"was Deep Blue necessary to beat Kasparov? Or could we have gotten away with an off-the-shelf 1996 desktop running a smarter chess engine? "
Tl:Dr yes, and it's not even close
https://nicholas.carlini.com/writing/2026/algorithmic-improvements-chess.html
@tomstafford great project and article, thanks for posting. As the author concludes at the end, there's a potential lesson for AI here. Energy efficient AI that can be trained with commodity hardware to performance levels comparable with current state of the art might not be far away. What happens to all those trillions of dollars of investment then?