Stockfish is the unchallenged strongest, but pure strength is not always all there is to it because in opening preparation a player’s aim is to reach a position that he is familiar with, but not his opponent. At least in over the board play; correspondence play is another matter.
Of course, such a position must have possibilities to play for a win. To that end, analysis by a weaker engine can actually be preferable even though the position might be won by Stockfish. This kind of opening preparation only applies at the upper levels where surprising the opponent is more important than playing the absolute best move. At the non-Master level such things are far less important because most players don’t follow the “book” more than five moves anyway! In fact, for many of us winning a won position is not assured.
Correspondence GM Jon Edwards, winner of the 2022 World Correspondence Championship, has said that in modern correspondence play which permits unlimited engine use, in order to get an edge, a player must make better use of all the tools available than his competitors can.
In addition to Stockfish and Leela, he found that Houdini would sometimes be able to generate ideas not found by the other engines, but you had ti run it for several days.
Also, of great importance was a database of correspondence games and he noted that even elite OTB players rely on the games of top correspondence players.
When it come to over-the-board play, Sadler says that using Leela with its Win-Draw-Loss evaluation can generate some interesting opening ideas that are difficult to refute over-the-board.
The Win-Draw-Loss Evaluation is an interesting concept. Traditionally, chess engines evaluate position in terms of Pawns. Leela developers clam that is not the most logical way to evaluate a position. Unfortunately, most chess GUIs are not able to display WDL probabilities.
My ChessBase programs have Fat Fritz (based on the LCZero engine), but it needs a very high performance graphics card which I do not have in order to achieve its full playing strength.
My chess programs have 25 engines available and in a two minutes per game tournament with five engines competing that I ran, Stockfish 17 walloped them all except for a draw against Drain by Komodo. Because of that I played a match between the two and Stockfish scored a decisive victory (+8 -1 =7). Consequently, there is nothing new to report...Fritz and Stockfish are still the winners.
Back in the days of infancy of engines I was playing e-mail chess with the Correspondence Chess League of America which did not allow engine use. Nevertheless, it was obvious that some opponents were using them anyway. It didn’t matter a whole lot though because at first engines could be beaten. Twenty years ago Correspondence GM Robin Smith published a book telling us how to do it.
Eventually though two things happened: 1) stopping engine use became a losing battle and 2) engine got too strong and Smith’s book was all but useless. As a result of engine use I switched to playing on Lechenicher SchachServer, a site that allowed engine use.
I stopped playing there when almost all the games were colorless draws. A while back I checked out the site and noticed that unlike in the old days the site has little activity and it takes forever plus a couple of days for tournaments to fill up.
The below game was played in 2008 when some of the top engines were Fritz, Shredder, Rybka, HIARCS, Junior, Zappa, Jonny and Fruit.
Rybka was released in August of 2008 and was available from both Chessbase and Convekta. It was strong enough that it won a match against GM Joel Benjamin that year.
Rybka won the World Computer Championship in 2007, 2008, 2009 and 2010. But then in 2010 it was striped of its titles and banned from future events by the International Computer Games Association because the developer was accused of plagiarizing code from other programs, especially Fruit.
HIARCS 12 supplanted Rybka as the top engine. Stockfish 1.0 was released in November, 2008. It was an offshoot of the open-source engine Glaurung.
In the following game I believe I was using the Fruit engine which was commercially available from September 2005 until July 2007 when it became free and its further development ended.
T.N.1975–Tartajubow20500–1D00Lechenicher SchachServer Open2008Stockfish 17
D00: Queen Pawn Opening 1.d4 d5 2.f3 f6 3.c3 This innocuous move suggests that my opponent was letting whatever engine he was using select his moves without any input on his part. f5 4.d2 This move reinforced my belief that my opponent was allowinf his engine, and a rather weak one at that, select all of his moves. 4.h4 g4 5.h3 h5 6.g4 g6 7.g2 e6 8.g5 fd7 9.xg6 hxg6 10.e4 c6 11.h4 b6 White has the better position, but he was apparently content to accept the draw offered by his higher rated opponent. Young,A (2390)-Golod,V (2552) Santa Monica 2004 4.h3 e6 5.g4 g6 6.e5 bd7 7.xg6 hxg6 8.g2 c5 9.0-0 The position is about even. Sluka,R (2319)-Sosna,J (2364) Czechia 2015 4...e6 5.e3 c6 6.a3 d6 7.e2 bd7 8.g3 0-0 Somewhat better was 8...Bg6, but rightly or wrongly, I reasoned that white had lost time with his N maneuvers and if he took the B then I could gain even more time advancing the K-side Ps. 9.xf5 exf5 10.d3 e4 11.c4 df6 12.c5 c7 13.0-0 Srockfish 17 evaluates this position as equal and recommends either 13...Ng5 or 13...g5 e7 14.e1 This rates a ? and after the recommended 14.Be1 (apparently preserving the two Bs) the position is equal. Stockfish recommends plating the immediate 14...g4, but the text is also quite good. e6 15.h3 White is dawdling and with this move he slightly weakens his K-side. More to the point was 15.b4 although it must be admitted that white’s position is not very promising. g5 Initiating a strong K-side attack. 16.c3 g4 17.e5 For better or worse white had to minimize the damage to his K-sid and play 17.hxg4. gxh3 18.g3 Played with the forlorn hope of keeping his K-side closed. 18.gxh3 h8 19.f3 g8+ 20.h1 xe5 21.dxe5 g4 22.hxg4 xg4 wins for black. 18...xe5 Eliminating white's only well placed piece. 19.dxe5 g4 20.xe4 fxe4 White has a won position, but there is no way to force a breakthrough on the K-side. However, white's next move is a gross blinder that loses immediately. He could have tried to hold on with any reasonable move. 21.f3 h2+ 22.g2 h1+ White resigned. 22...h1+ 23.xh1 23.xh1 h6+ 24.g1 h2+ 25.f1 f2# 23...xe3+ 0–1
No comments:
Post a Comment