Author: Uri Blass
Date: 16:14:24 05/20/02
Go up one level in this thread
On May 20, 2002 at 18:25:49, Albert Silver wrote: > >>>I think it is easy to make an argument that permanent memory is written >>>material. If you store a openingbook on your harddrive it is written material in >>>my view. Its there to read for anyone with a PC to connetc to the HD. :-) And as >>>such against the rules! >> >>I have no problem with rules that limit the hardware. >> >>You can decide that the program should use only memory in the RAM of the >>computer but you cannot practically prevent programs to use opening book. >> >>opening book can be stored in the RAM and not in the HD. >> >>I think that for the future it may be interesting to make limitations about the >>hardware of the computers. >> >>Most humans have no chance against the top programs on the new hardware so if we >>want to see interesting comp-human games also in 2020 we should decide >>that for Fide tournaments the hardware that is allowed is only the palm of >>today. >> >>Another option is to make rules that allow humans to use notes and to use books >>during the game and another option is to use slower time control. >> >>I see no reason to decide that 2 hours/40 moves is the slowest time control in >>human-computer game and it is possible to decide about slower time control(I >>remember that 2.5 hours/40 moves or 2 hours/24 moves was used in the past). >> >>Another option is to allow both sides to take back moves but they lose time in >>the clock for their opponent after every take back so they cannot take back >>moves forever. >> >>It is going to be a fair game because the computers can also get the right to >>take back moves(today they do not use it but if there is going to be a serious >>game with the right to take back then I expect programmers to implement that >>option and also to teach their program to take back moves when they need it). > >You have to look further than that to realize the impossibility of this (not to >mention ridicule). I won't restate that machines don't *compete* so you are >basically try to outrace a fancy calculator no matter how much joy and emotional >stress that may give you. You can do the same by racing automatic motorcycles. >*You* may be competing, but it certainly is not. It's just a machine. > >Still, let's suppose we handicap the hardware to something that is within human >reach. Let's say we do this now, and set the maximum to..... 500 MHz and 128 Mb >of Ram. We estimate then that the programs are performing at roughly 2500-2550 >Elo. The exact value is unimportant so don't start arguing it is more or less. >Fine. So we let the machines in knowing they won't possibly do better than that >unless the humans playing it get careless. Then there is the opening book. What >do we do about that? If you limit it too much, players will be able to clearly >outbook the machine, but let's suppose a number of moves is found and accepted. >Fine. > >So for 5 years, the programs continue to progress in quality, always exploiting >their limited hardware, and they play at 2650 now. So after 5 years, the >programs are once more unacceptably strong, and talk of banning them is once >more heard by the players, so we reduce the hardware to 400 MHz. Ah! *Now* they >can play in our events. Why? Because we are assured they won't perform over 2550 >Elo. Our top brass still shine in comparison. Ah!... > >In 20 years, the hardware is only 350 MHz (software writers are running short of >ideas to continue improving software that exploits the ever DECREASING hardware >limits) and the machines must be custom made since no one really makes such >absurdly slow processors anymore. Even the latest Casio wristwatch goes faster >than that! BUT, we can STILL beat the machines. Ah!... > >Looks and sounds terribly silly doesn't it? That's where it would lead to. If >you limit the hardware to not allow a performance beyond a certain point, what >exactly are you achieving? > > Albert What I achieve is an improvement in the software and an interesting competition. I think it may be interesting to know what is the worst hardware that machines need in order to beat humans. I expect to see in the next 10 years a commercial program that is going to beat every player in the world and in this case the question is what is the future of computer-human games. I think that in this case it may be interesting to see what is the worst hardware that machines can use and still be better than humans. 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.