Author: KarinsDad
Date: 10:40:02 02/03/99
Go up one level in this thread
On February 02, 1999 at 15:16:11, James Robertson wrote: >On February 02, 1999 at 13:47:25, Jay Scott wrote: > >> >>On February 01, 1999 at 02:53:38, Jeff Anderson wrote: >> >>>I think that they should have a version of Hiarcs 7 playing that anyone could >>>buy. >> >>I think they should have a version of the grandmaster playing that anyone >>could buy. If the human can buy the program and prepare, then the program >>should be able to buy the human and prepare. >> >> Jay > >Humans (as far as I can see) have very few intrinsic advantages over a machine. >The most important of these is the ability to prepare for their opponent. >Computers have the large intrinsic advantages of never being tired, never being >fatigued, etc. My opinion is let them deal with their weaknesses, let us deal >with ours. > >James If that were the case, then Hiarcs 7 should be started in "analyze this particular GM" mode and left to it's own devises. I have no problem with that. Oh, you mean that mode does not exist and that Matt will have to use the opening book to tweak the program's opening book? The experiment is flawed. Matt can only tweak Hiarc based on previous games by the GM. If the GM takes this really seriously, he has several advantages over a tweaked Hiarcs 7. GM Advantages: 1) The GM can use Hiarcs 7 to analyze his own games and find where Hiarcs "thinks" the GM has flaws and correct them accordingly. Hence, the GM could if he wanted to, create a similar opening book to what Matt will create. 2) The GM can play dozens of normal time games vs. Hiarcs 7 and hundreds of faster games to determine Hiarcs' playing style and weaknesses. 3) The GM can analyze games from SSDF that Hiarcs has played against other computers. 4) The GM can look at Hiarcs' logs to determine what it "thinks". 5) If the GM finds a set of lines which lead to a loss for Hiarcs in standard times, Hiarcs has a high probability of playing thoses lines during the tournament if it can be led into them in the opening. An example of this may be games from the SSDF. The experiment proves nothing more than whether the GM prepares enough to win. The win of any game is non-deterministic, but the win of the match should be guaranteed IF he takes the 5 steps above and seriously prepares. Matt's contention is that the program should be capable of defeating the GM. This contention is probably only valid if either both parties would have been kept in the dark as to their opponent, or if the GM does not properly prepare for the match. I hope the GM wins all of the games, just to show that this type of experiment is flawed. This is more of an experiment of whether Matt can outwit the GM with preparation than whether Hiarcs 7 can beat the GM. KarinsDad
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.