Author: Bruce Moreland
Date: 01:40:40 02/21/99
Go up one level in this thread
On February 21, 1999 at 00:31:37, KarinsDad wrote: >On February 20, 1999 at 10:40:26, Robert Hyatt wrote: > >>On February 20, 1999 at 04:25:58, Mark Young wrote: >> >>>On February 20, 1999 at 03:29:57, Bruce Moreland wrote: >>> >[snip] >>>> >>>>In a 5 0 game, a program will have less time per move, obviously, than it would >>>>in a 5 5 game, but there is also a mad scramble at the end where the human, even >>>>has little time and has to deal with something that has more time and won't roll >>>>over and die. >>>> >>>>In a 5 5 they can get an advantage and convert it. > >How? Because the program is not designed to handle this? This seems incorrect. >I'm sure you could think of 3 or 4 ways for a program to gain time when it is in >a semi-forced set of moves or whatever. It could even repeat a move, just to >gain 10 seconds in some circumstances. Granted, a human would have an advantage >over a program if it is a stupid program with regard to time management. But why >should any program be that way? Seems strange to me. Imagine you are up a pawn and have 60 seconds left to win a 5 0 game. All you get is 60 seconds. The program doesn't undergo mental collapse and even when it is down a queen it doesn't resign (mine doesn't, at least). You have 60 seconds to increase your advantage, queen a pawn, kick my pieces out of the way, and checkmate me. It is often extremely hard to do this without running out of time. In a 5 5 you get 5 seconds additional per move, so those moves when you are queening your pawn and I can't stop you are essentially free, etc. This seems very obvious, I don't see where the contention is. >>>>If you automate a program and tell it to play only humans at 5 0, and take >>>>another one and tell it to play only humans at 5 5, the first one will usually >>>>have a significantly higher rating, I think. > >Only if it has weak time management algorithms. No, see above. Programs will flag the hell out of humans at 5 0 if they are silly enough to want to play this insane time control. They won't flag in 5 5. There are lots of games that the program will draw or lose rather than win or draw. >[snip] >> >>yes, but the 'automatic' programs have problems as well. They have to adjust >>their time _by themselves" without a human to say 'speed up' you are getting low >>on time. And even more important, an automatic program has to contend with an >>IM that plays 30 games in a row, fishing for an opening bust. Manual operators >>won't allow that and stop after they see what is happening. > >Programs can be changed to disallow this as well. They can keep track of who >they played and when, and with which opening, etc. The IM played e4 last time, >so instead of e6 I will play c6, and next time I will play c5. > >There are no disadvantages to automated programs, just disadvantages to what is >coded within them. Machine learning is very important if you play on the net, but it's hard to make it bulletproof. If you have flaws in your machine learning, you'll lose some games before you can catch this stuff. If you are playing manually, you'll try something else immediately, assuming that you are not acting as an absolutely neutral intermediary between the program and the server. bruce
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.