question of looking at signatures of alien civilizations. Let’s assume that [aliens are] very concerned with maximizing the efficiency of how they communicate. They’re going to be limited by results from Shannon about information theory, and you can ask what it’s going to look like. An old joke, going back many decades even prior to that, is that it’s going to look like blackbody radiation, so we’re not going to be able to tell. But they point out it doesn’t look like blackbody radiation in three dimensions—it looks like it in one dimension, so you could actually get a signature.

I haven’t thought about this in 15 years and I’m probably butchering it, but there are funny connections between things looking a lot like heat and things actually being done extremely efficiently. The point is you can compress down so that all of the redundancy is eliminated, so of course it starts to look like noise.

Joe: In some sense, that picture is quite continuous with Land’s thing, in that he’s really into efficiency in some sense. But the problem is that efficiency qua efficiency doesn’t look on its face like it’s especially good. You need something that it’s efficient for.

I think there are some interesting questions here as to whether maybe the process of life and growth and intelligence is really quite nice and benign by default, even as it becomes more efficient or more powerful. Maybe power is actually really deeply unified with goodness enough—that’s the best-guess version in my head of an actual empirical worldview that could falsify deep atheism at the empirical level.

Competition, cooperation, and protecting the weak

Michael: This is a very Steven Pinker kind of argument—actually, the way to succeed in a world once you’ve achieved some minimal level of law and order is to be very trustworthy. It’s to trade, it’s to communicate well, all the good boundaries kind of stuff. And it is interesting that there are at least some simple models in which you get an emergent ethics which is surprisingly positive out of what seems like quite a brutish world.

Joe: Yes, that’s the sort of thing I’m talking about. The question is how does that work? Does that sufficiently protect the weak? I think that’s an important aspect. We do see predation, raw predation.

Michael: Have you ever read the Code of Hammurabi? It has the most amazing passage very early on where he explicitly says that the purpose of it is for the strong to protect the weak. My understanding is that it was probably codified very early, as he was getting started. To some extent it was justification, but it does this amazing rhetorical move of getting its moral force from that argument. I’m going to rule over you. And part of my moral justification for doing it is I am going to protect the weak from the strong. Whatever that is—3,800 years old or something.

That’s a very sophisticated and at some level very encouraging idea. It suggests that human beings are sufficiently good at coordination and talking to one another that the way for me to get power is to act on behalf of the powerless. That’s kind of the implicit theory, at a very early stage. You can go forward in time to Locke and whoever—maybe there are similar ideas much later. But I think it’s very interesting that it’s already so strong. You can tell a story about Christianity that’s a bit similar.

Joe: Certainly. Christianity is famously interested in this pattern of concern for the weak. The question is whether that pattern of concern is suitably connected with power such that, if you really selected for power in the world over time, you would persistently get beings that cared about the powerless. The argument for why you wouldn’t is: the powerless, by hypothesis, don’t help you. They’re not very useful trade partners. You could be like—and we also do see this. Concretely, we fund vaccines for diseases in developing countries way less because people don’t pay for it in the same way. And there are non-human animals and all these patterns of ignoring the interests of tons of beings that are stitched into even the relatively cooperative structures we see.

Michael: I don’t know the history of the Code of Hammurabi—it’s so remote in time that probably nobody does. But you can say, what’s actually going on? It’s not that the individuals are powerless. They have a little bit of power. And Hammurabi has realized that he can aggregate that and use it to support his claim by acting on their behalf. So there are two distinct ways of looking at it. The other would be that he really is concerned out of some kind of pre-Christian ethic, doing this out of the goodness of his heart and his sense of empathy. And maybe both of those things may actually be true.

Joe: Yeah, it’s probably a mix. The question is—you could think, it’s really hard for people to have no trade value or to be bringing nothing to a coalition. And so you would expect that it’s quite advantageous to be nice to tons of people in a way that allows for productive trade and other forms of cooperative relationship.

But there’s a bunch of gnarly empirics there. It’s sort of related to comparative advantage. And unfortunately, it seems contingent. It wouldn’t necessarily be a super deep feature of the universe, because the costs of cooperation and subsistence—you don’t hire everyone at your company. There are lots of people who could contribute, but it’s not actually worth it. We do see cases where domination might just work. I’m Pinker-curious, but not Pinker-confident about where this actually goes in the limit. And I’m concerned that there’s a shallowness of the atheism at stake in some of the Pinker/Roser liberal optimism discourse, where they’re kind of like, I’ve looked over a century of history and it sure looks like something like liberal democracy also converges with the most economically or militarily competitive arrangements. And I’m like, well, yeah, maybe until the ‘90s. But even if it’s true, it’s been true for a few centuries over this very small set of selection processes and different experiments. Trusting in that over the longer-term selection processes is a much tougher and higher standard.

Michael: I think I understand you. What you’re saying is we are unfortunately in an n-equals-one case here. You can maybe make a case that China developed somewhat independently of India developed somewhat independently of the West [etcetera]. But there was a tremendous amount of mixing. And it might be that if something slightly different had happened 10,000 years ago in the Indus Valley, the world would actually look completely different. Even in the last hundred years—Philip Roth, I think, wrote this alternate history in which the Nazis won World War II. Even that, on the scale of 10,000 years, is probably not the largest intervention that was possible, but it does lead to a very different world and a very different set of values. Same biology, different culture. And yet there does seem to be that level of contingency.

Joe: Yeah. There’s a few different ways that example could function. It could be that there’s a bunch of contingency where the Nazis won World War II and then, because there just wasn’t that much competition—so it’s a question of who would be—at least when I think about this in the context of whether goodness is compatible with competition, which I think of as quite related to deep atheism. Competition is one of these fundamental forces driving the evolution of agents interacting. And if you can’t trust in competition, then you need to do a bunch more control. You have to shut down competition in a bunch of ways if it ultimately leads to very bad places.

The concern would be: if things can just go in all sorts of directions, if you’re concerned about competition, then path dependence is your friend. You’re hoping you can be the path-dependent one where you create a nice world and it just doesn’t get selected against super hard because the selection pressures aren’t that bad. My concern would be something more like: if you looked across the universe, the prediction of someone who’s really optimistic about liberalism or norms of cooperation is that everyone’s being treated nicely, everyone has property rights, there aren’t unhappy workers, there are no slaves anywhere, because slavery is economically disadvantageous. And I’m just not sure. I think empirically, if you did a bunch of alien civilizations, it totally could be that slavery is part of the most efficient arrangement. And that’s bad. It’s not that that’s good.

Michael: You think about E.O. Wilson and his two great examples of eusocial animals on Earth—the primates and the insects. I don’t know how good life is for an individual ant. I don’t think it’s particularly good. And there are a lot more ants than there are people.

Joe: It certainly doesn’t look like a paragon of liberal property rights in an anthill.

Michael: No. And you have to go out to a separate level. And maybe at some other level it actually makes sense. You can repeat that argument at the primate level as well. An individual cell in my body does what it’s told at some level. It doesn’t have a whole lot of agency.

Joe: It’s true. And what are the immune cells killing themselves for you all day? Who’s paying them? What are they getting out of it?

Michael: Exactly. This is definitely some sort of very extreme dictatorship, much more along the lines of the ant colony. There are different levels of abstraction. We’re focusing at one particular level and saying we get property rights there, but it seems as though in order to get it there, at some other levels you actually have these little dictatorships going on. The cytokine response is doing its thing and your immune system is responding. I don’t understand very well how to think about those separations. You have a particular type of morality at one level and then it goes away completely at other levels. And that maybe might actually be necessary.

Joe: I think to some extent it’s unsurprising if you think that the right kind of norms—the dynamics that shape the equilibrium at a given level—are empirically contingent and alter at different levels. Then you should just expect lots of variation in terms of what morality emerges at a given level of organization, because different dynamics apply.

This would be my best guess. If you look at nature: sometimes there’s a bunch of cooperation and sometimes there’s a bunch of predation, and it really depends on the case. If you say it’s a fundamental law of nature that everything is predatory—well, not only, there’s some nice stuff. And if you say everything is cooperative—no, that’s not right either.

There is an interesting question of whether there’s a step change at some level of intelligence. My friend Katja Grace has this great post about why we don’t trade with ants. And her take is: well, you can’t talk to them. If you could talk to the ants, we could totally trade with them, get them doing stuff for us.

Michael: But we do know a lot about the chemicals. We know an awful lot about how they communicate. I’m not sure—maybe the trouble is on the other side. Maybe we can sort of send information to them, but they’re not necessarily that interested in talking to us.

Joe: Well, there’s some set of—we can manipulate the ants, and we do this with sheepdogs and stuff. We can make an animal useful to our purposes by training them. And I can imagine a case where we put down little chemical trails—I don’t know, what would you want an ant to do? Clean out your pipe or something?

Michael: Very often people just want them out of their house. They do use signaling to do that.

Joe: Right, you could just lead them out. As opposed to killing them. I was thinking something a little more like, if you’re trying to apply comparative advantage arguments to the ants, why—

Michael: Yeah, but I’m just poking a little bit at Katja’s argument to say that in fact we can communicate in one relatively limited way, but maybe not vice versa. They don’t really have any concept of us, as far as I know.

Joe: Yeah, that’s interesting. I think the barrier is that even if you can do some communication, there’s some notion of all the cognitive structures that need to be in place to have something akin to a deal. As I’m saying that, though, I’m like, well, how do we understand other forms of symbiosis? There’s a way in which the deal can take place at the level of selection rather than at the level of communication.

But I think it’s at least interesting: if you think there’s something unique about the cognitive equipment a certain level of intelligence gets you, such that you can do deals in some more full-throated sense, then maybe as we go forward, there will be more and more deal-making such that you’ll see more advantage to respecting boundaries or incorporating someone in a—

The fundamental dynamic that I think you would expect to crop up, that seems very real, is: it’ll often be the case that two agents will do better by both their lights by not fighting. I think that’s going to be a robustly real feature of the world. And it’s a happy feature. You have this great writing about happy features of the universe—things that aren’t deep-atheist-y. The existence of gains from trade, of mutually beneficial stuff.

Michael: Even just the fact that threat displays are so often much better than actually fighting in so much of the animal kingdom is at some level surprisingly encouraging. It all mitigates against a deep atheist point of view. There are a lot of non-trivial, very surprising ways in which the universe is actually better than you would have naively predicted.

Joe: Yeah. Some of the Adam Smithy invisible hand stuff is a bit like that. There’s a bunch of this. It starts to feel especially non-deep-atheist if it’s a decentralized, uncoordinated process that nevertheless leads to a good outcome. Because then it’s a kind of uncontrolled default.

Elinor Ostrom and governing the commons

Michael: One of my heroes is Elinor Ostrom, who wrote this amazing book, Governing the Commons—about exactly this, how decentralized groups can sometimes govern themselves for hundreds or even thousands of years, despite the fact that it looks like they should be in bitter competition with each other. Very interesting. Definitely very green, not deep atheist at all.

Joe: Elinor Ostrom is very green.

Michael: Yeah. It’s a funny word to apply to her book. She reads in many ways as a political economist—she won the Nobel Prize for economics. But fundamentally, really quite green in your terms. Not just in terms of the attitude she brings to it, but maybe more in what she discovers in the world. She goes out and points out things like, there are these irrigation areas in parts of Spain which have been very functional for nearly a thousand years—I think it’s 800 years or something. And there’s no real governance structure in the conventional way. It should have failed long ago, but they’ve figured out how to do it. They’ve developed these amazing governance structures that don’t eliminate competition and don’t eliminate argument. Instead, they’ve figured out how to resolve that, in ways that have basically worked for 800 years around Valencia.

That’s a discovered type of greenness in the world. And maybe even more positively, it’s interesting that you can take what she’s done and start to apply it in other contexts. I’ve heard people say this is the best book written about open source software.

Joe: So what do you take from it? Because it strikes me as a little bit like the predation-symbiosis case, where you have things that go in one direction and things in the other. And the real question is how do you predict or cultivate the conditions under which you’ll get the good one?

Michael: It’s a while since I’ve read it, but I don’t remember her talking very much about the instigation process. She has these—I think it’s six principles—which these structures have to satisfy. She’s more interested in finding extant examples in the world. She goes through a ton of different cases, cases where things have utterly failed, cases where things have worked, and tries to sort out what makes it work versus not.

But she’s not so interested in the genesis question. My general impression is that very often it didn’t require that larger group of people to really be focused on establishing a lot of the preconditions. Some of them are just classic liberal things—property rights which are respected. She wants very well-defined resources. She points out that when there’s a lot of ambiguity about what counts or what doesn’t, it tends to rip the systems apart. So she wants very clear boundaries. And she has a number of examples where that fails to hold, with disastrous consequences. So from a philosopher’s point of view, conceptual clarity is very important.

Joe: But not actually—that’s slightly incongruous with a broader green aesthetic. Interesting.

Michael: Yeah, it’s very much: you can make small, apparently self-organized groups do much better than you might expect without a traditional notion of property rights. There’s this discovered set of design principles that you can use. But she doesn’t really do a deep history of how you set these systems up from scratch, how they come into being. And that seems like an incredibly interesting question.

Joe: I’ve been trying to figure out how pessimistic I am about some of these self-organizing decentralized processes. I want to name that you can actually go the other direction in terms of deep atheism, where you have more allegiance to decentralized competitive uncontrolled processes but you don’t think they’re inevitable in their victory. Both of us have written about this in different ways. There’s a way in which some of the threats to the future come from a kind of uncontrolledness of the world. You’ve written a lot about the vulnerable world—having a lot of actors who have access to very dangerous technology that can destroy everything. That discourse leads very quickly to this very intensive need for control.

Michael: Yeah.

Joe: And in particular, political control—an especially scary form of control where you need to lock down the access of more and more actors to more and more stuff.

Michael: Or do surveillance. There doesn’t seem to be any particularly good solution.

Joe: Yeah. It’d be interesting to see if you get an Ostrom-like thing. There’s no iterative potential with these one-strike-and-you’re-out cases. I imagine self-organizing systems do better when there’s a chance to be robust and iterative and messy, where—no, you’re just never allowed to have this thing happen ever. It ends up being a worse fit.

Michael: No, yeah, all the violations in her examples are relatively small and relatively local. So you get a lot of benefit from that. You also get a tremendous amount of surveillance, but in this very friendly way where a lot of it is just being done by neighbors. They know each other. And the rules tend to be of the form: if two people have a conflict, get a third-party witness. If you’re dealing with fisheries or forests or these kinds of things, it’s not going to work a hundred percent of the time. But if it’s working 95% of the time, you’ve solved 95% of your problem. She leans a lot on that kind of argument in a very compelling empirical way. She’s not just talking about it in theory—she’s done the hard slog of finding out how it works in practice.

There is actually a really interesting thing in connection with your suggestion. The examples she chooses involve between 50 and I think 15,000 people, and they’re never huge. And I think maybe that’s quite discouraging for attempts to do it at real scale.

Joe: You’d mentioned open source software as a potential example.

Michael: Yeah, but again, open source projects tend not to involve many people. I don’t know how many people are contributing to the Linux kernel—I’m sure it’s a large number. And even there, they’ve transitioned more to something like a government model with relatively centralized arbiters who have a lot of power. The ability to veto or accept commits is already a governance structure, which maybe doesn’t have quite the same analog in an irrigation system around a city where you just have a lot of farmers separately irrigating. There’s certainly still some centralization of power, but the whole point was it wasn’t being delegated to the city—it was actually being done collectively by the farmers.

Vulnerable worlds and the limits of control

Joe: I’m curious—I feel some concern that I go around saying, there are risks from too much control and risks from too little control, you’ve got to look at it case by case, reality has a lot of detail. And then from there you can kind of—people have an intrinsic allegiance to one versus the other, they want to come in with a prior about, well, that involves control. I’m kind of like, is this even productive as a frame? It’s sort of, you can notice these tensions, notice the dialectic, but it’s not very productive to be like, whose side are you on?

Michael: I think my instinctive response is: you want an allegiance to closely watching the world and then responding contingently, not holding on too closely to your ideological priors. If it turns out that classic liberalism is taking you in the right direction, great. But if you start to see a whole lot of exceptions, you want to think pretty hard about that.

Joe: Yeah, I agree. Though—for example, the vulnerable world stuff is interesting, and I feel like you’ve been admirable in your attention to the painfulness of some of the tradeoffs that the vulnerable world dynamics implicate. Though I haven’t heard you go as hard as Bostrom on, the only solution to this is some intense—I remember you had a deep discussion of privacy at one point. Depending on how you set up the empirical assumptions, the vulnerable world thing can just—Bostrom does this by definition. The vulnerable world hypothesis is basically that set of circumstances such that you need this horrible surveillance apparatus in order to avoid existential catastrophe. And he’s just like, guys, there are ways the world could be such that that’s the only way you avoid existential catastrophe.

Michael: Or his other solution is a singleton, more or less. So you get—

Joe: Or isn’t—I think there’s some detail where he’s like, either you lock down the motivations or you lock down the access to the tech. He has a disjunctive definition.

Michael: Yeah, I think you’re remembering it slightly better than I do, but there are multiple alternatives, and you read through them and think, I’m not keen on any of this, thank you very much.

Joe: None of them are good. And sometimes the empirical circumstances are such that one is not keen on the necessary solution and one is just in a bad place.

There’s an empirical bet that one makes, which goes to your point about being attentive to the actual circumstance. But interestingly, when you talk about it, you often draw on your intuition. When we’ve talked about this in the past, you’re like, well, as a physicist, it’s very hard for me to argue about exactly why the world would be vulnerable, but I just have this deep intuition that physics will make this sort of technology available. And that’s an interesting case where it is a question of how much we want to bet on that intuition. I’m often—people’s threat models for the future become sufficiently like the vulnerable world, and then I sort of tap out, where I’m kind of like, well, that’s the vulnerable world case, that’s going to lead us to a global surveillance state. There’s some way in which I’m actually bracketing some of those as an especially hard case—let’s see if we can win in the other ones. But it’s interestingly difficult to make those bets at the level of rigorous empirics, because you just don’t have that.

Michael: Yeah, it’s of course about any possible future at this point, and that’s part of what makes it difficult. You’re implicitly taking points of view which are quite strong and quite hard to justify, no matter what your intuition says.

Death and sincerity

Michael: I want to come to a set of stuff around death and sincerity that you’ve written about that I found really interesting. You have this great phrase—I think it’s your review of Atul Gawande’s book /Being Mortal/—you talk about this observation that death often brings life into focus. Encounters with death: the person who gets cancer and all of a sudden they know exactly what they need to do, or they have very strong goals where maybe before they didn’t. It brings people into focus, and it also—I don’t know if you used the word there, but you’ve used it a lot elsewhere—it makes people more sincere sometimes.

I thought this was really interesting. You talk about having it as an aspiration for yourself. The most basic question, which I have some instinct for but don’t quite understand, is: why does it do this? And on the other side, why is it so hard to be sincere some of the time? You talk about fake thinking versus real thinking as well—a separate line of thought, but clearly very connected. It is very tempting to do fake thinking when you’re in creative work, which is a little bit detached from the world. Why is that the case? Why is the threat of death so helpful here, and how do we simulate it in other ways?

Joe: I love this question. Partly—it’s interesting, with some of these terms like sincerity or fake thinking, it’s almost this question of: why would you do the other thing? What’s good about the other thing?

Michael: You can give signaling answers to that question—it’s about achieving social goals, so you pretend. And you give some nice examples where it’s very hard to believe that’s really what’s going on. Somebody has chosen to do something low-status that other people aren’t very interested in, that they’re not going to get paid for. And then they still find themselves doing fake thinking. Why? The social explanations just don’t make sense to me. I think you point this out quite compellingly.

Joe: I think one hypothesis is that—

When I’m doing bad fake thinking as a philosopher, the reason I’m doing it is because it’s easier. It takes a kind of energy to—instead of working with some framework that is familiar and kind of chopping the logic—be like, no, what is true here? And then I’m like, but that concept is a little fake, isn’t it? And I can see myself wanting to be like, can’t we just work with that? And I’m like, what if we really—is that really right? There’s a way in which it’s burdensome and effortful to have to try to refactor your ontology to make sure that you really believe in the way you’re thinking about something.

This sometimes happens with other forms of spirituality too, where people—I don’t want to think about death today, I don’t want to try to live my best life, I just want to watch TV. And there’s a way in which that is easier. If you’re going to get out there and have a bunch of meaning, you’re putting pressure on yourself.

But the death case—let’s focus on that. That’s a great example. Why does the consciousness of death—and this is Heidegger, he has this whole thing about being-unto-death, it throws you out of the das Man/—I think /das Man is actually his analog of inauthenticity, similar to fake thinking. You’re caught up in some social plane, and the contact with your own death throws you into your bare existence and consciousness of your existence as Dasein, something like that.

But what is this thing where you get caught up in the das Man? What is inauthenticity? What is bad faith? What is shallowness? There are all sorts of non-ness—non-there, non-real—that are a deep part of our lives and our baseline ways of being. Things I would guess: one is there’s clearly something going on with not wanting to look at things. If your brain is constantly doing “I can’t look at that, I can’t look at that”—or if it has a bunch of inner conflict that it doesn’t want to resolve—there’s a way in which, if your mind is not able to be coherent and sit with itself, and is instead bouncing off of things all the time, that contributes to various forms of unreality.

Michael: Can I give what might be a concrete example? A lot of scientists, particularly often in an academic context, will get quite frustrated because they’re very curious about a set of things, and they get used to the idea that if they apply for grants to do this, they get turned down. Instead they’re doing something else that they know they can get funded for. But it’s adjacent—it’s not actually what they want. I have some sense they’re having to fool themselves all the time, and so the work becomes a little bit insincere, it’s a little bit more fake thinking. Does that seem like an example where this kind of bouncing off is likely almost a survival strategy?

Joe: Yeah. The way I would put it is: when something is harnessing fewer aspects of your being, it often becomes faker. The paradigm of being really present with real thinking in my head is: your whole system is firing on all cylinders. You’re really fully present. I think same with something like an encounter with a terminal diagnosis. There’s a way in which it unifies your system towards an end. Whereas something like doing work that you don’t actually want to do—there’s some part of you that is like, I’m not bought into this project, I’m out. And so you’re losing a part of your brain.

This is a hazy model, but I think at least one dimension of being real, authentic, sincere is getting all parts of your cognitive resources to be together and look at each other and be willing to be in the same room and doing something together. And if half of them are noping out, or pretending they’re not there, or going to sleep because they’re bored—then you’re less alive.

Michael: It seems as though at least part of what’s going on—I mean, we both do this very abstract creative work, and at least some of the time, you don’t know what you’re doing. You’re pursuing something because you have to. Parts of this conversation have been very speculative, but you’re entertaining possibilities in the sincere hope that it does lead to progress on goals which you really deeply care about.

I’m just thinking about death—I suspect a lot of people essentially quit their job when they get a terminal diagnosis. Sometimes just for practical reasons, but sometimes for meaning reasons. It’s interesting that for some other people, it’s the opposite. I believe George Orwell wrote 1984 after being diagnosed with tuberculosis. And there are a number of examples where somebody really—it seems to have concentrated their attention enormously in the way you’re describing.

Actually, both of those things are consistent with what you’re saying. It’s forcing you to choose. Am I really all in on writing this book? Or do I really want to go and spend every waking minute with this person or that group or in this location? It’s a symmetry breaker.

Joe: I still love this question of why does it do that, and why do we not ordinarily do it? Because we’re all dying all the time.

I think one thing playing a role is resource preservation. When I’m doing bad fake thinking as a philosopher, it’s because it’s easier. It takes energy. And I think there’s a way in which trying to have a certain kind of spiritual orientation stably—it’s burdensome and effortful.

Michael: Kevin Kelly has this great story about how he decided to set himself a deadline six months in the future, and that was going to be his death date. And he was going to try and live literally as though he was going to die then. He gave away his money. He really massively restructured his life and he tried as hard as he could to prevent himself from thinking about things after that date. So he’s gone through a very—it’s a LARPy kind of thing, but which he took very seriously—a practice death. And I wonder to what extent that becomes a reusable resource, in the way you’ve talked about wanting for yourself to live with that kind of sincerity all the time. Maybe that’s a way of permanently increasing your sincerity or living more fully.

Joe: That’s really interesting. How much in the future was it?

Michael: Six months. Which is a good period of time. You can do a lot in six months, but you’re also going to be thinking a fair bit about it if you know you’re going to die in six months.

Joe: Yeah, that’s crazy. I feel like if you were actually doing that, you would just make all sorts of mistakes. There’s something often wrong about pretending that something is high stakes when it’s false. But it certainly seems interesting.

I’m sort of skeptical that it’s possible to—at least in my experience, the effort and the project of trying to have a certain kind of spiritual orientation stably does not ever involve a single durable “and now I am always something.” It’s more like diverting a river, where your mind has some—you can have epiphanies that change how you are, and it has some structure, almost like a key that opened some lock in your mind. That does change the structure of your mind. But if you keep hitting that motion over and over, it wanes in its impact on your orientation.

That idea—at the time you might have thought, if I just remember that I’m going to die in 30 years—sometimes I’ll watch a movie and come out with some intense sense of existential urgency. And then why does that go away? There’s some way in which your brain—it’s a novelty thing maybe, or it gets the message and moves on, or your physiology changes. But in my experience, it’s much more about having an ongoing injection of effort and new ways of coming at it, and continually, slowly, patiently becoming more the type of person you want to be, rather than a thing that always works.

Michael: For Kelly, at least in this particular case, there was an ongoing process. There was always this event slowly marching closer one day at a time, where every day he had to get up and think about how he was restructuring his life. He apparently sent a whole bunch of checks for $500 or $1,000 to friends anonymously—basically spending down his relatively meager wealth at the time. And just the choice of where am I going to be, who am I going to spend time with—it sounds like he considered that very deeply over six months, reconsidering and actually moving himself around. So in some sense it was a stable focus that required continual decision-making, which is a slightly different orientation than what you’ve described. Maybe it was very helpful.

Joe: Yeah, I’m often more optimistic about practices—something where you do a thing and you actually have to be doing something each day that is renewing. Just having had a thought, or even an experience, or even a quite profound experience—at least it’s harder for that to have a lasting impact. And even the thing I was thinking—I don’t know what happened to Kevin’s attitude towards death after he passed this point—I imagine it was a quite formative experience, but I would also imagine that it is nevertheless an ongoing challenge, insofar as there was some sincerity that he gleaned from that, to hold that going forward over many years.

Michael: Can I ask you then: what do you feel has increased your own sincerity or your own depth of engagement with the world the most? Is there any practice that you’ve found that you’re unexpectedly pleased with?

Joe: Certainly death is one instigator. I’ve done a bunch of meditation and adjacent practices, which I think were a little less about cultivating sincerity and more about something like presence—a kind of prerequisite poise in engaging with the world.

I think actually some kind of ethical inquiry has made a difference for me, in the following sense. A lot of what can feel like insincerity is if you have a kind of background agenda that you’re not acknowledging. There’s some exercise where you go: what actually am I doing, all of it? And you try to have a full accounting, where there’s no secret extra thing that you’re not looking at. You’ve brought to mind and integrated all of your motivations. Obviously you can’t excavate every unconscious whatever, but something where you’re not holding back.

There’s no “I’ve got my life, and then there’s this other thing I’m doing on the side.” It’s like, no—there’s only this one chance. If the thing you want is power and status, or you want to play video games—whatever your secret agenda that you just want—you can do that. Go for it. Sincerely. You can integrate the full range of your motivations into a sincere orientation towards your life. I think that helps partly because it starts to seem less like some motivations are privileged as part of your official internal narrative and some have to be in the background. It’s more like you can have your whole life and really ask yourself, what actually matters? How do I actually want to make tradeoffs? You can ask that question with your whole being.

Grace as integration

Michael: I just happened to see a little video involving the soccer great Ronaldinho. Something that has struck me often about him is how relaxed he is. He shows no hesitation in his physical movements. He seems to speak very freely. Obviously I have no idea, but it seems very consonant with what you’re describing. He seems very integrated in at least some ways. I wonder if that’s connected with his ability to bring everything he had to his football.

Joe: I think the notion of grace is a very interesting analog of some kind of integration. I always use the word poise, and this is often a word I use for a kind of mental integration. But I think you see it in really beautiful dance and beautiful physical movement—a kind of non-conflict between parts of the system. The whole thing is working together.

Maybe he’s managed it at the mental and the physical level, and they’re integrated. It’s interesting—often beautiful, graceful movement—but often there’s a deep connection between the mental and the physical.

Michael: I think we’ll end there, on a balance between mental and physical grace. Seems like a very nice place to end.

Joe: Awesome.