On Thu, 2008-08-28 at 17:00 -0400, Robert Waite wrote: > * If my ego were hurt by the fact that Mogo scales better, I > * could easily construct a theory that explained it away. This is what we > * tend to do when we don't want to believe something. That's what I > > * think is being done with the argument that improvement against computers > * doesn't translate to improvement against humans. Sort of a false > * modesty - when I beat a much stronger player once I made excuses for > > * him, my brain was not ready to fully accept the win. > > Is this a stab at me? ; )
Of course not. I'm talking about chauvinism in general - chauvinism is a kind of prejudiced belief in the superiority of his or her own kind and in this case our kind is the human race. We find it strangely "uncomfortable" to believe that computers really can start competing with us. > > I think my ego being hurt is less of a factor in my opinion than you > say but we tentatively agreed to disagree so I won't bring that whole > > mess back up. I think there is evidence and statistics that can be > interpreted in different ways by different people. I wouldn't think that > Occam's Razor really can be applied here to make your solution the winner. It's funny to me that whenever I've brought Occam's Razor up to anybody about any thing, they don't believe it applies in their special case. But it's not a rule anyway - it's just a guiding principle and nothing more. Occam's Razor doesn't make me right and I never claimed it does, it just keeps me honest, I don't entertain every possible explanation that might happen to appeal to me aesthetically. Give me some evidence that it's really more complicated than this (not a hunch or superstition) and I very well might change my mind. > > "All other things being equal, the simplest solution is the best." > I don't know that all other things are equal in this case.. > and it sort of seems that Occam's Razor could be used to squash opposing > > opinions a little prematurely. For example.. if we were back in the 1200's, > many people would say that the Earth is flat because it looks flat and if it > were not flat.. you would fall off if you went to the other side. Given the > > available information at the time.. a theory of gravity and other various > applicable theories would be quite a bit more complicated than the flat earth > model.. and would probably be rejected since the simplest solution with the > > evidence at the time would point to the Earth being flat. And rightly so. With no evidence whatsoever, a good scientist would have made flat earth their default assumption but would always remain open-minded. > * I'm willing to be proved wrong but we all know how > * difficult it is to get reliable evidence from human played games. It > * really would require thousands if not millions of fairly played human > > * computer games at a huge variety of controlled levels and conditions to > * make sense of this. > > That could be true if you are trying to prove with great confidence or > to get an accurate model of the scalability... but I think a decent answer > > will be given with a much smaller sample of games. If you got a group of > 5 pros... and they each played 10 games vs. the computer at various > strengths.. you could quite > possibly get data that would let you know the general scalability.. it > wouldn't > > be the final model that works in all cases.. but it would at least give the > go world an idea of how powerful these programs are when scaled against > humans. > If the computer did very well... I think that alone would wake the human go > playing > > world up... and they would not be so quick to dismiss computer go. The evidence would be interpreted subjectively. You really need pretty airtight evidence or more of the same happens, we believe what appeals to us the most. Haven't you seen this? In the match played 2 or 3 weeks ago we heard a lot of excuses for Kim, but not the computer. People had the impression that Kim was experimenting, not playing up to full strength, etc. Someone did come along to correct that notion but you saw the spin machine in motion. I personally don't have anything against doing that - I do it too. I'm only observing that it's a phenomenon. Watch 2 presidential candidates have a debate, then watch which people say the democratic guy won, and which say the republican guy won. Guess what? The democrats always think the democratic guy won and the same for the republicans. I don't think they are lying - I think they each believe they saw a win. I saw Deep Blue beat Kasparov and I believe Kasparov was the better player and I can give you lots of plausible reasons why he lost and why it was unfair. I couldn't do that very effectively if they played a lot more game and Kasparov still lost. I agree with you that if the effect is overwhelming, you might get by with less data. What can happen however is that a conclusion is reached before it should have been. Happens in science all the time - the standard accepted theory sometimes gets cut and pasted from some original paper written 50 years ago that was just believed without enough of a real test. > > Certainly.. if we were able to scale up to the level that is presumed to be > required to surpass all humans... 10 games at this level.. with the computer > > obliterating the human every time would be reasonable evidence that > scalability > will own humans. > > Kind of like with Chess.. there weren't a terribly large number of games > between > the blue machines and Kasparov. It felt to me that the world in general > accepted > > that computers dominate humans in chess after those two matches. That is > debatable.. > but I think many people's opinions were changed just by those two matches. I wrote what I did before I finished reading your words and we definitely agree on this. In the same sense that nature abhors a vacuum, I think ELO abhors wild intransitivities. It would be difficult to find natural ones although you can artificially create them. It's hard to be 500 ELO stronger than Joe, and not also be significantly stronger than Fred who wins 50% against Joe. When such wild transitivities exist, there must be some horrible imbalances in whatever skills one or both players possess (such as Joe who has an irresistible compulsion to resign against players whose names start with the letter H.) All of this remains speculation of course. I merely appeal to the principle of Occam's Razor which says not to gravitate towards the more convoluted explanation until you see that it actually explains things better. - Don > _______________________________________________ > computer-go mailing list > [email protected] > http://www.computer-go.org/mailman/listinfo/computer-go/ _______________________________________________ computer-go mailing list [email protected] http://www.computer-go.org/mailman/listinfo/computer-go/
