Bureaucrat. Penguin enthusiast.
JohnofCharleston
Karma: 953
AI 2040 Discussion Group
probably less than a dozen economists who are taking the transformative AI seriously at all
I think you’re pointing at something meaningful and interesting, but you’re skipping an inferential step that I think leads this analysis off-course. I agree that not many economists are publishing technical papers or pre-prints taking Transformative AI seriously, but it’s not because the field ignores the possibility. The Dallas Federal Reserve published on this last year:
Notable counterexample from the Dallas Fed, note the date is June 2025.
https://www.dallasfed.org/research/economics/2025/0624The trouble is that… there’s just not much to say, economically, in each of these three scenarios. Economics at its core is about allocating scarce resources. If everyone’s dead (purple line), nothing is scarce. If we don’t get Transformative AI (green line), the world economy grows slightly faster but otherwise things stay as today. If aligned cognitive labor is no longer scarce (red line), our world looks very different than today, in ways that are genuinely quite hard to reason about.
Some can and do publish on what a post-singularity world would look like, but that work can’t be very technical. There’s no data to work with, the models are wild guesses, it’s all informed speculation at best.
I think this piece would have gotten to a more useful place if you described your observation more directly. “I cannot find much published technical work by economists on Transformative AI. What does that tell us?,” rather than skipping to the assumption that it’s because they don’t take the premise seriously.
Note, fixed end date being five months before the start date. Oops.
This argument from @Elizabeth and @Ben Pace, the Vacationing Vagabond fully convinced me. We have started referring to the project internally as “Lighthaven DC”, in part to clarify our intentions, in part to leave space for another version that isn’t policy-focused.
I’m considering editing the post’s title and body to replace each mention of “Lighthaven East” with “Lighthaven DC.” What are the norms about that? Encouraged, discouraged, pedantic nuanced third choice?
Edit: Post is now changed to reflect this, with a note explaining why. Thanks for the feedback!
Tyler Cowen is a fan of 80,000 Hours: The Book, calling it ‘the one book that really matters.’ I have not had the hours (ha!) to read the book, but I do worry the thinking is growing rapidly obsolete.
Note this is actually Alex Tabarrok, not Tyler Cowen. Classic MR mistake.
I’ve heard “what about a rural retreat location, the land is much cheaper” pretty often in response to this pitch. While I think that’s a good idea, it’s not this idea, which specifically is focused on gaining policy influence in DC. Lighthaven DC needs to be within an easy commute distance of policymakers, that is ultimately the point.
I do think the retreat center thing is also worth doing. I wanted to note this not-so-dissimilar scheme to buy Hampshire College, by people at least adjacent to the community. Substack post here, there was some discussion of it at LessOnline:
The Real Hampshire College Has Never Been Tried
Zillow Alert: $700M+ College, available for just $175Mcc: @sparr, though I’m sure you must have heard.
Commenting to note that I reached out via email and haven’t heard back.
I’m not sure what to do with generic expressions of interest like this. Any thoughts or recommendations on how to capture and build on people’s enthusiasm if they haven’t engaged with the local community before?
I wasn’t aware of that site, thanks for sharing. The Henderson buildings look fantastic, though zoning may be a problem. Definitely on our radar now.
Edit: Sadly under contract, as of the end of June.
I pulled on this thread last week when talking to @habryka about his design decisions. As you might have guessed, this wasn’t an accident, was a very deliberate design choice.
I’m DC-pilled enough to see the value in tables. I too enjoy ridiculously over-engineered board games that take all day. One of my formative moments as a bureaucrat was printing out 11x17 sized slices of a complicated econometric model I’d made to analyze a question, then tracing various assumptions through the pages spread out around a big conference table. They’re also quite helpful for eating, especially when you have food that you need to cut.
Lighthaven implicitly asks you to “just use several whiteboards,” or “plan for catering that doesn’t need knives.” That can be annoying… but I see Oliver’s point. Part of what makes Lighthaven work is optimizing for conversations in the 3-6 person range. Part of that is lowering switching costs, encouraging people to leave conversations that aren’t a good fit or have grown too large. In his example, putting your food down at a table ties you there for a while, you pay a large social cost to get up from a table and take your food elsewhere, in a way that you don’t if you’re holding or regularly picking up your plate anyway.
My current synthesis, weakly held:
I think tables are great in break-out rooms, for planned workshops and where conversations that spark “wait let’s sit down and talk about this more” can go.
I think Lighthaven is right to minimize tables in the main social spaces.
Curious for more table thoughts!
I think this is onto something, EA and Objectivism share a lot of qualities, and this describes it well. But I’d be careful introducing and using so many acronyms and terms of art. I get that’s a staple of certain branches of philosophy, the practices has its uses. But it can hide subtle problems and differences that the chain of formal definitions flattened, then paved over with symbols. Using full words, rather than reducing things to acronyms, can keep your analysis grounded and help spot where it might not fit.
Summer Camp added many additional hats, including cardinal hats, a pope hat, crowns, and a turban. All were excellent.
Dean Ball—Leviathan Waking: On Anthropic/USG, and a new era in AI governance
The Takeoff—A Conference in Washington, DC
Good option but tends to work better indoors than outdoors. Also requires having glassware and cutlery, which meetups often do not. But I agree this is better in contexts where it’s doable.
I suggested organizers should occasionally use this tool they already have for the purpose of resetting the noise level at events they are responsible for.
A low monotone. You can imagine it as an “om” hum if that helps. The idea should be everyone joining in to the same hummed note.
This is a good and interesting point. I’m familiar with think tanks as institutions, but not as physical spaces. I will say, the output of most think tanks isn’t something we want to emulate, it seems fine and on topic but not particularly intellectually lively, with the notable exception of new entrants like FAI and IFP. And it just so happens that those are, at least, influenced by this community. A significant portion of each have accounts here, not something that’s true of Brookings.
So I believe think tank interior design could be interesting, but I suspect it wouldn’t have much to teach a project like this.
I already have a job that I like.
Yeah, I agree with and was tracking most of what you say. In drafts, I got a comment that laid out a similar point:
[...] The options I see:
1) Lighthaven East wears the rationalist/EA banner loud and proud and leave it to adjacent organizations to decide how they feel about the association risks.
2) Lighthaven East works carefully to brand themselves as something more generalized—a neutral tech-policy event space, for instance, to aid in broader coalition building[...] I think that many politicals especially on the right would prefer option 2.
I replied: “you’re right that I need to lay out both paths, rather than just the one I prefer.” I started writing up option 2… and made almost no progress on the draft for a week. This past weekend I realized option 2 just didn’t move me, so I abandonedthat alternative. There may well be a “Constellation East” in the works. Someone could also just scale up NET to a similar effect. I’m sure either would be feasible and great community resources. They’re just not for me, not what I was analyzing here.
I disagree on a key point, though. I wrote this post to say that Option 1, the full fledged “doing something with ambition on the scale of Lighthaven, with all the weirdness that entails,” seems feasible to me. People are welcome to disagree, several friends in DC do, but I see this city as much more able to consider weirdness than the cynical rationalist caricature would have you believe. They may not ultimately agree, they may not think they can build a coalition for something weird yet. But they don’t turn off their brains to even considering the idea as often as our community thinks.
Last year, I wrote:
Many people misunderstand the problem with pushing for policy that’s outside the Overton Window. It would be difficult to find a policymaker in DC who isn’t happy to share a heresy or two with you, a person they’ve just met. The taboo policy preference isn’t the problem; it’s the implication that you don’t understand their constraints.
I still believe that’s true. In fact we’ve passed that point, because their constraints have started to change. Six months after writing that, I watched a Senator open for @So8res at a press event. Last month I saw another Senator introduce @David Scott Krueger and @Max Tegmark. This town is openly discussing some of our weird ideas already. It’s more ready than you think.
“Spero”
I’ve been mulling over “Spero”, from Dum Spiro Spero; Latin for “While I breathe, I hope,” the motto of South Carolina. For our purposes, think Spero House, Spero Hall, or even Sperohaven (though that mixes linguistic roots).
“Dum spiro spero” in a stained glass window at Beverly Unitarian Church in Chicago, from wikimedia.
This has several features I like:
It’s distinctive; rare but not unique.
It’s meaningful; earnest and self-serious.
Uses Latin, a subtle callback to intellectual traditions and a bit of outreach to the modern right.
Everyone likes Cicero.
Easy to pronounce, easy to spell (unlike Schelling or Feynman).
Very short, only five letters and two syllables. I want to avoid acronyms wherever possible. In practice people would shorten the name in common speech down to just “Spero” rather than the awkward acronym “SH”.
Helps combat the “Doomer” narrative. The headlines of even potentially negative hit-pieces wouldn’t necessarily be bad: “Fancy ‘Doomer’ Clubhouse in DC is Called ‘Hope’?”
Personally resonant for me.
Downsides:
It’s earnest and self-serious.
We’ll have to explain it repeatedly in speech, again and again, until the end of time. “No, not the bird, S-P-E-R-O. It’s Latin for hope.”
This really only fits if we go for a building that’s old and in a traditional architectural style. It wouldn’t work for something modern, brutalist, or corporate. Art deco or other old futurist styles would be an interesting contrast, would depend on the site.
This came idea came to me a few days ago when I was mulling over the tone from “Nothing is Mere.” I ran through the Feynman quote looking for bits that worked a place-name, nothing quite fit. But it got me thinking about more abstract directions.