Thread

Amazing, are you telling me methods that had been around for decades and were state-of-the-art once still work?! Mind. Blown.

@tunguz On a less sarcastic note, I've interviewed hundreds of scientists since 2015 and could really see a shift from SVMs to xgboost to deep learning over the years for essentially the same toy problem I used. It pays of to know what to use when, even in the days of "AGI."