Author: James T. Walker
Date: 15:12:57 01/12/01
Go up one level in this thread
On January 12, 2001 at 14:27:08, Dann Corbit wrote: >On January 12, 2001 at 10:43:33, James T. Walker wrote: > >>Well I've enjoyed the discussion with you but take a look at your arguments. >>You want a mathmatical certainty that programs are GM strength but you can not >>or will not define what that certainty is. > >Certainty will always be a probability range. Consider this hunk of stuff from >the top of the SSDF list: > > THE SSDF RATING LIST 2000-12-28 76240 games played by 211 computers > Rating + - Games Won Oppo > ------ --- --- ----- --- ---- > 1 Fritz 6.0 128MB K6-2 450 MHz 2629 25 -24 845 67% 2506 > 2 Junior 6.0 128MB K6-2 450 MHz 2589 23 -22 1027 65% 2483 > 3 Chess Tiger 12.0 DOS 128MB K6-2 450 MHz 2572 28 -27 651 62% 2483 > 5 Fritz 5.32 128MB K6-2 450 MHz 2548 28 -27 658 60% 2475 > 6 Nimzo 7.32 128MB K6-2 450 MHz 2547 25 -25 791 60% 2476 > 7 Hiarcs 7.01 128MB K6-2 450 MHz 2541 42 -42 274 53% 2522 > 8 Gandalf 4.32f 128MB K6-2 450 MHz 2530 29 -29 584 52% 2519 > 9 Hiarcs 7.32 128MB K6-2 450 MHz 2529 29 -28 597 59% 2466 > 10 Junior 5.0 128MB K6-2 450 MHz 2528 26 -25 746 57% 2476 > >What this means is that: >1. When under autoplayer >2. When on 450 MHz K2 machines >3. When played against the SSDF competition >Fritz 6.0 has an ELO of between 2629+25 and 2629-24 with a probability of 2/3 >Hence, the ELO must be _at least_ 2605 and for two standard deviations (about >97% probability) it must be 2605-24=2581. I believe that it has been adequately >demonstrated that against this pool of talent, that ELO is "GM" level. If the >same sort of rating can be maintained against human GM's after the same number >of games, then it will have been proven with a doubt of at most 3% that the >program was of GM level against that talent pool. > "I believe" ?? Now you are having to make a judgement call based on some data which seems to satisfy you. You missed the point of my last post. It will ALWAYS come down to a judgemet call. There is NO 100% accuracy or probability. Sooner or later you have to use your judgement to determine the answer. Any mathematical bounds you set up are arbitrary. Any mathematical requirements you set are only in your judgement and not mine and probably not anyone elses. You can only decide for yourself when you think the GM strength mark is attained by any program. YOu will never have proof. It doesn't exist because you can't define it without making a judgement call. >Note Junior 5.0: >With only one standard deviation, the ELO is 2503, and so I would say that this >one is seriously in doubt. With two standard deviations, it is below GM level. >Therefore, our confidence level is fairly low. That having been said, since the >x-bar figure is 2528, we could still say that Junior 5.0 was "probably" playing >at the GM level. Now I am learning about statistics. "With one standard deviation, the ELO is 2503" ?? I thought it means that with one standard deviation as a boundary there is a 68% chance that the ELO calculated will fall within this range. There is just as much chance that it will be on the plus side as the minus side. I would go back to school but I'm too old to start learning all over again. Anyway I'm retired and I don't have to know anything. > >>Because if you do you will be using >>your own "Feelings" or common sense. So you can never know with certainty >>because you are afraid to rely on your own judgement. > >Aristotle relied on his judgement. That's why the earth was the center of the >universe. If you do not have facts, and yet the facts can be produced, and yet >you rely on your judgement, and then have the UNMITIGATED GALL to call that a >proof, then you are just plain wrong. > Why do you keep bring up these idiotic comparisons which have nothing to do with the discussion. You are using inductive logic to prove your point. Where did I say my judgement was Proof? Where did I say I have any proof of anything?Please quote the phrase for me. I'm getting old and may have forgotten already. >>You need a formula to >>tell you what is true but your can't decide what you require of the formula. > >You need math to tell you when it is proven. We may know that it is probably >true. But we don't know that it is CERTAINLY true. To say that it is certainly >true is just a form of lying. > >>You can't decide what the truth is. Is it one standard deviation or is it two? >>Do you need 68% or do you need even 99%. > >At one standard deviation, the confidence is fairly low. Suppose that life or >death were on the line. >Dr Jones: "There is a 2/3 chance that you will live with this procedure. There >is a 97% chance that you will live with the other." The second procedure has >been proven more clearly. Every level of confidence makes the proof more >secure. > >>What is the probability of the GM you >>are comparing against being "on" his game when the computer is played? There >>are many variables and you want an exact answer. Your math is failing you and >>you can't decide for yourself. You can't quantify the problem in exact figures >>so you will never have an answer. > >I will have an answer when it is proven. I will have a guess when it is not. >For someone to say that the guess is an answer is wrong. > >>That is untill you decide to use your own >>brain and judgement and forget about probability. > >Forget the math and go with your feelings then. I will stick to the math. > >>The answer is NOT in math. >>It is in your own human ability to reason. > >If you ignore the math, then you are UNABLE to reason. If the math says that it >is uncertain, then it is uncertain. To then claim that it is certain is >completly wrong. > >> But mathematicians cannot deal with that reality. > >The world is flat. Everyone knows it. Why require some kind of mathematical >proof of anything? Because when it is available, we can believe it. When it is >possible and we ignore it, then we are fools. Don't the people who believed in >a flat earth look like idiots? In retrospect we will probably all look like idiots. But you should understand most people believed the world was flat because that's what they were told by the scientist/mathematicians/educators of their day. The people who could prove beyond a doubt that it was true with some formula they concocted to back up their theory. Remember when Einstein was told by the Astronomers that the Universe was NOT expanding? He then invented a constant to make his theory work. Math is a wonderful thing! It is an exact science! :-) > >What people do not understand is that there is a difference between something >being generally accepted and being proven. The inability to understand this is >a fundamental lack in many educational systems. Apparently, many of the members >of this forum have never been shown the difference. Or, if they were shown, >they did not understand. Now you are showing some hint of understanding the problem but not quite. Between being generally accepted and being proven is where you are stuck. Because you cannot define GM strength exactly you have to either find a generally accepted number that you can apply your formula to or you are lost. Again you have to make a judgement call. The math will not work without that figure which does not exist. So in your case you will never be able to prove that computers are of GM strength untill they are way beyond that and then you can say with reasonable certainty that it has happened. Others here are claiming it has already happened but as you point out, you will not be able to prove it untill it is a certainty. Then in retrospect our ancestors may look back at us and scoff at the idiots that could not recognize a GM when they saw one. :-) Jim P.S. I'm finished. I enjoyed playing with the subject but I have other things to do now.
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.