|
|||||||||
Some observers predict that the progress of chess-playing computer software will likewise cause the end of the Royal Game, by finally answering the question, “Is chess a win or a draw with best play – and for who?”. While initial silicon efforts at pawn-pushing were amusing in their time, current titles can beat most people, and some programs can beat all players. Against that background, consider Gary M. Danelishen and his The Final Theory of Chess. The Final Theory of Chess is an attempt to lay a solid foundation upon which further analysis may be built in order to reach the first goal of a partial solution to the game of chess. Between mid 2004 and 2008, daily computer analysis was conducted and The Final Theory of Chess slowly was written. During this time, a network of six computers running the Fritz family of computer chess programs continuously calculated around the clock. Each previous round of analysis laid a foundation upon which future analysis was conducted… The early foundation of the book rests upon analysis using Fritz 7. For consistency purposes, the decision to continue with the Fritz family of chess software for any future upgrades was made. As soon as the network was up and running, the switch to Deep Fritz 8 was made in order to provide the best available analysis. Over a thin primary layer of Fritz 7 analysis, sits a thick secondary layer of Deep Fritz 8 analysis. A significant portion of the book owes itself to Deep Fritz 8’s analysis which was conducted between mid 2005 and late 2006. As soon as Deep Fritz 10 was introduced in late 2006, it replaced Deep Fritz 8 as the primary chess program. Deep Fritz 10 produced far superior analysis to either Fritz 7 or Deep Fritz 8. The superior quality of Deep Fritz 10 enabled greater productivity between late 2006 and late 2007, when analysis concluded prior to publication. If this seems a bit odd to you – a bit 2001: A Space Odyssey, HAL 9000-ish (another Arthur C. Clarke creation) – recall the news concerning the recently completed World Chess Championship match between Viswanathan Anand and Veselin Topalov, quoted from Chess Base online:
At the
time nobody knew the exact specifications of the cluster. Today we
know that Topalov had a seriously powerful set of computers, and in addition
access to a Blue Gene/L super-computer that had 8792 processors and could
execute around 500 teraFLOPS (one thousand billion floating point operations
per second)… Once we found out that he had some super computer, some cluster version of Rybka, something that can scale very well, it was obvious to us why Rybka 4 was being held back. So we had to do something in a hurry. Then the people from Hiarcs got in touch, Harvey Williamson and Mark Uniacke. Harvey had access to a very powerful computer, and he let me have it for the whole match. That helped – at least we could check some of the more critical areas with a really powerful machine for a couple of hours. That improved matters quite a bit. It’s one thing when you have it a couple of months before the match, but at least a lot of the critical areas these guys would check very thoroughly and make sure that I had something to play with. Just to be sure, of course, World Champion Anand also had his “human cluster” for preparation assistance: Magnus Carlsen, Garry Kasparov and Vladimir Kramnik. So – if you have author Gary M. Danelishen, his book, and the “Final Theory of Chess Project” (more on this, later) on your side, what do you actually have? For starters, you have a book the size of the Yellow Pages for a small city, jam-packed with analysis that creates an opening repertoire. Certainly you’ve thought at one time or another, Wouldn’t it be nice to have an army of master-level analysts putting together my opening lines for me, just like the Big Boys (and Girls) have? Danelishen and Fritz & Co. have done this for you. There’s quite a bit of computing packed in the pages of The Final Theory of Chess – a lot of it probably not found elsewhere (except perhaps in secret databases of stronger players who have done similar investigations.) To be clear, the author has made a few initial choices to cut his computer network’s task down a bit. For example, if you are playing White, the analysis of recommended moves starts with 1.d4. None of that Bobby Fischer “1.e4, best by test” kind of thing. Playing Black, the second player is encouraged to play 1…f5 whenever possible, even against unusual openings – although he should answer 1.e4 with 1…e5, when he will be facing, among other things, the Ruy Lopez, the Four Knights Game, the Vienna Game, the Bishop’s Opening and the King’s Gambit. (Book reviewer John Elburg bemoans the book’s absence of coverage of his beloved Latvian Gambit, but The Final Theory of Chess recommends 2…Nc6 as the response for Black after 2.Nf3. For that matter, it recommends 3…Nf6 and the Two Knights Defense for Black after 3.Bc4 – no Giuoco Piano or Jerome Gambit, either; Mr. Elburg, I feel your pain!) At first glance a reader might hesitate at the book’s subtle implication that the Dutch Defense, 1.d4 (recommended for White) f5 (recommended for Black), is going to lead to the solution of chess. However, that isn’t the half of it: in Danelishen’s schema, White is playing 1.d4 so that after 1…d5 he can play 2.e4 and aim for the Blackmar-Diemer Gambit with 2…dxe4 3.Nc3 Nf6 4.f3! Again, I can see the Doubters and the Delighteds out there begin to shuffle in different directions. Meanwhile, both will have to add to their list: for starters the “system” will have to have a White line against some 1.e4 defenses, after all – Caro Kann, French, Nimzovich and Pirc, arrived at by transposition. Indian defenses, on the other hand, will be dealt with by following 1.d4 Nf6 2.f3, again with the hope of continuing 2…d5 3.e4 dxe4 4.Nc3, returning to the Blackmar-Diemer Gambit. The “Final Theory of Chess Project” adds to all of this. Danelishen has established an online chess openings encyclopedia similar to Wikipedia, which can be added to and edited by others who have done similar computer-based analysis. Thus, the number of researchers (human and computer) and the amount of research into The Final Theory can be further expanded. The large majority of The Final Theory of Chess is composed of pages of analysis and diagrams. I can give you a replica snippet (grabbed randomly from the section on the Advance Variation of the French Defense) D. 6…Nb4 7.O-O Nxd3 8.cxd3 (See first diagram) (Black seeks the immediate removal of White’s king-bishop while exchanging down material) a. 8…Bd7 9.Nxd4 Qb6 10.Nb3 Ne7 11.Bd3 Qc7 12.Bd4 Bb5 13.Rd1 Nc6 14.Nc3 (=(0.1)/21(DF8)) Danelishen, in his “Understanding the Notation” section, notes that the non-highlighted lines show analysis that he feels “to be trustworthy.” The lighter highlighting (I am using light green; in the book he uses a light gray) is for analysis that he feels “to be questionable” – “further analysis should be conducted to explore other options for White.” The darker highlighting (again, I am using teal; in the book he uses a darker gray) indicates “a second degree of questionability of the analysis” – its purpose is only to provide additional information for future analysis that may be conducted to improve the questionable analysis.” In addition, bolding (not in the above example) is used to indicate “thematic moves”. Diagrams are set with either White or Black at the bottom, depending on whether the section is focusing upon the White or Black repertoire. The book’s organization is in three parts. First comes the supportive material:
Including, at the end:
The second part of the book includes 95 pages of analysis on openings. This is a bit misleading, as many of the openings include deeper analyses that are referred to in the following, third part of the book: 132 different Appendices, covering over 270 pages (for a total of almost 400 pages of analysis). This arrangement allows the author to deal with sidelines succinctly, and introduce his main lines – which are then followed up in more detail further on. How to judge all of this? First, as news from the most recent World Championship indicates (see above), The Final Theory of Chess is cutting edge in its concept of using analysis from chess software on a computer network. (Were Danelishen to start his project today, no doubt he might use Deep Rybka 4, the current choice of top grandmasters around the world.) Philosophy Looks at Chess (Open Court, 2008) was completely correct in including the essay “Is Gary Kasparov a cyborg?” For those seriously into using computer analysis of openings, the online “Final Theory of Chess Project” offers the option of combining resources, not to mention communing with like-minded souls. Second, there is a lot in The Final Theory of Chess that may be new – inspiring or cautionary – for opening explorers, especially those interested in the Blackmar Diemer Gambit and the Dutch Defense and who don’t have analysis software running on a network (plus years of time to invest). I did not evaluate the coverage of every opening (one of the few times I broke my rule of reading every book twice before reviewing them). However, I surveyed the main lines of the BDG as covered in the book, and those players outside of the Gemeinde probably will not be surprised to learn that Fritz et. al. do not always give White an easy time. For example, the O’Kelly Variation (1.d4 d5 2.e4 dxe4 3.Nc3 Nf6 4.f3 c6 5.Nxe4) scores as poorly as -.50 (worse by half a pawn). Others: the Vienna Defense (1.d4 d5 2.e4 dxe4 3.Nc3 Nf6 4.f3 Bf5) -.66; the Gunderam Defense (1.d4 d5 2.e4 dxe4 3.Nc3 Nf6 4.f3 exf3 5.Nxf3 Bf5)-.38; the Teichmann Defense (1.d4 d5 2.e4 dxe4 3.Nc3 Nf6 4.f3 exf3 5.Nxf3 Bg4) -.49; the Ziegler Defense (1.d4 d5 2.e4 dxe4 3.Nc3 Nf6 4.f3 exf3 5.Nxf3 c6) -.48; the Euwe Defense (1.d4 d5 2.e4 dxe4 3.Nc3 Nf6 4.f3 exf3 5.Nxf3 e6) -.29; and the Bogoljubow Defense (1.d4 d5 2.e4 dxe4 3.Nc3 Nf6 4.f3 exf3 5.Nxf3 g6) -.48. Again, Defenders and Doubters can make of that what they will – and those who do not or will not play the Blackmar Diemer Gambit will probably seek erudition elsewhere – but club players should keep in mind Geoff Chandler’s fanciful Blunder Table before scoffing too loudly at White’s chances (masters will have skipped the previous paragraph, anyhow). All players should be able to spot their opponent leaving a mate in one on.
A 1200 player should win if an opponent blunders a Queen or a Rook. The Final Theory of Chess is not a first resource for a beginner who is just learning about openings (not enough explanation; try Chess Opening Essentials, Volume 1) or even someone first exploring the Blackmar Diemer Gambit (ditto; try IM Gary Lane’s The Blackmar-Diemer Gambit). It can be tiring on the eyes, if consumed in large doses – although the 850+ diagrams are a help. It is hard to know how much of the opening lines were chosen by the analyses of the Fritzes, instead of by their opening books (computer opening books give guidance masked as understanding, but they also mandate direction) or by the author, himself; and what effect all of this has had on pruning the branches of analysis. Still, the
opportunity to see what computers have to add to a whole lot of openings
(not just the BDG) – it is attractive, is it not? And, supporting a
consortium of researchers chasing the “Final” question of chess?
Besides, I can think of many more dissipative uses of $25... As a final note on the “Final Theory,” one of the members of Chess.com posted this on their website:
|
Chessville
The
Advertise to Single insert:
|
||||||||
|
|||||||||
