testing beliefs against reality

Remote work buys time for housing reform.

forecasting bots will transform the field

Yes, great comment. Getting adoption will be a challenge (although the concept might appeal to fledgling applications).

it’s not a technical limitation, it’s a business model

I found the argument of the counterfactual to be persuasive. Do you have a sense of how much of the time would people actually be doing something better with the extra bandwidth?

I think it's a good argument to say that we don't know for sure. But people who are addicted to youtube today would probably have done something more meaningful in the 1990s, 2000s, 2010s...?

I'd be curious about forecaster estimates of monetary damages from the drone strikes.

I did briefly look into how much the estimated $7B was as a proportion of GDP: 7B/2T = 0.35% of GDP, which feels like a lot. For 41 planes it'd be $170M per plane, which seems reasonable when compared to US bombers, but unclear for Russian ones, but my guess is it's not too far off. This source (https://www.france24.com/en/europe/20250601-ukraine-says-it-...) says $2B for the planes alone, but then you also had the airbases &c, and maybe a submarine base <https://newsukraine.rbc.ua/news/major-explosion-hits-russian...>. I'd still expect it to be a bit exaggerated for propaganda effects, but it does seem reasonable all in all.

Possible, but seems high to me. I'm not sure if those planes were really destroyed or just damaged. My knee-jerk estimate would be somewhere in the hundreds of millions. Still a very successful operation by Ukraine though.

The explosions seem to have blown one wing off of the planes. That's not destroyed - the main fuselage is there, and the other wing is attached - but it's in need of a major rebuild. I suspect that it's a factory-only level of rebuild, but I don't actually know.

It's at least months to years before they're operation again, or years to decades until the factory can make that many new ones.

> not destroyed - the main fuselage is there, and the other wing is attached - but it's in need of a major rebuild

> months to years before they're operation again

Good point!

Good point. For me the issue is the variance which governs the time that I have to leave. For the subway, if I give myself a 10 minute margin beyond the median time, I'm quite likely to be on time, for walking or biking if I give myself a 5 minute margin, I'll almost certainly be on time. For driving on the other hand... as per the article, a trip with a median travel time of 17 minutes can take me over an hour when traffic backs up.

Thanks for the feedback! I'll try to add something about automated cameras and tag violations (which I didn't address) in my next deep dive.

The idea was not that San Francisco does everything perfectly, but instead that there are things they've tried that have been demonstrated to be successful we can learn from in NYC.