I am definitely on board for "the 2020s could be a technological sea change decade". Really important not to kill the golden goose here - frankly with better incentives for inventors we could probably be doubling R&D in the areas while keeping expected social return high.

Great ex of why theoretically a very good reason to expect AI productivity to most complementary for mid-level skills. Paper coming soon but basically "need to know what you don't know enough to clean things up/guide the AI". Ethan Mollick's framing of this as "an assistant" is good!

Wildly unknown fact internationally: Canada had essentially zero pc GDP in past ten years. Big jump in the capita and homeowners selling to the new arrivals made out like bandits. Everyone else just wound up with no real income gains and if you spend abroad, a 30% drop in the CAD

What every voter should know about housing policy 1. Rents reflect the balance of supply of apartments and demand for those apartments in a given area. That’s it; there’s no magic. If you want lower rents, you can hope for a recession that destroys jobs and, therefore, demand.

Paul's short note here on Brian Chesky has interesting axiom beneath it: most people provide little value to orgs, hard to know who, and team incentives (Holmstrom) means hard to self select. Well performing orgs: principal agent prob w/ costly info flow. Pure delegation not optimal!

The linked paper seems to misunderstand what a model is. Sure, there are only presidential elections every 4 years. Every Super Bowl with that exact set of players is unique. This does not make forecasting impossible, it means you need a model and assumptions.

The basic economics of AI - Ilya and friends raised a billion for a company started a couple months ago, in a very tough fundraising environment. Revealed preference in the Bay about AI (and foundation models may be worst value capture segment of the market!)

No academics in my family; I thought all univ teaching was like this, or cognac at Faculty Clubs! Still a great job, but my classes ain't like this one! (That said, I do teach in same building as the math class from Good Will Hunting- let's see the kids throw down the gauntlet!)

On why surveys about how people feel about economic policy are not terribly insightful...big gap between long run equilibrium consequences and "vibes"

RCT introduction of a pollution market in Surat, Gujarat. Pollution fell by 20–30%, and efficiently — gains came from the firms who could cut pollution most effectively. Mortality benefits are *25 times* costs. An incredible state/research collaboration. 1/2

We Arnold Ventures were frustrated that the US publishes data on the # of hogs slaughtered each day with a one day lag but national crime stats only annually, 9 months later. Since the FBI can't/won't publish more timely data, we funded the creation of realtimecrimeindex.com

Maciej Cegłowski called super-intelligent AI “the idea that eats smart people” and I think it is a large part of why discussions of AI devolve, sucked into the gravity well of existential risk & reward. But there are lots of more immediate risks & rewards that need urgent focus.

I imagine you folks know, but if you care about business/industrial history or factors underlying progress, Works In Progress (worksinprogress.co) & Construction Physics (construction-physics.com) are must reads. Wish WiP had a print version I could put on my coffee table!

Turns out GenAI actually can solve this problem (and so could literally any economist, or home builder...some problems are intractable but California housing is not one of them...)