 "Chess (#1069)")
Among the more crucial aspects of preparation for the world title match will be the effective deployment of electronic resources. Both teams will use many engines and access multiple databases.
The selection and management of silicon help is critical. Chances are both teams will have the normal commercial engines and souped up beta versions as well. Their databases will not only include the normal offerings but also games from obscure, high quality correspondence events.
Match preparation isn't about brute force analysis though that is important. It is about tweaking engine parameters to unearth and examine ideas, which are likely to turn up and to examine lines of play which look logical to a human being. Computers can play strange "inhuman" moves, which may be perfectly feasible in analytical terms but are not likely to be tried by humans.
Different engines have different preferences - their search and pruning algorithms are different and so are the weights set for variables like material, king safety, mobility, etc. To some extent those settings can be "tampered" by experienced players. The Junior series for example, are hyper-aggressive chess engine, Houdini is the best in terms of all-round strength, HiArcs and Shredder are more positional in their outlook.
Both players present unusual problems in terms of prep work. Viswanathan Anand has a massive range in terms of the openings he does play and he finds highly complex tactical ideas in all sorts of positions. Magnus Carlsen avoids early complications and has an amorphous repertoire. He often digs out historical openings which don't offer objective advantage but postpone the struggle until both players are "out of book". A lot of inspired guessing will be required to pick the systems which are likely to pop up.
The diagram, WHITE TO PLAY AND DRAW is a study by Djaja. Even very strong humans struggle to solve it. Computers find it tough for different reasons. Computers can use tablebases, which means perfect play in positions with up to seven pieces.
The solution is 1. Nf5+ Kd8 Here computers will defend by 2. Nxd4 a2 3. Rb7! a1=Q 4. Rxb6 Qxd4 5. a7. This is a 7 piece position, the computer "knows" as drawn.
The elegant "human" solution is 2. Ra8+! Kxd7 3. a7 Ra4 4. Rg8 Rba6 White relies on 4.--a2 5. a8=Q Rxa8 6. Rxa8 or 4.-Rxa7? 5. Rg7+. Black doubles on a-file and now, how does white save?
Answer 5. Nh6!! - Perpetual check on the g-file is forced with the white rook protected except on g5,g7 and those squares can't be hit due to the knight's control of f5,f7. Black can't avoid this except by pulling a rook off the a-file and losing to a8=Q. Computers find this idea eventually.
The selection and management of silicon help is critical. Chances are both teams will have the normal commercial engines and souped up beta versions as well. Their databases will not only include the normal offerings but also games from obscure, high quality correspondence events.
Match preparation isn't about brute force analysis though that is important. It is about tweaking engine parameters to unearth and examine ideas, which are likely to turn up and to examine lines of play which look logical to a human being. Computers can play strange "inhuman" moves, which may be perfectly feasible in analytical terms but are not likely to be tried by humans.
Different engines have different preferences - their search and pruning algorithms are different and so are the weights set for variables like material, king safety, mobility, etc. To some extent those settings can be "tampered" by experienced players. The Junior series for example, are hyper-aggressive chess engine, Houdini is the best in terms of all-round strength, HiArcs and Shredder are more positional in their outlook.
Both players present unusual problems in terms of prep work. Viswanathan Anand has a massive range in terms of the openings he does play and he finds highly complex tactical ideas in all sorts of positions. Magnus Carlsen avoids early complications and has an amorphous repertoire. He often digs out historical openings which don't offer objective advantage but postpone the struggle until both players are "out of book". A lot of inspired guessing will be required to pick the systems which are likely to pop up.
The diagram, WHITE TO PLAY AND DRAW is a study by Djaja. Even very strong humans struggle to solve it. Computers find it tough for different reasons. Computers can use tablebases, which means perfect play in positions with up to seven pieces.
The solution is 1. Nf5+ Kd8 Here computers will defend by 2. Nxd4 a2 3. Rb7! a1=Q 4. Rxb6 Qxd4 5. a7. This is a 7 piece position, the computer "knows" as drawn.
The elegant "human" solution is 2. Ra8+! Kxd7 3. a7 Ra4 4. Rg8 Rba6 White relies on 4.--a2 5. a8=Q Rxa8 6. Rxa8 or 4.-Rxa7? 5. Rg7+. Black doubles on a-file and now, how does white save?
Answer 5. Nh6!! - Perpetual check on the g-file is forced with the white rook protected except on g5,g7 and those squares can't be hit due to the knight's control of f5,f7. Black can't avoid this except by pulling a rook off the a-file and losing to a8=Q. Computers find this idea eventually.
Devangshu Datta is an internationally rated chess and correspondence chess player