Paul GrahamVerified account

Parlaying bets

Error correction also shifts foundation of later builds creating stranded assets and changing direction and pace of future development

Another underrated issue is that getting the right result is not a prerequisite for having made the right prediction.

Reminds of integrating acceleration from an IMU to get position estimate. It's hard.

Future developments will require new knowledge, so predicting them requires most of that knowledge, too. If we had it we'd almost be there.

The reason it’s hard to predict the future is that our physical and social environments are poorly designed.

policy beats planning

Proof: tape-measure & cut a length of wood (1). Use (1) to estimate 2 & cut. Use (2) to estimate (3) & cut. Rpt to (10). Tape-measure 10. ?