jts Posted November 12, 2015 Author Share Posted November 12, 2015 This is game 22 of TCEC season 8, superfinal.http://tcec.chessdom.com/archive.php?se=8&sf&ga=22On white's move 62, Stockfish's eval is 26.13. An advantage of 26 pawns.Stockfish is calculating 28 plies deep, with a search tree of 400 million nodes, and 900 thousand table base hits.Komodo is calculating 34 plies deep, with a search tree of 3000 million nodes, and 50 million table base hits.Stockfish has 10 minutes on his clock; Komodo has 30 minutes.On white's move 64, just 2 moves later, Stockfish's eval is 0.00, dead draw.Stockfish is calculating 39 plies deep, with 2800 million nodes, and almost 50 million table base hits.Komodo is calculating 32 plies deep, with 4000 million nodes, and 64 million table base hits.Stockfish has 7:37 on his clock; Komodo has 28:58.The time control is 180' + 30". This means each side starts with 3 hours for the whole game plus 30 seconds added after each move.Stockfish's eval dropped from 26.13 to 0.00 in just 2 moves.How the 773H did this happen? I doubt that anything like this happened before in the entire history of TCEC. Link to comment Share on other sites More sharing options...
jts Posted August 30, 2016 Author Share Posted August 30, 2016 They are doing stage 3 of season 9 of TCEC and I just saw something so stupid that you have to marvel at the stupidity. No human who ever lived can play chess at the level of the top chess engines but sometimes they play really strange moves. Houdini had rook and 3 pawns against Gull with rook and no pawns. Houdini sacrificed 2 pawns for nothing. Just simply threw away 2 pawns. No compensation. Why? Because it reached a table base win of rook and pawn against rook. Sometimes a table base makes chess engines play stupid moves. Houdini won the game by table base win. Link to comment Share on other sites More sharing options...
jts Posted December 7, 2017 Author Share Posted December 7, 2017 TCEC season 10 superfinal finished just now with Houdini beating Komodo. The champion from season 9 was Stockfish but it was only 3rd place in stage 2 of season 10 and didn't make it to the superfinal. But the latest greatest supersensation that is rocking the world of computer chess is AlphaZero. This is a modification of AlphaGo for chess. It beat Stockfish in a 100 game match with a score of 28 wins, 72 draws, 0 losses. You don't believe me? Click on this link on chessdom.com and read for yourself. Google's AlphaZero Destroys Stockfish In 100-Game Match Link to comment Share on other sites More sharing options...
BaalChatzaf Posted December 7, 2017 Share Posted December 7, 2017 1 hour ago, jts said: TCEC season 10 superfinal finished just now with Houdini beating Komodo. The champion from season 9 was Stockfish but it was only 3rd place in stage 2 of season 10 and didn't make it to the superfinal. But the latest greatest supersensation that is rocking the world of computer chess is AlphaZero. This is a modification of AlphaGo for chess. It beat Stockfish in a 100 game match with a score of 28 wins, 72 draws, 0 losses. You don't believe me? Click on this link on chessdom.com and read for yourself. Google's AlphaZero Destroys Stockfish In 100-Game Match How do the top computer chess machines compare to the best human players? Who is beating whom? Link to comment Share on other sites More sharing options...
jts Posted December 8, 2017 Author Share Posted December 8, 2017 13 hours ago, BaalChatzaf said: How do the top computer chess machines compare to the best human players? Who is beating whom? The very strongest human chess players have an Elo rating above 2800. https://2700chess.com/ The strongest chess engines, apart from AlphaZero, currently are Houdini, Komodo, Stockfish. Their strength is a bit hard to measure because it depends on the hardware they are run on. Also they often play stronger than their rating because the programmers update them. Rating is based on past performance, current performance might be stronger. TCEC puts them at around Elo 3200. Chessdom.com apparently puts Stockfish at Elo 3400 if you look at their website. I don't know how good those numbers are for comparing against humans because they seldom or never play against humans except maybe odds games. For what it's worth and I don't know how much it's worth, here is a little chart showing the rating difference and expected result. The scoring is 1 point for a win, 0 for a loss, 0.5 each for a draw. difference in rating stronger player score 100 64% 200 76% 300 85% 400 92% Link to comment Share on other sites More sharing options...
jts Posted December 10, 2017 Author Share Posted December 10, 2017 More information about AlphaZero from chessbase, https://en.chessbase.com/post/the-future-is-here-alphazero-learns-chess Link to comment Share on other sites More sharing options...
jts Posted December 17, 2017 Author Share Posted December 17, 2017 AlphaGo gets beaten 100 to 0. Link to comment Share on other sites More sharing options...
jts Posted December 23, 2017 Author Share Posted December 23, 2017 AlphaZero beat Stockfish 28 wins, 72 draws, no losses. According to that it seems currently AlphaZero is the strongest chess entity ever to play chess. It uses neural net technology and has a totally different way of evaluating positions. Instead of evaluating positions in centipawns (0.01 pawn) like normal chess engines do, it figures out probabilities by playing whole games (or so I heard). It does a much smaller search tree, 20,000 nodes per second instead of 70 million. All the commentators are saying it plays in a style that was never seen before on a chessboard. It plays like Morphy and Anderssen but at a much higher level. It doesn't give a rat's ass about material, all that matters to it is piece activity. This is a risky way of playing chess because if you miscalculate you lose but it doesn't miscalculate. When most chess players (Morphy, Anderssen, Marshal, etc) sacrifice anything more than a pawn, it is usually part of a combination and it is usually calculated to checkmate or return of material. But AlphaZero sacrifices 2 pawns or a piece and it is not humanly possible to calculate it and there is no checkmate and no return of material and the positional compensation for the material sacrificed is sustained. In this video Sagar Shah explains some of the moves in a few of the games. There are many videos explaining AlphaZero's games. I picked this one because it is more entertaining than most. Link to comment Share on other sites More sharing options...
jts Posted December 25, 2017 Author Share Posted December 25, 2017 There was a time when chess masters gave lessons to amateurs for a price. In this video Kasparov, world champion 1985 - 2000, gives 7 hours 20 minutes of chess lessons for free. It is hard to ask for a better deal than that if you want to waste your time on chess. Link to comment Share on other sites More sharing options...
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now