it's crazy to me that neural networks learn ~arbitrarily well regardless of initialization. you can even embed patterns in the weights and learning works fine you could encode an image of your face into the layers of a language model and no one would ever know

I wanted to share something I built over the last few weeks: isometric.nyc is a massive isometric pixel art map of NYC, built with nano banana and coding agents. I didn't write a single line of code.

california love

A GREAT PREDICTION MARKET* *if anyone were trading it. In today's newsletter. The potentially very useful snowfall contracts have virtually no volume. While nearly $1 million is on the line for the Louisiana-Monroe Warhawks vs the Appalachian State Mountaineers basketball game.

San Francisco is the warmest city in the lower 48 this morning

Kyle Chan Arnaud Bertrand Without a paywall: archive.is/OZwF2

In any other time or presidency, this story. reported by The Wall Street Journal, would be an earthquake of a scandal. The size, scope and implications of it are unprecedented and mind-boggling. But will this Congress, which brayed endlessly about the "Biden crime family," probe it? Not a chance.

Had an MBA student email this morning to say she wouldn’t make it to class bc the subway train was stalled.. then sent a follow up email giving her specific train car number

In 2026 there are two corporate comms styles.

Let me reframe this entire story: A Chinese company invests in a factory in the US that was previously a shuttered GM plant, employs 3,000 workers in the US with only 1% on visas, and outcompetes a nearby plant that is owned by Vitro, a Mexican company. wsj.com/business/tarif…

Eagles made on hole 1 at Pebble Beach in the AT&T Pebble Beach Pro-Am 2005 through last night: 2 Last 2 groups: 2

This is my first time watching a non-negligible amount of the Winter Olympics since, I think, 2010, when the star competitors were Yuna Kim and Apolo Anton Ohno

Josh hits the 🎯 here. Reminds me of this quote from Arthur Schopenhauer, Counsels and Maxims, 1851:

This is not workable, one could lower the percentage by adding large numbers of probability near-zero markets (like the second coming) and inflate volume by wash trading 🧵

With Russian, Ukrainian, and Iranian airspace closed, there's just a tiny needle eye left for international flights from Europe to Asia. Some airlines are profiting from the current situation more than others. Source: zeit.de/wirtschaft/202…

Big missed opportunity for stats folks to call this a “marginal sensitivity model” rather than “bounded confoundedness”:

🚨Replication alert🚨 I'm pleased to announce that my replication of Moretti (2021) is now accepted as a comment at AER. I find ten issues in the paper. My comment focuses on two major problems; in the appendix, I document eight (relatively) minor problems. 1/

Gradually unfollowing and usually blocking people who post AI writing passing it off as their own. Several hundred blocked so far Writing has historically been useful as proof-of-thought. AI writing tends to be proof-of-performing-thought. These are not the same. Indeed, the