Author: Uri Blass
Date: 13:28:09 11/16/01
Go up one level in this thread
On November 16, 2001 at 15:48:15, Christophe Theron wrote: >On November 16, 2001 at 14:16:14, Uri Blass wrote: > >>On November 16, 2001 at 13:56:14, Roy Eassa wrote: >> >>>Agreed: having the strongest chess program for the PC will lend MS significant >>>prestige. >>> >>>Anybody think they have a chance at accomplishing this without buying out Tiger >>>or another of the current top engines? >> >>I believe that it is only a question of money. >> >>It is maybe not something that one programmer can do in a year(I am even not >>sure about it because maybe someone who has the right idea can do it) but if >>they hire 100 good programmers to work together then they may succeed in doing >>something clearly better than the programs of today. >> >>Every programmer should get a specific task in order to finish the program in >>few months and after doing it they can use a lot of computer time to test >>changes in the program by using many computers at the same time. >> >>Uri > > > >That's the theory. In practice, putting a lot of people on a task like this is >extremely inefficient. > >That's one of the main problem in software engineering. What one programmer >alone will do in one month, two programmers will do in one month. It's a common >joke, but it is close to the truth. > > > > Christophe It is possible that you are right that more than one programmer is not going to help much but I still believe that a lot of humans can help if you give them the right task. I believe that there are ideas that no programmer tried because they are too big and people are going to need too much time for them but if you put a lot of people for the task they are suddenly not too big. An example for it is better prunning rules It may be possible to define hundreds of rules to decide when a move is not logical and they can help to do the program faster because of better branching factor. One programmer probably has no time to find all the rules but many people can do it. I already posted ideas for rules in a previous post that it seems that nobody payed attention to it: An example for a possible rule is: the move white Bg2-h1 that is not a capture is illogical if there is no black bishop at h3. It is possible to hire many people to discover the relevant rules that practically are correct in almost all the positions that we can find and after that programs may use them in order to prune lines with too many illogical moves. The people who discover the rules do not have to be programmers and they have to look at a lot of games. one of them is going to get a database of all the position from games when Bg2-h1 was possible and his(her) task is going to be to define a rule when Bg2-h1 is not logical. Other humans are going to have the same task for other moves and together they will give pruning rules that they suggest the program to use. Uri
This page took 0.01 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.