We’ve looked at a lot of information over the last few posts to this blog, so we’re going to take a short pause today and catch our collective breath a little bit before proceeding. Admittedly I did cover things a little backward in those past posts, as I’d first received numerous requests for information on the analysis features of the Fritz “family” of chess playing programs (Fritz, Rybka, Junior, Hiarcs, and Shredder). Partway through that short series of blog posts, I began to receive requests on game input (how to add your personal games to a database). So that’s why we discussed analysis first and game input second.
Either way, that ground’s been covered, leaving us with some additional questions:
- What are the best settings for game analysis?
- What do you do with the analysis after it’s been generated?
We’re going to hit both of these two topics as this blog moves forward, but today we’ll take a quick little (useful) detour and see some interesting things which can occur when Fritz, and other engines analyze your games.
The first interesting item concerns endgame analysis. Here’s an Internet correspondence game which Fritz analyzed in “Blundercheck” mode:
…along with the board position at the highlighted move (20.Bd3+):
Do you notice anything unusual about the gamescore from move 20 onwards? Right – every move (except one) is a forced mate (assuming perfect play). What’s interesting to me is that Fritz12 found all of those forced mates in a couple of seconds – not a couple of seconds for each move, a couple of seconds total. When I started the analysis, Fritz12 began at the last move (you’ll recall that ChessBase chess engines analyze games starting with the final move and then work backwards through the game) and went “ZIP!” back to 30…a5 in less time than it takes to tell about it. It stopped for a minute or so to analyze and present the better alternative to 30…a5, then went “ZIP!” again back to move 19…Kh7. Now before we attribute this speed solely to endgame tablebases, we need to look at the final position of the game:
That’s a 13 piece endgame (remember that Kings count as pieces when we’re talking about endgame tablebases), the largest existing tablebases are for 6 piece endgames, and the largest tablebases I have on my computer are for 5 piece endings. So it’s safe to say (and I double-checked this manually using the “infinite analysis” feature) that if you allocate a generous hash table allotment to Fritz12, the engine by itself can find certain forced mates (even long sequences) really quickly even without the help of tablebases.
What we learn from this game’s analysis, by the way, was that Black was sunk fairly early in the game. Despite being behind in development, Black started swapping off material and by the time the smoke cleared when both players were finished Hoovering the board, it was too late for Black. Even the “better” 30th move wouldn’t have helped Black much; sure, it wasn’t a forced mate but having a lone King against a Rook, two Bishops, and four pawns ain’t much of an improvement.
Here’s another interesting bit from this game, which you may see from time to time in your own analyzed games:
Normally the move marked “Last book move” (highlighted in the illustration) would be the last move analyzed; the preceding moves would be skipped by the analysis engine since they’re in the opening book (that is, they’re part of known recognized opening theory). But notice that the 8th move for both players has been analyzed by Fritz12. What gives?
Very simply, White made a non-book move (resulting in a position which isn’t in Powerbook 2011) at move 8:
…and Black replied with another non-book move. But at move 9, White played a move which results in a position which is found in Powerbook, thus transposing back into recognized opening line:
So it’s important to note here that Fritz and other chess engines do identify transpositions into and out of the opening book, which is another reason why a large encyclopedic opening book such as Powerbook is a valuable tool in chess analysis.
We’ve previously looked at what the settings in the various analysis dialogues mean; next time around I’ll offer some tips on the actual settings themselves. Until then…
Have fun! – Steve Lopez
Chessplayers who have purchased their ChessBase brand chess computer software from USCFSales can receive free technical support and advice on their purchases straight from me; just shoot me an e-mail (email@example.com), but please remember to include the USCFSales order number from your ChessBase software purchase. – Steve
Copyright 2011, Steven A. Lopez. All rights reserved.