Author: stuart taylor
Date: 16:34:12 11/29/05
Go up one level in this thread
On November 28, 2005 at 14:30:50, Sandro Necchi wrote: >On November 28, 2005 at 08:15:39, stuart taylor wrote: > >>On November 28, 2005 at 02:04:17, Sandro Necchi wrote: >> >>>On November 27, 2005 at 17:01:28, stuart taylor wrote: >>> >>>>How's the Shredder tuning coming on now? Does it look like it will again take >>>>the competition with a storm? >>> >>>Stefan did not started yet, but he feels he can do it. >>> >>>>Just those few improvements, and it might make for the REAL companion for true >>>>expert super GM analysis and other values. >>>>It should also be honest knowledge and understanding programmed in it, and not >>>>just something to outsmart the competition, subjectively, due to the particular >>>>mechanical weaknesses with present programs, or book competitions etc. >>> >>>Stefan will work hard on the chess engine and I will improve the book. We feel >>>optimistic about the outcome. >> >>You are admitting what other programmers maybe do not. But it doesn't matter to >>me if he works on it for only half an hour, if the results are fantastic. BUT, >>nevertheless, I would have prefered that all the time I'm waiting, forthe whole >>year, it is being worked on. >> >>I mean, if it was, then wouldn't it get to be 200 elo stronger? whereas if he >>only does it for a few weeks, it will only get 50 elo stronger (if that)? > >Well, you do not me enough...what I would like is a program that scores 100% >against the competition or better (120%). I am joking but in reality I am never >satisfied, so if it was for me 500 points more would be the minimum...is it >enough? > >> >>> >>>> >>>>I really would have liked to feel that it's being worked on the whole year, not >>>>just a few small improvements, just to get to the top, with plenty left for the >>>>next upgrade (Shredder 11). >>> >>>The next upgrade will be Shredder 10...I am sure it will be stronger than 9...it >>>must be. >> >>I mean, Shredder 9 might be 50 elo better and Shredder 11 another 50 = 100 elo >>stronger than 9? >>Why not put it all into Shredder 10, so that it would be 100 stronger, if not >>200? > >We need to keep the gap that there was before, so you can guess how much we >need. > >> >>> >>>>Waiting in great anticipation! >>>>S.Taylor >>>> >>>>n.b. Or other programmers, are any of you REALLY expecting of breaking away from >>>>all competition? >>>>S.T. >>> >>>the fight did not started yet...wait Shredder 10... >> >>We hope! But I'm not so interested in just about winning the fight overall. I'm >>waiting years already for a really finished product, which is much stronger in >>every way, espially when computer is alone without books, and that the strength >>comes from great and correctly applied chess knowledge. > >Well, I do feel the same as you, but WITH BOOKS! Do you want to remove my hobby? >I am joking! >I think the books add something, but the engine is KING, so the engine must be >strong even without books. Obviously. Indeed, the (opening) books today have a job trying to be sure that they are better than the engine would play without them. And obviously, the books should make it that much greater, but not to be INSTEAD of the engine. The existence of books in fact, make it take much longer to see the strength of differnt engines and to get an accurate rating. But it is still a necesity, and the pride of the program, and if it auments the particular program, then that is also a great art. Why not 500 elo more? I'm also trying to be realistic, in saying that a year's work might yield a clear 200 increase, compared to a 50 increase after a few very concentrated weeks of hard work. And we ARE waiting, the whole year. Yes, It's good to want to atleast keep the gap. But firstly, that might not always succeed, (but then you would make a patch wouldn't you/he, like from 7 to 7.04), and secondly, an overall topping, might still have weaknesses, which make a person prefer a different program for analysing a different type of position, and you can never know which program will find a best move, because if the top ssdf proram is weak in some cases, then you are frustrated for another year and so on and so on. e.g. as I've pointed out many times, if you are playing correspondence, and you are playing Shredder 9 moves and the other is playing all Junior 9 moves, and you are sure you will win because you are using the strongest program on the market, you are likely to lose, because Junior 9 was a bit stronger when playing against Shredder. So therefore, you should have used Shredder 7.04 to have a much better chance to throttle Junior 9. But people keep forgetting, and don't think too much, and will never believe that anything Shredder can do, Shredder 7.04 cannot do better, because Shredder 9 is better, whatever that means. Anyway, I'm being too long-winded now. But I don't think too highly of a 50 elo increase, when some positions it plays much weaker than its predescessor. 200 elo, however, should be enough to cover all aspects and to be a stronger engine in every aspect, not just a different priority of style which happens to succeed more in the competition at that present time. > >> >>But all this is perhaps not the comercial way of seeing things! > >Yes, a little!:-) Christophe Theron seemed to be quite open about it when I challenged him on that question. He was saying that an improvement of 20 elo points IS an improvement, and was very happy about it. That way, you have to keep waiting for 10 years before you get 200 elo more, and what if I want it now, and in ten years from now I will not be young enough anymore, to appreciate it, and to improve my skills by practicing/analysing with it, or will already be too busy for chess anymore? In that case, I'd prefer to enjoy a program from previous year which has 20 elo less, and not keep dreaming of something which will not come until 10 years later, and just to leave that for a younger generation, or even for those who aren't born yet. 200 elo would therefore be 2 things. 1). A real and noticeable improvement in all aspects of playing, which would really be a pleasure. And 2). It would probably clinch the prowess of chess, and move beyond the criticisms for which chess programs are still being chastized, and will make a REAL challenge to find any further weaknesses which are still there. I'd much prefer to hear that "top programs do everything very well, and are REALLY worthy GM's in every meaning of thet word, but only, there are still a few things which can still be improved upon, which some computer expert GM's can still exploit at times, but they are not THAT bad even at that". But not to hear that "computers still don't have a clue in some positions, and although in some positions they are like super GM's or better, im other positions, they are below 2000 elo level of understanding". With a 200 elo increase from present, I believe this (the last statement) will no longer be sayable. And when a strong player losses, he won't keep thinking that it was just because he missed some wicked tactic. S.Taylor > >>S.Taylor >>> >Sandro
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.