4 ms·

I prefer watching this world chess championship between the two strongest engines in the world: http://tcec.chessdom.com/live.php http://tcec.chessdom.com/live

by tromp 10y ago

I prefer watching this world chess championship between the two strongest engines in the world: http://tcec.chessdom.com/live.php http://tcec.chessdom.com/live.php Stockfish 8 squaring off against Houdini 5, 100 games back to back, no rest days. Go, Stockfish!

Kenji 10y ago
Wow, that is a very pretty website. I really like that Windows 95 style grey window.
schoen 10y ago
That's really cool. It's amazing to see how much progress and interest there is in this area. A couple of questions: Are these Elo ratings really comparable to human Elo ratings? So it's pretty clear that these engines would all beat any human player in history? (Edit: I know computers have been able to beat human champions for what xkcd points out is about half of my lifetime, but Deep Blue had a lot of resources behind it that I would imagine these engines don't.) How do they deal with differences over time in the hardware that the engines run on? (both in terms of Elo calculation consistency and in terms of historical comparisons between engines' performance ... how do we know how much improvement is due to programming improvements vs. faster hardware? do we know how earlier generation engines would fare today running on the same hardware that these are running on?) Do the engines in a particular match run on the same or different hardware? (I thought I saw people in the chat talking about this but I didn't exactly understand what they said.)
- schoen 10y ago
  Also, when the tournament interface says Stockfish is working at a depth of 41 plies, does that mean it's performed an exhaustive search to that depth (!?!!) or just that it's examined some line of play to that depth?
  - danielvf 10y ago
    Only the good lines of play. The secret to modern chess engines us that they are able to not evaluate most most moves.
- hyperpape 10y ago
  Yes, the ELO difference is pretty real: https://www.chess.com/news/komodo-beats-nakamura-in-final-battle-1331 https://www.chess.com/news/komodo-beats-nakamura-in-final-ba...

by tromp 10y ago

Kenji 10y ago

Wow, that is a very pretty website. I really like that Windows 95 style grey window.

schoen 10y ago

That's really cool. It's amazing to see how much progress and interest there is in this area. A couple of questions: Are these Elo ratings really comparable to human Elo ratings? So it's pretty clear that these engines would all beat any human player in history? (Edit: I know computers have been able to beat human champions for what xkcd points out is about half of my lifetime, but Deep Blue had a lot of resources behind it that I would imagine these engines don't.) How do they deal with differences over time in the hardware that the engines run on? (both in terms of Elo calculation consistency and in terms of historical comparisons between engines' performance ... how do we know how much improvement is due to programming improvements vs. faster hardware? do we know how earlier generation engines would fare today running on the same hardware that these are running on?) Do the engines in a particular match run on the same or different hardware? (I thought I saw people in the chat talking about this but I didn't exactly understand what they said.)

schoen 10y ago
Also, when the tournament interface says Stockfish is working at a depth of 41 plies, does that mean it's performed an exhaustive search to that depth (!?!!) or just that it's examined some line of play to that depth?
- danielvf 10y ago
  Only the good lines of play. The secret to modern chess engines us that they are able to not evaluate most most moves.
hyperpape 10y ago
Yes, the ELO difference is pretty real: https://www.chess.com/news/komodo-beats-nakamura-in-final-battle-1331 https://www.chess.com/news/komodo-beats-nakamura-in-final-ba...