Open Thread 236
1: Lots of people have asked for my thoughts on CFTC shutting down PredictIt. I’ll write something on this next Monday, but for now I tentatively agree the conclusions of this Karlstack article. Short version: the guy on Twitter claiming responsibility is probably a troll; more likely this is Kalshi/Sequoia Capital weaponizing the CFTC to shut down competitors and get a monopoly. PredictIt’s parent group has a link to contact your member of Congress and other relevant officials here, but the website doesn’t work very well and I doubt the political campaign will work much better.
2: Comments of the week: John Schilling tries to calculate the cost of building Neom, and comes up with an optimistic estimate of $4 trillion (compared to its $0.5-1 trillion budget). But he uses the cost of WTC1 as a base; this is unfair because it’s one of the costliest skyscrapers ever. If we use the similarly-sized Princess Tower in Dubai, one of the cheapest, we get a base cost of $2 trillion before applying economies of scale, and ??? afterwards (potentially much less because they can mass produce, or potentially much more because they exhaust the global concrete market). But the $1 trillion Neom budget also has to cover the Floating Octagon Of Clean Industry, the giant ski resort, etc, so it’s still a long shot. Also, Reader talks about his experience working in one of the companies that designed Neom. And Honourary notes that the Saudis are big investors in hyperloops, and a working hyperloop could make good on Neom’s otherwise-hyperbolic transportation claims. Maybe one way to think about Neom is as the first city designed from the ground up around hyperloops, in the same way that Levittown was the first city designed from the ground up around cars. This makes me update very slightly towards plausibility - but it’s still based around combining ~5 things that have never been tried before, and hoping none of them goes wrong or exceeds projected costs.
3: Michael Trazzi has a YouTube channel where he interviews people about AI (including alignment). Pretty high context, but if you know the basics it’s a good way to keep abreast of the latest developments. I especially recommend the episode with Connor Leahy.