Bits and bricks: Oliver Habryka on LessWrong, LightHaven, and community infrastructure

I'm joined this week by Oliver Habryka, who runs Lightcone Infrastructure—the organization behind both the Less Wrong forum and the Light Haven conference venue in Berkeley. The conversation ranges from reviving an internet forum to the fractal geometry of a hotel built through organic accretion, to why Berkeley's building codes list "plugging in a lamp" as one of twelve permitted activities, to the surprisingly difficult work of fundraising for intellectual infrastructure.[Patrick notes: As always, there are some after-the-fact observations sprinkled into the transcript, set out in this format.]

Complex Systems is producing more video episodes like this one. In addition to this site, you can access them directly on YouTube. My kids inform me that I’m supposed to tell you to like and subscribe.

Sponsor: Mercury
This episode is brought to you by Mercury, the fintech trusted by 200K+ companies — from first milestones to running complex systems. Mercury offers banking that truly understands startups and scales with them. Start today at Mercury.com 
Mercury is a financial technology company, not a bank. Banking services provided by Choice Financial Group, Column N.A., and Evolve Bank & Trust; Members FDIC.

Timestamps for Video:

(00:00) Intro
(01:44) The origins of LessWrong
(04:28) Challenges of running an online forum
(06:28) Reviving LessWrong
(15:12) The importance of conference venues
(25:55) The complexities of venue management
(37:45) The impact of wide roads on American car culture
(38:12) Fire safety regulations vs. traffic fatalities
(39:24) Microeconomic analysis in policy making
(40:47) Building codes and staircase requirements
(41:57) Berkeley's shift in construction policy
(43:48) Fundraising challenges in the nonprofit sector
(44:58) Effective altruism and fundraising dynamics
(47:02) The collapse of FTX and its impact on fundraising
(52:52) Strategies for successful fundraising
(55:08) Lessons from religious fundraising practices
(01:05:49) Fundraising experiences and challenges
(01:13:37) Wrap

Transcript

[A brief note on our transcripts] 

Patrick McKenzie: Hi everybody. My name is Patrick McKenzie, better known as patio11 on the Internet. I'm here with my buddy Oliver Habryka.

Oliver Habryka: Nice to meet you, Patrick. Well, good to see you again. I'm glad to be here.

Patrick McKenzie: So you run Lightcone Infrastructure, which runs both the LessWrong site online and also this conference venue, gated community... what was the word you used for it?

Oliver Habryka: The New York Times loves to call it our "walled, surveilled compound." Because we have walls and we have security cameras. And it's more than one building, which of course technically makes it a walled surveilled compound.

[Patrick notes: That was the Guardian, not the NYT, which I mention so that Complex Systems doesn’t have to issue a half-dozen corrections. We have editorial standards at this publication.] 

Patrick McKenzie: It is a conference venue and it's a very good conference venue. We'll talk about conference venues in the future, but A++, would hold a conference here.

Oliver Habryka: I'm glad.

Patrick McKenzie: It's a nonprofit entity, as I understand?

Oliver Habryka: That's right.

The origins of LessWrong

Patrick McKenzie: We'll talk a little bit about the joys of nonprofits later, given that both of us have had the experience of fundraising for them. But let's start out with talking a little bit about LessWrong for an audience that might not be super familiar with it.

I think it is relatively well known in the tech industry in the same fashion that Scott Alexander is relatively well known in the tech industry, but the further you get from it, the less it is appreciated. LessWrong is kind of the beating heart of the rationality community as it exists in online forum form. But as someone who is not a member of the rationality community, but has read you guys' stuff over the years, I think it is an enormously useful resource for the world. Do you want to chat about it a little bit before I give people reasons why, if you read the New York Times, you should also read LessWrong occasionally?

Oliver Habryka: Yeah, sure. As a very rough summary, just so people who aren't familiar: LessWrong is an online forum that got started around 2008 to 2011 originally. Eliezer Yudkowsky, who just published his latest book—you might have heard about him in that context—started blogging together with Robin Hanson, a relatively widely known economist, on his blog Overcoming Bias. He started becoming more and more of the daily posting schedule until approximately Robin one day told him to get his own place, at which point he started LessWrong and figured at the same time he might want to invite other people to blog with him similarly.

Around 2011, this forum slash community blog got started, originally using a fork of the Reddit code base. So it was based off of the basic Reddit system where you have long form posts but people can vote on them and people can submit them to a forum. During that time he wrote lots of stuff about AI and importantly the core of the rationality community—a lot of ideas around cognitive science, a lot of ideas about how to think better, a lot of trying to get into the practical details of what it means to form accurate beliefs about things.

He wrote a lot of stuff. He started attracting a lot of other writers. Indeed, Scott Alexander around that time started showing up on Overcoming Bias and then LessWrong and started his blogging career more properly there. He posted a bit on a few other small LiveJournal blogs, but I think LessWrong is where he probably got started.

Roughly until 2015-2016, Eliezer at the same time was writing Harry Potter and the Methods of Rationality, which people may also be familiar with as a very absurdly popular fan fiction about a scientific slash STEM-y approach to what Harry Potter might look like. He started writing that up until 2015.

Challenges of running an online forum

Roughly after that, he realized the truth that many people will realize at some point in their lives if they decide to run an online forum, which is that there's approximately nothing as stressful in life as running an online forum. I actually find there's a quote by Paul Graham about it who writes about it on Twitter. He said all the most stressful parts of founding and running Y Combinator—90% of the time when something would be on fire and I would be stressing out and unable to sleep at night—would be the result of Hacker News stuff going on. Not any of the companies which were reaching billion dollar valuations or engaged in highly transformative technologies. Hacker News was by far the most stressful part of running Y Combinator and a stressful part of my life.

I think similarly running an online forum is stressful. There was a fun situation around the time that people might have also heard about—Roko's Basilisk, where a user called Roko suggested a thought experiment about what if a hypothetical super intelligent AI in the future might want to punish everyone who didn't help bring it into existence. Eliezer was basically like, "Well, I don't think that works out, but also aren't you just making things worse if you talk about this? So please go talk about it somewhere else." He told Roko to shut up, to which Eliezer, at the time very naively unfamiliar with the Streisand Effect, produced one of the greatest memes of I think the modern internet, 1000x-ing the attention that Roko's random post would've gotten.

Importantly, he had to deal with the moderation fallout of that for multiple years, enough that he basically gave up and was like, "I'm not dealing with this stuff anymore." He left roughly around 2015 and then LessWrong started a pretty intense decline, with the site basically completely dying to trolls and spam. The Reddit code base was no longer properly functioning and the website wasn't properly working on mobile, now that during the 2010s basically everyone started using mobile phones to browse the web.

Reviving LessWrong

That's roughly when I entered the picture. I was like, "Oh, actually there is this really great resource, just one of the richest places on the internet for interesting intellectual discussion of many things that I deeply care about." I basically asked Eliezer for permission and a few of the moderators whether I could give it a try to revive the site. That's where I entered the picture around 2017. I started LessWrong 2.0. I had just been in the last year or second to last year of my CS undergraduate degree. I was not a very good programmer and tried my best to make an okay working website and then did indeed successfully revive it.

These days, LessWrong continues to be one of the... it's one of the places on the internet. I don't know what top X places it is, but my best guess is something like top 50, top 30 places that is part of what makes the internet the internet.

Patrick McKenzie: Yeah. I think one of the least remarkable things about it in the sense people would say at least is that all forums go through a Great Filter, and the vast majority of them are garbage collected by that Great Filter.

[Patrick notes: “Garbage collection” when used by a computer engineer is a metaphor for the automated process which removes things from memory after no program is using them any more. I mention this because occasionally people hear the metaphor to mean referring to the GCed thing as “garbage” in the casual and pejorative sense, which is not my intent. Almost all literal garbage was once something useful, and the literal garbage collection is part of the economic lifecycle of that thing, which removes it from being a present burden on the person/firm possessing it and releases it to its future economic life in a waste stream.]

Patrick continues: There is Hacker News that made it through. As a longtime power user of Hacker News, whether it is fantastic in any given year is heavily dependent on the year. But there is an installed user base that has continued forward into the future.

LessWrong has continued forward into the future, but there were tens of thousands, hundreds of thousands of forums on the Internet, and most of them never achieve a nucleus of a community that can sustain it after the initial group leaves. Most of them never develop a distinct culture around them. So it's already an achievement to have any forum survive into the present over a 10 year period.

This particular one is perhaps pound for pound, perhaps character for character, one of the intellectually densest and most consequential web forums, or even might be one of the intellectually densest and most influential “scenes” that exists in the world. I sort of see it as the inheritor of the Parisian salons. They were a site for discourse. Much of our understanding of that century comes down filtered through a relatively small number of people in a relatively small number of rooms drinking a relatively small amount of coffee together.

I am not merely saying these geeks are my tribe and I love them for that. The observable track record on LessWrong, because everything is on the Internet forever, is impressive. If one compares LessWrong as a method of understanding, say, COVID in early 2020 to the CDC's official pronouncements or the pages of the New York Times or Vox, LessWrong stacks up very credibly in that analysis. I don't think that's the only time that that has been true.

[Patrick notes: I have previously described Vox’s reporting as “analysis that aged more poorly than perhaps anything in the history of the written word.” Apparently some reporters at Vox bristled at that, because they reject the notion that the publication doing important journalism was responsible for, quote, the social media team, endquote. Those of us in the tech industry who were right and early have long memories for being called cryptoracists by professionals in the pages of their publication. Sorry folks, you were wrong and late, and that does go on the permanent record if you pretend to the mantle of journalism.]

Patrick: Obviously it's been quite a focus of the community for a very long time that AI was going to be here soon and be incredibly powerful, for better and (according to some people) for worse. The community did successfully collect a lot of people's attention on that prior to that being the consensus pick. To the extent that it is not the consensus pick, I think that people who have lurked for 10 years as I have have successfully anticipated some of the things. [Patrick notes: “Lurked” is forum lingo for “passively participated via reading but did not themselves post.”]

Patrick paraphrases, in the voice of a longtime LessWrong participant: Stochastic parrots? Anybody who says that phrase is an idiot. That'll last for weeks if that. 

Patrick continues: You can 10 years ago have figured out things that people will be surprised to learn six months from now, and it gives you a collection of mental tools for the surprises that we're all going through at the same time.

Okay, that's enough of my sales pitch for your free service that is available on the Internet.

Oliver Habryka: Greatly appreciated.

Patrick McKenzie: The fact of it being a free service available on the internet—there's any number of institutions in society that could have done it and didn't. I love universities. I love them as a place for exchange of intellectual ideas, for educating the upcoming generation, for fostering an open culture of debate and research. And yet, you know, Berkeley didn't do LessWrong. Oxford didn't do LessWrong.

Oliver Habryka: I have a whole set of rants about this. I think the craziest thing is something like—arXiv is a university run project, but they're notoriously underfunded. It's basically one guy who's running arXiv. It kind of drives me insane. If you are a major modern university and you find yourself being the university that hosts arXiv, elevate it to a department. It's obviously the backbone of a large fraction of the modern scientific endeavor. But somehow the university can't... successful infrastructure these days just doesn't get built at the universities anymore. It's a very weird and very intense fact. Even outlier successes like arXiv just fail to get support that would be at all reasonable.

Patrick McKenzie: I don't want to make this the complaint about the rationality and universities episode of the podcast, but there are ways in which we make clearly suboptimal decision making at the individual organization level and at a societal level of infrastructure projects to back. Universities spend how much money on journal subscriptions by Elsevier? [Patrick notes: Plus or minus three billion dollars a year. I’m not particularly offended by that, because knowledge development is valuable for humanity, but that’s an interesting anchor relative to scandalously underfunded mechanisms for the same.]

Oliver Habryka: They actually have their office in downtown Berkeley. We considered renting right next to them.

Patrick McKenzie: Elsevier is widely perceived by academics to be a rent seeking institution. But Elsevier has a sales process and a channel worked out with universities where every university builds a few million dollars of subscriptions into its cost base every year. No problem. And then people have trouble keeping the lights on for arXiv, where arXiv is carrying a large portion of capital-S Science on its back with one guy. [Patrick notes: That situation has likely been rationalized in the last decade or so, based on a quick skim of their Wikipedia article.]

In a similar example which is also adjacent to an Internet forum: while less important than science with capital S, a thing that universities put no small amount of work on in the United States is debate as a practice and academic activity. The debate community had a forum, and every year there was a worry that the debate community's forum would go under for want of $20 a month of hosting costs. This is an activity where universities are sending teams of undergrads accompanied by a salaried sponsor at the university—typically a communications professor or something, often a tenured individual—sending them to another state with all the attendant costs associated, and then the community organizing method slash watering hole, the primary way you would find out about new tournaments, went offline multiple times because nobody involved had $20 on a purchasing card available.

When I graduated university and went to Japan and earned the standard 20-something salaryman fortune of $2,000 a month, I said it is absurd that our community has not self-funded the $20 a month reliably in the last four years that I've been involved in this activity. Just allow me to be the administrator emeritus please, and I'll make sure the $20 a month gets paid. That is how I ended up hosting a university web forum for 10 years while gallivanting around Japan. [Patrick notes: I eventually handed over the reigns to a new administrator who had more recently been in the debate community. He tragically passed away, and I do not believe the community was able to sustain that particular watering hole.]

The importance of conference venues

Going back to infrastructure, the nonprofit has a couple of activities. One of them is covering for systemic irrationality in people funding web forums. And the other one is running the space Lighthaven at the one that gets so much skepticism in The Guardian and the New York Times. It's just a conference venue. But I don't want to undersell the importance of a conference venue. The Parisian salons were a place to pay money for getting coffee and ended up being a major impact on the world. This has ended up hosting some conferences, making some relationships happen, et cetera, and supporting that is a whole lot of work inside the world of atoms.

Oliver Habryka: So much.

Patrick McKenzie: Do you want to tell some war stories? Let's give people some context. What is this space?

Oliver Habryka: Lighthaven, previously known as Rose Garden Inn, was a hotel in downtown Berkeley. I actually found this place in something like 2014. The Center for Applied Rationality, which was an outgrowth of LessWrong at the time, was running workshops for rationality and trying to help people think better based on cognitive science. They used to run workshops and one of the places they used to run at the time was Rose Garden Inn. That's how I found it.

The thing that makes it really, really interesting is that it is this—you're close to downtown Berkeley, so you're in a quite urban environment, but it's actually around five buildings, all in an essential courtyard. Those five buildings are all extremely weird. There's a way I like describing the Rose Garden Inn as a historical property, which is that it is mostly cancer, in a sense that it is an organically grown object that at every point in time was trying to eat itself and trying to expand.

I think literally most—a large fraction, my best guess is something like 50% of the current rooms in this property—are the result of an evolution where there was a house. The house at some point wanted a porch, so they added a porch. Somewhere in that house wanted an additional room, so they took that porch and then built walls onto that porch and made a room, at which point they wanted to have a new porch, so they built another porch. At which point the second floor of the house above maybe also wanted some kind of balcony, so they built a balcony on top of the previous porch that is now a room.

That has, as far as I can tell, explains about 50% of the building volume and mass, which produces a kind of construction style that I would only describe as “fractal.” Every single wall you look at will look different from every other wall. Every single nook you look at will have a completely unique shape. There's probably not a single grid you can overlay anywhere on this whole property such that anything will conform to the grid in a nicely neat way.

Patrick McKenzie: There's one of the buildings that has various Harry Potter themed names of rooms in it, but it feels like stepping into a Harry Potter movie set, where you're abiding by some sort of magical logic as to where stairs should be. That's how one gets from point A to point B.

I think—and you're welcome to your own point of view on this obviously—that is one of the reasons why it works so well as a conference venue, because it is not a university campus that was laid out by a particular designer that was optimizing for the ease of transit between Hall A and Hall B, each of them named after a sponsor. As a result, the sort of accidental, metaphorical, or literal collisions of attendees with each other as they're going between talks creates the best sort of structural accelerant for the hallway track of any conference venue I've ever been in.

Oliver Habryka: I have this whole rant. I find it almost amusing when people try to be like, "The hallway track at conferences is the best part." I'm like, doesn't that scream insanity at you? You're telling me that you have your conferences and the part that is the best one is the part that's happening in the hallway. Have you considered moving that part somewhere that isn't the hallway? It feels like almost as if it's treating it as an accidental thing, where I'm like, the thing that you're supposed to be doing is you're supposed to make your whole conference the hallway. So that it becomes a place where people run into each other and meet with each other and stuff like that.

Then these events should be—it makes sense sometimes you want to listen to a talk if you want to have these anchor events where cool people show up to create an initial ability for people to realize the conference is real and give it credibility and all that stuff. But Hallway Track always felt to me as this thing where, well if you are still talking about hallway track, you clearly aren't running a good conference because a good conference would've made it so that you had any other place to have the good conversations that isn't a hallway.

Patrick McKenzie: If we've had 20 years of running various conferences in various venues and repeatedly coming back to this, we should have some level of meta rationality as conference planners to say, okay, given that I know attendee-on-attendee interaction is a focus of the conference for many people, I should pick a space that will make attendee-on-attendee interaction better than the typical space available in Las Vegas or similar.

Conferences are a business for many people that run them. Venues are certainly a business. We have the venue we have available, selling it is our business. When 40 of us in a city make that calculation, your choices are one of the venues that is available, and they're all designed to essentially be not carbon copies of each other, but extremely samey, because any deviation from the norm makes it harder for them to get the marginal sale.

It's nice that of all the conferences that I've attended, and many of them are quite wonderful and been important to my life, the ones here at Lighthaven have a particular vibe to them. I think it is a vibe that a lot of people would actually enjoy and seek to have in their conferences if they knew it was available.

 The complexities of venue management

Oliver Habryka: Yeah. It kind of reminds me—we'll probably go into this later in the construction world and also the conference world—the thing that keeps coming up, also related to the name of this podcast, is just an extreme pressure that you have against complexity in the venue layout.

One of the things that's the case is that conference organizers, the vast majority of conference organizers tend to be very far from the decision maker who organized a conference. Conferences are relatively unique in the sense that usually organizing an annual conference is a pretty high level executive decision. Stripe will run an annual conference—that's a pretty big executive decision. [Patrick notes: I previously worked at Stripe, am still an advisor there, and will not interrupt a guest to say “Stripe does not necessarily endorse representations about its decisionmaking process made on Complex Systems.”]

Oliver continues: That's a big deal that we'll do a lot to our brand. But the people who run and actually make the venue booking decisions tend to be quite low level. The average person who shows up tends to be someone who is a random office person. Maybe they have some event running experience, but they tend to be—especially in Silicon Valley, my basket would be on average five levels of delegation removed from the person who makes the call to run a conference.

So now it's extremely hard if they want to show up and be like, "What if we run our conference at this weird farm or this place that has a bunch of different buildings and different houses and has this fractal structure to it?" Propagating that information up to the executive level so that they can make an informed decision about that being what they do with the big conference is approximately impossible.

You have a huge "you can't get fired for buying IBM" problem. You will buy the conference venue whose layout—where you can describe the layout of your conference and the structure of your conference in three pictures. Because you can be like, "Well our conference venue has eight rooms, which allows us to have four tracks. The biggest track will be 200 person size, the second track will be 100 person size. Then we'll have a ballroom where we'll facilitate networking events or whatever." And you can be like, "Yep," and the complexity of the venue is literally there's a rectangle, you can see it fits 400 people, there's a rectangle, you can see it fits 200 people. That's the complexity that you get at selling venue decisions.

Patrick McKenzie: I think the industrial organization of this in Silicon Valley is interesting too, because for internal teams that are running a conference, that conference is very likely once a year. The person who might be say 24 or 27 years old, several levels removed from management who is running this conference—running that conference is their only job for the year. It is their only job because running a conference is just a huge amount of work. Every detail that shows up at the conference is something that someone needed to put in a planning document and check through the Gantt chart of doom for six to nine months at least.

Given that, and given that your job is on the line every year on successful execution of the conference, you are incentivized to—”nobody gets fired for buying IBM”—do the lowest risk things. You get two or three points of differentiation for this year's version of the conference. Maybe if you're lucky, you get 20 minutes with the CEO to discuss each of those two or three points. [Patrick notes: Or at a very special company perhaps quite a bit more than that.]

Patrick continues: But you are probably not going to put venue selection in as one of your things, both because of low perceived impact on whether you do a good job this year, and also huge outsized risk. Because if you violate the standard playbook, then everything else in the playbook is downstream of that.

Am I going to have to reinvent everything we do for catering here? Well, I've just bought myself eight weeks of calls with vendors. Am I going to have to reinvent everything we do with respect to physical access of the conference venue, et cetera, et cetera, et cetera.

So the Rose Garden Inn, this beautiful venue in Berkeley, you were able to just buy it. You sign the check, you change the name on the door and you're done, right?

Oliver Habryka: Yeah, it was perfect. Basically no blemishes at all, no problems. [Patrick notes: Oliver is executing a perfect deadpan here.]

Definitely not all bathrooms filled with mold, definitely not me training myself to become a literal mold dog, in which I learned the skill of trying to lie on the floor in the right places of the venue so that I would get a headache at roughly the right rate so that I would notice the headache in time and not get too much of a headache so that I couldn't then do the rest of my day of work of trying to fix the associated problem.

Patrick McKenzie: How many bathrooms are there on this property?

Oliver Habryka: Well, it's a hotel. It has around 50 bathrooms. It has about 50 bedrooms, approximately each one of which has an en suite. Some of which were broken. So about 45 actually, probably 50.

Patrick McKenzie: The thing I learned recently when talking to someone is that there's a full-time maintenance staff now, and they explained the math of that. I'll let you take it away with that prompt because it is a useful thing to know about the world.

Oliver Habryka: Yeah. I guess the basic way I would think about this is roughly just in terms of error rates. At a hotel venue that has roughly 50 to 60 bedrooms or a hundred beds or something in that order, I think just on average we have a plumbing problem... it's kind of hilarious. You can imagine plumbing problems happen when you're dealing with 200 people at a venue with a hundred bathrooms. We're talking about a toilet is clogged somewhere, a sink is overflowing somewhere, literally every 45 minutes or something. At the busiest events, we would end up having a Slack channel that's just plumbing problems. Just that alone keeps half a staff busy during the busiest events.

There's really a lot of fun maintenance. One of the things that I'm honestly glad about is the very temperate weather in Berkeley. But even at temperate weather in Berkeley, just especially for a property that's constructed like ours, you will get some small amount of leak. You will get some small amount of flooding. You will get some small amount of a wall isn't working anymore in the way it was intended, so you need to fix it and fix those issues. Not to mention the insanity that people do themselves in the rooms.

Patrick McKenzie: You have the standard ship of Theseus problem for any large constructed artifact. You have people who are not at their home or workplace and therefore might be doing things that they would not do at their home or workplace, or not exercising the level of care they might at their home or workplace. For every X, you have vastly more instances of X than people will typically have in their built environment. You get to experience the mean time between failure of X at a much higher rate than people do in their personal lives.

If typical consumers call a plumber once a year in our personal lives, even assuming nothing but steady utilization of the property, you would be calling a plumber weekly here. But the business model is the opposite of steady utilization on the property—extremely intense use of the property periodically. It is a never ending project on that. Then I assume you layer on that the complexities of running commercial real estate and the complexities of building anything in the physical universe.

But thankfully Berkeley is the most pro-developer, pro-building jurisdiction in the United States, right? [Patrick notes: One good deadpan deserves another.]

Oliver Habryka: That's right. It is beautifully perfect. I mean, basically there exist no laws as far as I can tell. You can just erect a building and then you can live in it. Ultimately, all the cost will happen basically as a result of labor and materials and not in the one and a half years of average turnaround times it requires for the city of Berkeley to approve the construction of anything that remotely looks like commercial real estate.

Patrick McKenzie: Maybe being less sardonic for people who don't understand—that is not exactly how it works. Formally in many places in the United States, if you touch basically anything that is load-bearing in a building, you have to get your plans reviewed by a city office for that. Lead times are what they are. Living in Chicago, I'll not say that the lead times are optimized for homeowners in the state of Chicago, but homeowners vote, businesses do not themselves vote. To the extent that anyone is having a happy experience, homeowners get a happier experience than commercial real estate typically does. Although perhaps commercial real estate has a savviness advantage or ability to bring professionals to bear on their behalf.

You did quite extensive reconstruction of this property, which was in not great repair. You changed it from a business that was focused on selling people nights in a hotel to a business which does sell people some amount of lodging, but is built around the conference business. So you changed the character of many of these buildings in the walled compound. Do you want to give any anecdotes on here's the thing we did, it seemed easy, and then...

Oliver Habryka: Yeah. One of the things that helps to understand is there's a website on the city of Berkeley which is "things that do not require a permit." It is an exhaustive list of all things according to the city of Berkeley that you can do to a building that do not require you to get a permit. That list is about 12 items long. One of those items, which it explicitly specifies, is you're allowed to plug in a lamp. Just so you know, lighting receptacles, plugging into pre-established and permitted outlets does not require a permit, just in case you didn't know. It's on the list of 12 things that do not require a permit.

[Patrick notes: This list very much exists, but is slightly longer. And indeed, “replacement of lamps or the connection of approved portable electrical equipment to approved permanently installed receptacles” is exempt.] 

Patrick McKenzie: We've used one of our 12 slots.

Oliver Habryka: One of our 12 slots. The kind of thing that is on that 12 item list is not at the level of abstraction of... it mentions "repairs" as one of your items, which is important, and that actually covered practically everything we did because repairs were really almost a central thing that we had to do. But it's very important to understand—the list of things that you're allowed to do to a building without the city getting angry at you is really very limited.

It's a bit hard to say what were ultimately... I think because we ended up doing repairs, we actually ended up very luckily interfacing surprisingly little with the city of Berkeley permitting system. Partially because we took it as an optimization criteria. When we worked on this and when we looked at this, we tried to understand really in a lot of detail what can we do without ending up with issues, in terms of falling outside of the things that are okay to do, and which things are we not. That ultimately basically determined the full project plan of everything. There's huge changes and huge differences I would like to have made.

Patrick McKenzie: I'd love to underline this point that there is a legitimate reason why cities have planning offices and some level of permit review. It protects both the current owners of the property, the future users of the property who might not be owners, and then future owners, given that it maintains a I Can’t Believe It’s Not Blockchain of state changes to the property.

[Patrick notes: The more technical term for that in computer science is an append-only log, but if you are currently storing your append-only property records in a fast database, there are many people not too far from Berkeley who would love to sell you a slow database to replace it. Though, in fairness to the crypto people, more clueful people in the community have realized there is minimal merit to this sales pitch, and they might feel aggrieved that I’m bringing it up to nosetweak them.] 

Patrick continues: However, the laws that we make very literally construct our built environment. There are compromises that you had to make due to the combination of the law and Berkeley's level of resourcing slash level of competence in enforcing the law that suboptimize for the experience of Lighthaven attendees, conference organizers and similar.

Oliver Habryka: One of my passions for the last—I think basically starting with Lighthaven—is to try to chase down the consequences of fire regulation. I think it's easy to think about... there's a few different things. Basically when you think of permitting stuff, the city of Berkeley as most municipal districts basically will divide the city's relationship to a property into three buckets.

It will talk about zoning, which is about what commercial activities can you run where. It determines various things like noise levels. It determines some amount of what businesses are allowed to run where—if you want to have an adult entertainment business, you can't put that right next to a school or whatever, and various stuff like that. Also often that's where single family zoning comes from and then suburbanization. It determines what you can do in a property. In some sense it's the most restrictive, but we interface with it very little because it's a hotel, so basically everything you do is very neatly covered.

Then you have building, which is basically the thing that you just mentioned about trying to make it so that the property is structurally safe, follows some basic rules or whatever. I have a huge number of disagreements with those rules. But the thing that I think people vastly underestimate is just a whole additional department—fire. Fire is ultimately the thing that determines the vast majority of constraints on a property.

For example, we have things like the Americans with Disabilities Act, which has a huge effect on a large amount of construction regulation.It actually turns out that almost all the complexity of those regulations is not a result of trying to make your everyday usage accessible. Almost all the complexity is the result of needing to combine the standards for fire safety with the standard for disability accessibility. Because the question is how do I make it so that if my building is on fire and everyone is trying to desperately escape, all my disabled people can still escape in time, which is a much, much harder problem than how do I allow general disability accessibility to the property. So fire is really where all the regulations come from.

The impact of wide roads on American car culture

Patrick McKenzie: There's an interesting and understandable historical reason for this, which is quite a bit of path dependence. The Triangle Shirtwaist fire in New York, for example, kicked off both fire regulation generally in the United States, but also quite a bit of building codes. The influence of them doesn't stop at the shell of the building either. The physical design of American cities is in many cases—the width and other dimensions of roads are dictated by the fire department in a lot of cases.

[Patrick notes: Fire apparatus access roads must have an unobstructed width of not less than 20 feet and vertical clearance of not less than 13 feet 6 inches. When fire hydrants are present, the minimum width increases to 26 feet. (Sources: 2021 International Fire Code, Appendix D; California Fire Code 2019). There is recognition among some urbanists that mandating these across a city cedes the life of the city to fire defense. Fire defense is a worthy goal, but policy is the art of tradeoffs among worthy goals.]

Oliver Habryka: That was one of the big things. I think recently I saw an interesting Twitter discussion which then caused me to make some estimates. Somebody was like, "Yes, the reason why American roads are so extremely wide is a result of lobbying from both local and national fire departments that try to make it so that their giant, extremely big fire truck can fit on any road in the country," which is then crazy.

Fire safety regulations vs. traffic fatalities

I did this estimate where it's plausible to me that 30 to 40% of traffic fatalities in the US are a result of building fire and road fire code regulations. Roughly the history as far as I can tell is: American fire departments lobbied for extremely wide roads. Extremely wide roads enabled America to have by far the biggest cars in the world. It's one of the primary reasons why America has such big cars. When you look into statistics about why American road fatalities are so much higher than almost anywhere else in the world, the size and volume of the car is one of the top primary determinants, next to a few other things around driver education.

It's not crazy to look at that and be like, oh my God, I think it's plausible to me that 30% of traffic fatalities in the US are downstream of building fire code regulations.

Patrick McKenzie: 30% of traffic fatalities is probably 10,000 deaths a year, whereas the total number of deaths to fire in the United States is probably what, a few hundred maybe? [Patrick notes: Traffic deaths in the U.S. is indeed about 40,000 a year. Self-driving cars can’t possibly come fast enough.] 

Oliver Habryka: It's less than that. I think per year you probably end up in the low hundreds. It wouldn't be surprising to me if every year we have less than 150 deaths.

[Patrick notes: Oliver and I both underestimated civilian fire deaths substantially. It’s actually about 3,500-4,000 in most years.]

Patrick McKenzie: Granted, that represents an enormous societal success in that Triangle Shirtwaist by itself would have blown through that number. But the people who are making the decisions on fire trucks—I'm sure they're making the decision partly because they want the safest fire environment possible. I'm sure they're also making it because they love big fire trucks.

Patrick, in the voice of an eternal fire truck enthusiast, which is not not his own voice: I loved them since I was a child. I wanted to be a firefighter when I was a kid. Doesn't everybody?

Patrick continues: But nobody's presented with both sides of the ledger. You're saving X number of lives this way and then costing this way, and I should make a rational cost-benefit calculus between the two.

Microeconomic analysis in policy making

Oliver Habryka: Oh yeah. My colleague—this was in the UK, not in the US, but very similar—there was a hilarious case where I think they were trying to estimate something about regulation around maybe stairways or maximum height of a building in the UK. There's this government report where for once, in a rare case, they actually do a reasonable microeconomic analysis. They check what are the safety implications, what are the economic benefits. You just have this report which outlines pro, case in favor, case against. They do a microeconomic estimate and it's like, "We estimate the benefits of not implementing this regulation exceed the drawbacks by a factor of five in terms of basically all the costs that we have."

Next paragraph: "Therefore we decide to recommend that you do instantiate the regulation." With no connecting word, just like: microeconomic analysis gets the wrong answer, throw away the wrong answer.

Building codes and staircase requirements

Patrick McKenzie: This is one of the examples of the policy pipeline in the United States. Someone in—not a journal article, but literally a blog post, could have been a LessWrong post—did the analysis of whether, for a very long time in very many building codes, including I believe the model building code in the United States, there was a requirement that above a certain size of building in terms of number of units, you need to have two staircases.

This was justified by a folk intuitive belief that if one of the staircases is on fire, then they can escape via the other one. Turns out, staircases don't catch on fire that frequently. We are instead just sacrificing a tremendous amount of resources to have the second staircase, both in terms of—sure it takes construction resources to build it, but also because real estate is fundamentally in the business of selling square feet. We have no one living in the staircase. Sensibly, a staircase is a rough place to live. If we just repurpose those square feet as buildable space in our cities, we would be able to, with no other change to the built environment, have many, many more people living in the cities at basically comparable cost.

I will drop a link in the notes, but I believe that there has been some amount of movement actually because someone actually sat down and did the math.

[Patrick notes: Claude helpfully informs me that Stephen Smith of the Center for Building is the single public intellectual most associated with this issue, which I had imbibed from the grapevine (as a canonical example of the “posting to policy pipeline”) but had not successfully cached a name for. You might find Options for Building More Single-Stair Apartment Buildings in North America a good reference if you’re curious.]

Berkeley's shift in construction policy

Oliver Habryka: Actually as a point, Berkeley has hugely shifted its policy environment. Basically around I think two years ago, the Berkeley mayor that was elected was an extremely YIMBY mayor that basically won on a YIMBY platform. Indeed, we've been seeing a huge amount of construction. The Berkeley regulatory environment for new construction has vastly improved in quality in roughly the last two to three years. Which I think now as a property owner in Berkeley who probably isn't going to upend this property, I now must be staunchly opposed to.

Patrick McKenzie: Yeah, the political culture that is Northern California. When I was living in San Francisco during the pandemic, no less, for complicated reasons, I got a postcard in the mail from the city of San Francisco. "Someone wants to build an apartment in your neighborhood. Would you like to come down to the meeting to object?" Because my real home was in Tokyo at the time, my Tokyo brain just—this fails to compute. Why? Why would I go and check? One, there's a pandemic on right now. You do understand that, right? Superspreader events at the meeting would be very bad. But two, why? What? Why would I care? Why would anyone care? It's a neighborhood, there is more sky available.

But I'm glad that through the tireless effort of the YIMBY folks on both policy advocacy and convincing just the general population that this is important, that the cost of housing is one of the things that Americans should indeed weight as most important on their list of wants over the course of the next 5, 10, 20 years, that there is some positive movement there. And us landowners, we can take it on the chin, that's fine.

Fundraising challenges in the nonprofit sector

Let’s not-so-awkwardly segue into the money that pays for it all. You run a nonprofit, I've previoPostsusly ran a nonprofit. Both of them are large relative to a church bake sale, small relative to the United Way. I would love to talk about the mechanics and surprises of doing fundraising. One of the things that I was told many times is, "Wow, you are so lucky that you are in and adjacent to the tech industry because it must be so easy to get money for a nonprofit in the tech industry."

I will say we raised $1.2 million, fantastically fortunate by the standards of attempting to start from nothing. But oh my goodness, it was a slog for every single dollar. It was much harder than I would've imagined. I spent 50% of my cycles as a nonprofit CEO doing fundraising when starting with approximately the most advantages I could imagine starting with.

So Lightcone raised some money. How'd that go?

Fundraising challenges in the nonprofit sector

Oliver Habryka: Yeah. I think the history of Lightcone Infrastructure has two seasons of its fundraising, I think historically. We were part of a lot of the effective altruism ecosystem. That is a completely different beast to fundraise from than almost the whole rest of the world. I really want to be... I'm now slightly tempted to refer to it almost in the past tense, but maybe I shouldn't be as cynical. I think it was one of the greatest things in terms of a charity ecosystem that I've seen, in the sense that genuinely what would determine whether you would get funding is firm estimates about your impact.

You would be like, "Here's how much you're buying." Of course, for an organization like ours—we are working on public discourse forum infrastructure—it's much harder to be like, "Look, here's how many lives you buy when you buy marginal LessWrong.com moderation hours." But ultimately at the end of the day, I had a sense that things were downstream of how much impact we had, roughly in the sense that we would have a budget, we would try to make an argument that the impact we're doing, the impact that we are producing with our budget, is above the baseline of what a funder like Open Philanthropy or some other similar funder would generally fund. Then we would pretty reliably get our funding.

In general, there was so much goodwill going around that basically the first funder we convinced of this would end up filling roughly our budget. Because they would be like, "Look, I could probably call up the other funders in the ecosystem to get them to pay their fair share, but they are doing the same on their side and I don't want them to have to sum up all the cognitive context to evaluate your project. So basically I'm going to fund you. I'm going to check in with some other people whether they think maybe this is a terrible idea to see whether maybe I want to apply a higher standard and see whether I would fund this all myself. But there's a lot of trading going around, there's a lot of goodwill going around. If it looks good by my lights, we'll probably just fund it fully because it doesn't seem worth it for you to fundraise from three to four other people."

The collapse of FTX and its impact on fundraising

Then I think basically all of that very drastically changed with the collapse of FTX and the broader effective altruism ecosystem. Both had a huge collapse in the surrounding trust of what was going on. Definitely now Open Philanthropy isn't going to go and be like, "Oh, well I just kind of trust the other funders to pay their fair share," because turns out one of the funders that most—they fought that they were paying the fair share to—turned out to have stolen half of their money or approximately all of their money. Now all the organizations that they thought were paying their fair share are now coming back to me to cover the clawbacks that they're facing to give back the money. A lot of that trust evaporated.

Then more broadly, the whole domain became a lot more politicized. Reputational considerations about how easy it is, good luck to fund this organization started becoming very dominant. I think in that case, now you're entering the domain of nonprofit fundraising that is the standard world. Where I think it's really crucial to understand that the thing that practically everyone hopes to buy by giving money to a charity is to look good.

Patrick McKenzie: Mm-hmm.

Oliver Habryka: And often, and this is particularly pernicious, they expect that they want it to look good to the least informed audience. It's not the case that they're like, "Oh, I want it to look good to the people who then will try to write my autobiography and my legacy and try to understand what the consequences of my life were." They want to be like, "Well I'm hoping that if I give a bunch of money to charity that the Internet mobs are less likely to get angry at me or something of that kind."

Patrick McKenzie: Or say high-status journalists or similar.

Oliver Habryka: Or high-status journalists or something like that. You now have this—basically what determines whether you can fundraise from someone, a large fraction of it, is: does it look good to give money to your organization? Does it generally look like the kind of thing that can signal very robustly that it's not going to bite them back later?

Patrick McKenzie: Apologies for making this observation out loud, but LessWrong and associated communities are among the most extreme cases on the graph of realized impact on the world versus lack of institutional trust and heft in the world.

Frankly, this community does not scream "I am the safe choice." [Patrick notes: A term sometimes used in the community is “spending one’s weirdness points” and, as a sympathetic non-member, I would say the community is profligate with the weirdness points, to the detriment of its broader aims.]

The absolutely safe choices are the university system and whoever is most recently name-tagged in the Wall Street Journal or similar having received a hundred million dollar check from the usual suspects.

Oliver Habryka: All the things that look like, even if they turn out to be fraudulent or fake afterwards, nobody's going to hold you accountable for. Nobody is going to be angry at you for giving a marginal $2 million to Make-A-Wish Foundation. Nobody is going to be angry at you for giving money to your local children's hospital. Those are the archetypical acts of "nobody got fired for buying IBM." 

Patrick interjects: “Nobody got fired for buying BLM.” [Patrick notes: The BLM Global Network Foundation benefitted from sharing the name of a social movement which had a certain amount of influence in the halls of power back in 2020/2021, and raised $90 million, in major part from corporate donors (including a large swathe of the tech industry) worried about being targeted if they did not pay the danegeld. Those donations were misused. You should not model corporate leaders as having been fired because that money was misappropriated, because efficient use of the money was never a major goal.]

Oliver Habryka: Similarly, nobody got fired for giving money to a local charity children's hospital. I like children's hospitals. To be clear, I think they're good institutions. I haven't really thought much about whether they're underfunded or overfunded. It's just important to understand that whether children's hospitals are underfunded or overfunded, I think plays zero role in the decision of the vast majority of people on whether to donate to children's hospitals.

Patrick McKenzie: I think one of the intellectual contributions of effective altruism has been that you should at least spend some cycles on the question of: is this problem tractable? Is this likely to be uniquely accelerated by my donation? Or is the rest of the world is going to handle it anyway and so therefore I should pick something that the rest of the world won't have a lock on?

I sometimes think the community gets a bit monomaniacal on that question, but it's almost a useful corrective versus the previous extremely consensus-oriented use of charity where everyone is chasing approximately the same charitable goals. The largest charities in the United States almost drown in money to the extent that—I don't know that they would describe this internally, but I think some cause areas have difficulty metabolizing as much money as is thrown at them. They do not have enough smart coordinators in the space to make enough programs to justify the amount of cash that has been thrown their way. So they're somewhat profligate with respect to things or they choose relatively low ROI things to focus on.

[Patrick notes: A true thing SBF once said which I agree with: coordinators are scarce. I trust frequent readers of my work do not believe I’m minimizing his crimes due to my great love for the cryptocurrency community and huge hodlings. He struck (and strikes) me as a quite intelligent young man, and I think in this he was quite perceptive.] 

Patrick notes: Meanwhile, other cause areas which are sort of important languish because they are not as camera friendly, because they would not be as immediately legible at a party in Brooklyn which a New York Times reporter was at, et cetera, et cetera. Hate to throw the New York Times under the bus, but the New York Times... well, I'll link to their explanation here.

I have many wonderful things to say about the New York Times. It's great that the United States has a paper of record. I don't think they get it some of the time. I think they get it more than The Guardian does. That is a very low bar. I'm subtweeting a piece The Guardian did of a conference that was here, which I attended, which they made out to be a hive of scum and villainy, and then required something like seven corrections to the piece within a short period of time because it was reported incompetently.

Strategies for successful fundraising

Sorry, going back to the fundraising. We had the discontinuous collapse of FTX and that caused something of a fundraising winter in this end of the world. And yet I notice we are still open. So you clearly did a successful fundraising campaign after the fundraising winter. Can you tell me about some of what happened there?

Oliver Habryka: Yeah. I think the key thing that we then did as we shifted—I think from the very impact-based giving model to the central fundraising methods that we ended up utilizing—is we did a big fundraiser on LessWrong and the vast majority of fundraising came from people of the extended rationality community. I kind of think of it a bit as... I'm going to be chasing down the annoying comparisons to religion for the rest of this podcast, but there is a bit of a normal model that a church would have for fundraising. You have a lot of people who are adjacent to the church community. You ask them, do you want to give to it?

Especially given that we are part of the effective altruism community, this has produced very confusing states of how people want to give to us. People are like, "I think you're probably the most cost effective thing, but I don't know how I would justify that to other people in the effective altruism community. So instead, I want to give you from my 'donating to my local community and infrastructure' budget."

That de facto has resulted in us ending up in an okay-ish financial position—one that I'm very glad about in the sense that we shifted from, I think probably in terms of the nonprofit space, one of the most concentrated donor bases maybe of any charity. We basically had one funder, two funders—Open Philanthropy and Survival and Flourishing Fund—that funded us for many years. Then shifting to a funding base where we fundraised approximately $3 million last year with not a single donation above $150,000, with approximately a thousand people who make up the majority of our donor base. We easily skipped two orders of magnitude of concentration.

Lessons from religious fundraising practices

Patrick McKenzie: Have you discussed with someone who has run the fundraising for say a Catholic parish or a synagogue or something similar?

Oliver Habryka: I've never talked to someone who did that specifically. My sense is—I've talked to a few people who are involved in church organizing—but I think they are usually like, "Well, we have this strong rule of tithing and so that's the thing that there's buy-in for. Then mostly we go around and we remind people to please tithe and that seems to work reasonably well." Whereas that definitely doesn't exist for the rationality community and shouldn't.

Patrick McKenzie: A long-held hobby horse: people in tech and tech-adjacent spaces underrate what they can learn from the history of religion and religious practice. Without giving you the entire pitch at the moment, I think you would find if you engaged people like that, that there are some things that religions have figured out in the last 2000 years. For example, not all religions have access to the technology of "we've told you for the last 2000 years you must tithe 10%." There are different anchors, there are different practices.

The natural way communities work is there are people in different seasons in life and different attachment to the community. If you look at the annual capital campaign for say a Catholic parish—I'm using this example because I am Catholic and I'm occasionally tapped for that capital campaign, somewhat complicated by the notion of how a parish has a relationship with the diocese and potentially an archdiocese above it—there might be people who are known to the parish as being relatively generous or relatively well-moneyed, local community leaders for example, who will get a separate appeal made quietly to them.

There is some element of the usual and some element of status and some element of "from those who much is given, much is requested." Then there is the broader community where if someone is donating $10 a month in the collection plate every week, that is a thing that the church is both very happy to get, and the numbers would not work without that broad base.

There's many, many different charities in the world that have different sized donor bases and different concentration of whales versus the rest of the population. But I think it might be interesting talking to someone who has done the capital campaign for raising a new church or something on how much comes from the usual eight people and to what degree you are literally downstream of someone who has a wedding to pay for this year and so they're going to either bow out or not be able to do much versus how much of it is okay, it's that time again, ponying up for the annual thing.

Oliver Habryka: Yeah. I mean, I think one thing that's weird, and I think it's kind of... my guess is this is also present for churches, but was a very central dynamic in LessWrong/Lightcone fundraising. It really felt to me like roughly centrally I was playing game of chicken. When I would go to people and be like, "Look, we don't have enough money to run until the end of the year. By default, we'll declare bankruptcy in nine months," they'll be like, "I mean, that's crazy obviously. It would be crazy for that important pillar of both my local ecosystem and the world to fall through."

Then I would be like, "Great, glad to hear that. You're fulfilling maybe some non-trivial chunk of our remaining funding gap." And they'd be like, "You know what? Take me down as one of your donors of last resort. If it really looks like by the end of the year nobody—by the end of fundraiser nobody—has offered to fill the gap, then maybe I will step in and be like, 'Look man...'"

I have 17 donors on my "donor of last resort" list. Each one of which has been like, "Hey, can I please be the person who at the very last gets tagged to somehow fill in your gap?"

Patrick McKenzie: I think this is a classic collective action problem and everyone wants to be the donor of last resort. No one wants to be the first check-in, which has long suggested to me that we're missing social technology.

In Silicon Valley there's status associated with "I was the first angel check into that company." That's amazing.

Oliver Habryka: That's the thing that I've been dreaming of. I've been desperately wishing for roughly the last two years of fundraising to have lead donors similar to how you have lead—someone who leads a round—to be someone to step up and be like, "I'm going to be donating a hundred thousand dollars to this project. They have a funding gap of a million dollars. I am now going to be the person who sends emails to a bunch of other people and pulls things together."

Because you're in such a shitty position if you are the charity itself trying to pull that together. If I, as a charity or even as a startup—similar situation—just make an email thread with 10 funders CC'd being like, "Hey, we got a million dollars to raise, how you want to split things up?" Unless you're Sam Altman or someone else who tends to be extremely good at various forms of social maneuvering or is extremely oversubscribed in the projects that he runs, everyone will be like, "Look, you're just a random startup. I have no reason to trust you. Of course you're extremely incentivized to tell me whatever I want to hear in this moment in time."

This secondary relationship of there's a person with stake in the organization who has sent a costly signal that they care, but who has relationships to maintain with the other funders who can then pull them together to make something happen, is a very crucial piece of social infrastructure in Silicon Valley. It feels to me like one that could generalize to the charity space, but I have never seen this happen.

Patrick McKenzie: So spoiler alert, the religious communities are nodding their head like, "Yes, we call them fundraising chairs." If there is a capital campaign for 2025, there might be a person in charge of that capital campaign who's not the professional or semi-professional person that is organizing the campaign. That person—in consensual hallucination time—didn't buy their seat as the chair. They didn't buy the shout out from the pastor or similar at the annual fundraising party for being the chair and putting this together, but they didn't not not buy it.

They might be the largest check. They might be one of the larger checks in, but the expectation is both: one, I'm going to represent myself to other high trust peers in my community as I have skin in the game here. I'm not going to control you, I'm not going to guilt you. We've learned that cajoling and guilting doesn't produce the numbers every year. But it's that time again, I'm the person who is empowered and encouraged to take everyone out to coffee. I'm going to go down my list and—Bob had a great year in the trading business right now, I see the Wall Street Journal—so you were down for $5K last year, you're down for five again, et cetera, et cetera.

You rotate that position around the community, both so someone doesn't have the weight of the thing smack them every year, but also so that it isn't the same person spending their personal capital every year. A thing that you could potentially consider is—yeah, next year—who wants to be the master of ceremonies at a Lighthaven event slash who wants to spend a few weeks lighting up their network and saying, "Hey, are you good for help one of the bright lights on the Internet keep the lights on for another year?"

Fundraising experiences and challenges

Oliver Habryka: That's really interesting. Do you have a sense of how... it sounds in the thing that you just described that there's in some sense already an existing tradition that makes the negotiation here possible. But the thing that feels tricky to me when I'm imagining this is: it feels kind of hard to do in the sense that I'm asking someone to do a thing that's at a very awkward intersection of being something they do because I asked them to, or something I encourage them to do, and something that they're doing on behalf of the community or the ecosystem.

It seems in some sense that if I choose all the bits about who this is going to be, ultimately it's just another employee of mine doing more fundraising. So there needs to be a certain amount of independence. There needs to be a certain amount of community buy-in. Having it as a thing that a community can take as an object and as a role feels very important. So I'm uncertain how it would get started as a role.

Patrick McKenzie: Some of the social technology here is the person who is the fundraising chair is not necessarily chosen for that position because they are the best at doing fundraising. The formal or informal person who is the professional or semi-professional fundraiser is often giving them essentially a set of scripts to work with, both scripts in terms of "here are words that you can say that will resonate with people and get money out of them," but also the technology of being a fundraising chair. "We're going to make this easy for you. We'll set up some events, we will give you some suggestions as to people getting to those events and access to our mailing list, et cetera, et cetera. We'll often be writing those emails and you'll be bringing some of the things that you do so well—your personal network, your ability to schmooze people over coffees, et cetera, et cetera."

In year one, it's probably a bit of an experiment and it gets codified around the things that work. Who knows, maybe 10 years from now there will be a wall somewhere that has 10 photos of 10 years of fundraising chairs on it.

Oliver Habryka: That's really cool.

Patrick McKenzie: Yeah, thank you. Anytime. 

You too, Internet, can get a free consultation by coming on an episode of Complex Systems if you have something interesting to talk about, like a real estate project. 

A huge amount of my intellectual output in my career has been in having so very few original ideas. There's just some ideas that are appreciated in certain communities of practice that do not cross the osmotic barrier by themselves. And so I just pick them up from A, deliver them to B. Gains from trade, it's amazing.

Oliver Habryka: Yeah. I mean, immediately after this I will Google a bit and see: can I find any guides for how to make fundraising chairs, and how can I then translate that into, of course, the very, very different context in which I'm in, of a weird gaggle of rationalists.

Patrick McKenzie: Well, I wish you the best of luck and skill with that. When your fundraising chair wants to hit me up next year, they know where to find me on the Internet.

Oliver Habryka: I'm kind of curious how the thing that I described compares to your experiences. You just said you fundraised $1.2 million, roughly.

Patrick McKenzie: Oh yeah. I described this a bit in the Story of VaccinateCA piece I wrote for Works in Progress, but we were not in a position where we could just find one donor. We had a collection of donors from Silicon Valley. Some people that were donating in their own capacity, others where an organization was writing the check. It was very weighted towards large checks versus a large number of small checks.

I'm constrained a little bit about what I can say, but I will say that things like factors that felt extraneous to me did dominate the decision making of some funders, including some very well-moneyed funders. We were an emergency response effort, and I was very clear with everyone, "This is an emergency response effort. We will paddle like hell for as many months as is required—a year or two years if it is required. As soon as we are not required, we spin this down and go back to our day jobs." That led the start of every conversation I had with someone.

There was a large charitable funder, and I will not name them for obvious reasons. This large charitable funder put us through six weeks of a process that they had for qualifying potential opportunities. During those six weeks, they were extremely bullish on the opportunity for funding us. They mentioned that the target $2 million that I wanted to raise—the $2 million was not a lot of money for them. I was like, "Awesome. I will spend a lot of my time for six weeks jumping through hoops." I perceived myself as having a green light at some point in that six weeks. Possibly that was me misreading the situation. Possibly it was not.

The final meeting with that funder was, "Well, Patrick, we really love how much time you've spent with us, but we can't fund this because there is no ongoing institutional value created by VaccinateCA." You've known since the jump six weeks ago this is a crisis response. That is the only thing to fund here. They're like, "Well, we see as an organization that a priority is building institutions rather than just jumping between fires." I'm like, "Cool. Would've loved to know that six weeks ago because I had other things to be doing with my time." Didn't say it in as many words, but definitely how I felt.

On the flip side, there were other people who had done well in Silicon Valley. I feel weird about giving people shoutouts for charitable behavior. They're welcome to claim any of these anecdotes if they want to. But there was a particular person where he liked a tweet of mine on Twitter about the project. I sent him an email five minutes later and said, "Hey Bob, saw you like the tweet. We are fundraising for this right now, and anything would be helpful." Got an email reply back an hour later saying, "Meet A, B, and C on the team. Send in the white paper. I will get you a decision soonest."

That was at 4:30 PM California time. 9 AM the next morning: "I've decided to donate a hundred thousand dollars out of my personal account. You will shortly receive an introduction to my banker." Wire arrived at 10 AM. Love that bit of Silicon Valley—you can just do things, and when you decide to do things, do things fast.

But that was not a universal experience. Indeed we didn't catastrophically fail through inability to raise enough money, but we did have to mothball the project about a month, maybe six weeks before I would've liked to counterfactually if I had hit the fundraising goals.

The fundraising environment also changed under our feet. You mentioned how in your experience there was the FTX winter. In 2021 in February, when most people were individually not vaccinated themselves, it was like, "This sounds like the most important thing ever. Let's have conversations, et cetera, et cetera." May rolls around, I'm still fundraising and fundraising more than I was back in February, because what I perceived as an "in the bag" check is no longer in the bag nor a check.

In May everyone said, "Well, the supply issue for the vaccine is basically solved." They didn't say in as many words, but I think the math was, "I have my vaccine, my coworkers have their vaccine, my parents have their vaccine. Who doesn't have a vaccine anymore?" I'm like, "Well, going by the stats, that answer is half of the country." Perhaps more importantly, a very large number of people who are structurally right now queued up behind all of the Americans.

[Patrick notes: See the interview with Clara Collier for my succinct take on why so-called vaccine hesitancy was overrated as a cause of this.]

The faster we get the United States to the point where it feels politically secure that the vaccine is sufficient within the United States of America—which was not the case in May of 2021—the faster we release supply from American factories to the rest of the world, which will save lives externally and also decrease the amount of processing power that the virus has to run in people in the rest of the world and potentially come up with variants that would be bad for them and then eventually bad for us.

A chaotic time. I don't think I executed as well as I could have possibly executed. That's one of my major reflections on my performance as the CEO in that organization. But it is difficult for people who aren't the CEO to do charitable fundraising.

Oliver Habryka: Indeed.

Patrick McKenzie: Especially for one at the scale of that operation we ran. But lessons learned for next time. Hoping there is no next time for that, but hoping there should be next times for charity. I hope to not have an exact rerun of the global respiratory pandemic urgently needing money to fund something which honestly should have been done by the government. But that is an entirely different argument.

Oliver Habryka: Yeah. It's surprisingly related to our earlier conversation about underfunded infrastructure in a sense. That really was a lot of online, straightforward web development, online infrastructure and coordination of people that obviously—if any government bureaucrat was capable of doing the cost effectiveness estimate compared to anything—their internal spending decisions were being made.

Patrick McKenzie: We were spending tens of billions of dollars on various aspects of the pandemic response. Of course there's a trillion dollars up in the air with the economic impact. I was scraping for... I think the smallest check we got was $25,000. I was on phone calls for—get a $25K here, $25K there. It eventually adds up to real money.

Government being its own ball of wax, I had that same conversation with places in Silicon Valley that can lose $25,000 in the couch cushions and couldn't get them over the line. In some cases, we couldn't even get direct beneficiaries of our work over the line. I'm like, "Hey, you have a workforce, they would prefer not to die this year. You have a budget associated with them not dying this year. $100,000 is a very small amount of that budget. Put out a Slack message on how many of them found the vaccine through us, and then do the obvious math."

That was the least effective sales pitch ever. In some cases it was like, "Why buy the cow when the milk is already free?" But tech did not cover itself in glory. And yet, I will say, all of our funding was through the extended tech community. So we as a community made some good decisions.

But I don't want to end on a downer note like that. I think that one of the things that LessWrong demonstrates, that the larger nonprofit demonstrates, and that VaccinateCA demonstrates is that a small number of people can accomplish good in the world just by typing things into their laptops with enough frenzy until the world changes in the ways they wanted to.

I think I have some differences of opinion with a lot of people who are on LessWrong about specifics of the forecast for the world over the next couple of years. But I am immensely glad that it exists on the internet. By convention I end these with asking someone, where can we find you on the internet? But I think the answer is we've talked about that pretty thoroughly today.

Oliver Habryka: Yeah. LessWrong.com is the obvious place, and then if somebody wants to know more about the umbrella organization, LightconeInfrastructure.com.

Patrick McKenzie: Well, thanks very much for being on the program today, Oliver.

Oliver Habryka: Thanks very much.

Patrick McKenzie: And for the rest of you, we'll see you next week on Complex Systems.