In any case, at the end of the book he gives four games he played against Fritz 4 in 1997 and made some observations about the program's play. At the time he was convinced that offering Fritz some small material gain in order to get the initiative was the most plausible strategy. He added that in five or ten years humans may become helpless against chess playing programs.
Fritz, a German program, was originally developed for Chessbase by Frans Morsch based on his Quest program, ported to DOS, and then Windows by Mathias Feist.
In 1991, ChessBase approached the Dutch chess programmer Frans Morsch about writing a chess engine to add to the database program which they sold. Morsch adapted his program and ChessBase and it was released it for sale that year in the US as Knightstalker and Fritz throughout the rest of the world.
In 1995, Fritz 3 won the World Computer Chess Championship in Hong Kong by beating an early version of Deep Blue. This was the first time that a program running on a consumer level microcomputer defeated the mainframes that had previously dominated this event.
In 1998, Fritz 5 was released and it included a Friend mode which in which the engine attempted to adjust its strength over the course of a game to match that of its opponent. Fritz 5.32 was released soon after replacing the 16 bit with a 32 bit one.
In 2002, in Bahrain Deep Fritz drew a match against Vladimir Kramnik 4–4. Fritz 7, which was released that year, included the ability to play on the Playchess server. In November 2003, X3D Fritz, a version of Deep Fritz with a 3D interface, drew a four-game match against Garry Kasparov.
Over the years Fritz continued to improve with new releases. Fritz 15 was released on November 25, 2015 with new features, including switching to Vasik Rajlich's famous Rybka engine.
On March 30, 2022, Fritz 18 was released with a neural network engine. In July 2021, the developers of the open source Stockfish engine filed a lawsuit against Chessbase alleging that Fat Fritz 2.0 is a derivative of Stockfish.
So, what about that Fritz 4 program that Yermolinsky played? The program was advertised as combining "world-class playing strength with world-class training facilities...the only top playing program
actually designed to help you learn to play chess."
It was a "playing program, a partner, advisor, teacher and annotator, your own personal grandmaster to help you appreciate and understand the game."
By the way, it required a CD-ROM drive which was not available on all home computers at that time.
This fantastic program also boasted a standardized Windows interface with icons. drop down menus and buttons. It could be fully integrated with your other Windows or Macintosh programs like Word, Pagemaker and the ChessBase training software programs.
Fritz 4 also had a brilliant 3D virtual reality board, new coaching functions, automatic analysis in plain English, full database capabilities in ChessBase or PGN format. It had multiple chess engine options, increased speed and strength plus a knowledge of certain five piece endings.
In 1997 an upgrade from Fritz 3 costs $70 while Fritz 4 could be had from the USCF for $140 (that's about $260 in today's dollars), so it was pretty pricey. In 1997 that $140 would buy you 120 loaves of bread, 133 pounds of chicken fryers, 115 gallons of gas or 74 gallons of milk.
A game that I liked (Komodo 14)
[Event "Offhand 10 minute game"]
[Site "?"]
[Date "1997.??.??"]
[Round "?"]
[White "Alex Yermolinsky"]
[Black "Fritz 4"]
[Result "1-0"]
[Annotator "Stockfish 15"]
[PlyCount "73"]
[SourceVersionDate "2022.05.23"]
1. e4 e6 2. d4 d5 3. Nc3 Bb4 4. e5 Ne7 5. a3 Bxc3+ 6. bxc3 c5 7. h4 Nbc6 8. h5
cxd4 9. cxd4 Qa5+ 10. Bd2 Qa4 11. Nf3 Nxd4 12. Bd3 Nec6 13. Kf1 Nf5 14. h6 {
This is all book. White has a lot of play for the sacrificed and the position
is far from clear.} g6 {This move, surrendering the dark squares, is "a deadly
sin in such positions" as it was put by Yermolinsky.} 15. Bg5 O-O 16. Bf6 Qf4 {
This places the Q in grave danger, but it's loss was beyond the engine's
horizon. Stockfish 15 instantly recognized it as a bad move slapping it with
an evaluation of minus 6.5 Ps.} 17. Bxf5 Qxf5 18. Rh4 b6 19. Qd2 Ba6+ 20. Kg1
Rac8 {Yermolinksy noted that Fritz 4 was looking at a repetition with 21.Rf4
Qh5 22.Rh4 and evaluated the position at 0.00. Yermolinksy, like Stockfisdh,
realized that was totally wrong, but added that white doesn't have a lot of
time on the clock to work things out.} 21. Bg7 {Objectively not the best, but
due to time considerations he decided to capture the f-Pawn. White still has a
huge advantage though.} Rfe8 22. Rf4 Qh5 23. g4 Qh3 24. Ng5 Qh4 25. Nxf7 Rc7
26. Nd6 Rb8 27. Qe3 g5 28. Rf3 Na5 29. Rg3 Bb7 {The move Yermolinsky expected
as the logical continuation of black's last move was 29...Nc4. He gave some
analysis involving inferior moves the engine could have played that would have
made his task more difficult in the few minutes remaining, but engines don't
play that way! Psychology is not part of their thinking process.} 30. Qf3 d4
31. Nxb7 Rc3 32. Qg2 Nxb7 33. Rxc3 dxc3 34. Rd1 Nc5 35. Rd4 {Yermolinsky
called this a typical human move with no particular purpose! Immediately after
he played it he saw a better plan. Interestingly, Stockfish spots a mate in 16,
so the move played is not bad by any means.} (35. Qf3 {was Yeromlinky's
suggestion. Then Kg2 and Rh1 would force immeduate resignation. After this
plan Stockfish points out a mate in 13 moves.} {Stockfish 15:} Rc8 36. Rd4 Re8
37. Qc6 Kf7 38. Kg2 a5 39. Qxb6 Nd3 40. cxd3 c2 41. Qb7+ Kg6 42. Qf3 Qh1+ 43.
Kxh1 c1=Q+ 44. Kh2 Qf4+ 45. Rxf4 gxf4 46. Qxf4 Rf8 47. Bxf8 a4 48. Qf6# {
Compare this position to the one at the end of the game.}) 35... a6 36. Qf3 Re8
37. Kg2 {Faced with mate 13 Fritz began shedding material.} (37. Kg2 {
Stockfish 15:} Rc8 38. Qf6 Re8 39. Bh8 Qxh6 40. Qxh6 Kxh8 41. Qf6+ Kg8 42.
Qxg5+ Kh8 43. Qf6+ Kg8 44. Rf4 Nd7 45. Qf7+ Kh8 46. Qxe8+ Kg7 47. Qxd7+ Kh6 48.
Qe7 Kg6 49. Qf6# {The same mate as in the note to white's 35th move.}) 1-0
No comments:
Post a Comment