Author: Uri Blass
Date: 01:22:45 10/22/04
Go up one level in this thread
On October 22, 2004 at 00:46:45, Tony Nichols wrote: >On October 21, 2004 at 06:24:49, Uri Blass wrote: > >>On October 21, 2004 at 04:48:15, Tony Nichols wrote: >> >>>On October 20, 2004 at 19:57:45, Uri Blass wrote: >>> >>>>On October 20, 2004 at 19:06:31, Tony Nichols wrote: >>>> >>>>>On October 20, 2004 at 04:17:26, Russell Reagan wrote: >>>>> >>>>>>On October 19, 2004 at 22:00:24, Tony Nichols wrote: >>>>>> >>>>>>>We dont bow down to calculators do we. >>>>>> >>>>>>If the contest is to see who can do arithmetic calculations with the greatest >>>>>>accuracy and efficiency, then we would indeed bow down to calculators as they >>>>>>are clearly superior at this task. >>>>>> >>>>>Why do you think we dont have these competitions. There is no point. Just as >>>>>there is no point in seeing who remembers theory better computer or man. >>>>>> >>>>>>>They definitely do >>>>>>>math faster than humans. >>>>>> >>>>>>Right. At some point the calculators became faster than humans for doing simple >>>>>>math. Calculators continued to get faster, while humans did not, and now >>>>>>computers are billions of times faster than humans. Right now computers play >>>>>>equally as well or better than the best humans in the world, and computers will >>>>>>only get faster. Do you dispute any of this? >>>>>> >>>>>Yes! I don't think computers play = to the best human players and are certainly >>>>>not better. They will get better but I think they have a long way to go. By the >>>>>way speed is not important when the program is not even analyzing. What is >>>>>purley mathematical about opening books and endgame tablebases? We dont use >>>>>calculators with built in solutions. They have to compute each equation, so when >>>>>we talk of chess programs let it be that part that computes. >>>>>> >>>>>>>The point is chess programs are tools for chess >>>>>>>players. >>>>>> >>>>>>That may be what chess programs are *to you*, but that isn't what they are to >>>>>>everyone. >>>>>> >>>>>I understand this and it makes me sad. To think someone could be so involved in >>>>>chess and not be a player is weird. I suppose some people use calculators as >>>>>paperweights, but this is not their best use! >>>>>> >>>>>>>If we want to have matches between humans and computers we should not >>>>>>>forget this. By the way, I will not be "snickering at how it doesnt even >>>>>>>understand" I will be having good laugh thinking about all the people who think >>>>>>>it does! >>>>>> >>>>>>I don't think anyone in this discussion claimed that computers understand >>>>>>anything. What I am claiming is that chess is 100% purely mathematical and that >>>>>>some day computers will play the game perfectly, while humans will never >>>>>>approach perfection. My claim can be proven mathematically, at least >>>>>>theoretically. I doubt that your claim that humans will always have a chance can >>>>>>be proven, and I'm not even sure there is any evidence to support it. Your claim >>>>>>seems mainly based upon human ego, emotion, or hope. >>>>> >>>>>If we assume that chess is 100% mathematical then why does the computer need the >>>>>crutches of opening books and endgame tablebases? Surely modern computers can >>>>>compute better than humans. Let this be the competition and I claim humans are >>>>>far superior to computers! >>>> >>>>I think that the importance of tablebases is not high and they were irrelevant >>>>in most of the computer-human games. >>>> >>>>humans also remember book lines that they did not invent and I do not see the >>>>difference between it and computers. >>>> >>>>Remembering lines is not always an advantage even if the lines are correct >>>>because you need to know what to do later and I can say that even kasparov lost >>>>a game when he simply did not remember his home analysis at some point and could >>>>not find the correct line in the board. >>>> >>>>In other words even if you remember some theory it may be better not to use it >>>>when you do not remember enough because you may get better position but lose >>>>because you do not remember how to continue in the better position. >>>> >>>>Uri >>> I think if you took away endgame tablebases we would see many more endgames. In >>>fact I think GM's would be lining up to play endgames against the computers. >>>Regards >>>Tony >> >>I think that you overestimate the value of tablebases. >>In most games at long time control programs can find the right moves without >>tablebases even when it gets an endgame. >> >>I do not think that some 5% chance that the program is going to play inferior >>moves is going to change the style of the opponents. >> >>Humans are not superior relative to top programs in simple endgames even if >>machines do not use tablebases. >> >>You could also ask to remove hash tables that are more important than >>tablebases. >>You can claim that hash tables are lists that the program write during the game >>and it is not allowed. >> >>Tablebases may be even counter productive in endgames because the program may >>avoid drawn KRB vs KR that it has practical chances to win and choose some drawn >>position when it evaluate itself as 0.02 pawns better but the human can draw >>easily >> >>Uri > >This might be true. I have read elswere that endgame tablebases are also >beneficial when transposing from late middle game to endgame. If this is true >then they become more important. >Regards >Tony This is exactly the problem that I talked about. Programs that use tablebases in the endgame may avoid going to drawn tablebase endgame when that is the only practical chance to win for them. Uri
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.