From: Larry Kaufman comcast net> Date: 21 mar 2007 Subject: Re: Bonanza vs Ryu-O Champion ----- Original Message -----=20 From: "Mark Schreiber" hotmail com> To: topica com> Sent: Wednesday, March 21, 2007 11:14 AM Subject: RE: Bonanza vs Ryu-O Champion Hi Larry Kaufman, If it's a desire to profit then has the nsr setup a professional fee structure to play computers? The prohibition was made after a pro almost lost. Kazuyoshi Nishimura said if 1 pro loses, it could create the perception that pro players are weaker than the software. It's clear the prohibition is because they are afraid of losing. If NSR didn't want pros playing computers, this match wouldn't be=20 happening now. I think they just didn't want pros playing computers without= =20 NSR getting a cut of any fees. It is unfortunate we don't have a shogi programmer here to give their opinion about the usefulness of computer vs titleholder without a handicap. Can you give details about the handicap matches between computers and the top pros with mixed results? I don't have the details, but I was told that a big factor was=20 hardware - the program tended to win with fast hardware and to lose with= =20 slower hardware. Apparently, it is now running on much faster hardware than= =20 before (probably a quad or octal), that probably accounts for the estimated= =20 2800 rating vs. 2640 actual rating on sc24. From my experience with chess= =20 programs, I can assure you that such estimated rating changes based on=20 faster hardware are quite reliable. So there is little doubt that Bonanza on the super-fast hardware can= =20 compete with at least mid-level pros in fast play (30" per move, the usual= =20 time limit on sc24). However the slower the time limit, the better for the= =20 human player. I didn't hear what the time limit is for this match, but I'm= =20 sure it's much slower than 30" per move, so probably the title-holder=20 remains a huge favorite. Have titleholder pros lost to 4 dan pros with a bishop handicap? There was a series of ten games around 1990 in which the top ten pros= =20 played ten new 4 Dan pros at bishop handicap. All the 4 Dan pros won except= =20 the one who played Oyama. There have been a couple other instances of pros= =20 losing when receiving bishop handicap -- Senzaki lost twice taking bishop= =20 handicap from Kimura a couple years ago, but these were 10" per move games,= =20 and Hanamura 9 Dan was known to defeat other pros giving them bishop at=20 times. Also the late Koike, the top Amateur who qualified by strength to be= =20 a pro but was turned down for ethical reasons, lost a game taking bishop= =20 handicap from the great Masuda 9 Dan. I am a confused. What are levels between amateur 6 dan and pro 4 dan? They are separate scales. Roughly speaking, I consider amateur 6 dan= =20 and pro 1 Dan to be about equal on average, but amateur 6 Dan is a much=20 wider rank. Some players of that rank are not terribly strong, maybe only= =20 around pro 3 kyu or so, while others are definitely strong enough to fight= =20 on even terms with a typical pro 4 Dan. The amateur ranks are not well=20 defined. By the way, the pro ranks are only meaningful up to 4 Dan. Beyond= =20 that, promotion is pretty much just a matter of age, as it's based on how= =20 many games you win. The division in which a pro plays has far more meaning= =20 than his rank. Mark Larry Larry Kaufman wrote: > > > ----- Original Message -----=20 > From: "Mark Schreiber" hotmail com> > To: topica com> > Sent: Tuesday, March 20, 2007 4:30 AM > Subject: RE: Bonanza vs Ryu-O Champion > > > > To Bernhard, Adam and Everyone, > > The conspiracy theory is not far fetched. Remember the NSR is > paranoid, they forbid pros from playing computers. > > This is not paranoia, it is simply a desire to profit from the > situation. > > Computers are allowed in the Ryu-O tournament. I remember the > computer program Gekisashi won 1 game in the 2005 Ryu-O tournament. Is a > computer playing in this year's Ryu-O tournament? It is interesting to > see if it does better this year. > > The bonanza hardware will not matter when it plays Watanabe? Even if > it uses a supercomputer it will lose. Today's programs are just too > weak. They play at about amateur 6 dan. > > I don't think the programmer can learn anything from playing a pro > that is so much stronger. It's the same if you or I were to play > Watanabe. > > Let's not exaggerate. The difference in strength is enough to predic= t > the winner, but not so great that nothing can be learned. > > If a computer plays someone of equal strength then some > strategies will fail and some strategies will succeed. The programmer > can see if his changes cause a failure or it causes a success. But if > the computer or anyone plays a pro that is 12 levels stronger, then > every move leads to failure. The programmer will never see his changes > make a difference. Even if the computer makes better moves that a weak > pro would make they will also fail. The programmer will never learn > anything from playing a player so much stronger because all moves lead > to failure. > > I don't think Bonanza playing a 9 dan pro with a handicap makes any > sense. The handicap would have to be tremendous. > > Actually the handicap in recent matches between computers and the to= p > pros has been bishop, with mixed results. I wouldn't call that > "tremendous"; > pros have been known to beat other pros at bishop handicap. Basically I > think you are overstating the difference between a top pro and a top > amateur. > > Better for Bonanza to > play a player with a similar strength, a 6 dan amateur without a > handicap. > > Maybe better, but both are interesting. > > Mark > > Larry Kaufman > > Bernhard M=E4rz wrote: > > > > > > Hi, > > > > > Bonanza will be slaughter. Can we learn anything from this? I like to= > > > see more computer vs man matches. But make them balanced. Shouldn't > > > Bonanza be playing a 6 or maybe 7 dan? > > > > I believe, WE (!!!) can not learn much. > > > > But I assume, the programmers of Shogi programs can learn from such > > games to improve there programs. > > > > Also (as you said) it could be interesting so see more computer vs man > > matches. > > Nevertheless it's not so important for me, to watch such a game. I like= > > to know the strength of computers (for this reason results against 6 da= n > > > > humans would be interesting) and like to play against it for myself. > > > > On the other side I would be more interested in handicap games, means > > make the games balanced and give Bonanza a 50% chance against 9 dan > > pros. > > > > Bernhard > > Adam wrote: > > > > Hallo Mark, Larry and all, > > > > do You really think that such conspiration theories are true, Mark? :) > > Psychological value of Bonanzas slaughter by Watanabe RyuOh for pros > > is undeniable. But isn't it a bit childish? Maybe NSR wish to > > publicaly demonstrate difference in strength and thus help in > > development of better shogi playing programs. Advancements in CS is > > unstoppable, and unlike human minds computers simply have "no" limits. > > So time when computer beat up pro will come. THis raise (imho!) an > > interesting question: RuyOh tournament is open even to amateur > > players. What about computers? Are they allowed too? How will NSR > > react to possibility of computer become 42th ryuoh... > > > > As for upcomming match. Do you know any configuration details? Will it > > be some special hardware/bonanza version playing with Watanabe RyuOh? > > Or just regular "Thoshiba notebook" and "bonanza from web". > > > > Cheers, > > Adam Skalny > > > > Quoting Mark Schreiber hotmail com>: > > > > > > > > Larry, > > > Thanks for the ratings clarification. I have a theory on why this= > > > match was arranged. > > > > > > I can't imagine the Bonanza programmer will want his program to b= e > > > clobbered. The Japan shogi association arranged this match. With all= =20 > > > the > > > talk about how strong the computer will be some pros might be getting= > > > scared. This match will give the weaker pros pointers on how to beat= =20 > > > the > > > computer. The match is also to give the pros a boost of confidence. I= f > > > they are more confident then the Pros will play better against the > > > computer. It is also to show that shogi is so much harder than wester= n > > > chess that it is in a different class then western chess. > > > > > > I think it will take about 20 years, 2027, when a computer can= =20 > > > beat > > > the best shogi pro. So the pros have time before they need to worry. > > > Mark > > > > > > Larry Kaufman wrote: > > >> > > >> It seems strange to me too. Bonanza is roughly on a par with to= p > > >> level > > >> amateur players, but such players would probably only have a 5%=20 > > >> chance > > >> at > > >> best against a pro titleholder. So unless Bonanza has been improved > > >> greatly, > > >> it's just a publicity stunt. > > >> Incidentally, Bonanza is considered to be on a par with Amateur= =20 > > >> 6 > > >> Dan > > >> players, not Pro 6 Dan players, so it's even more unreasonable than > > >> your > > >> > > >> question implies. > > >> By the way, the final score of the Rybka-Ehlvest pawn handicap > > >> chess > > >> match was 5.5-2.5 for Rybka (giving a pawn). > > >> > > >> Larry Kaufman > > >> > > >> ----- Original Message ----- > > >> From: "Mark Schreiber" hotmail com> > > >> To: topica com> > > >> Sent: Friday, March 16, 2007 12:56 PM > > >> Subject: Bonanza vs Ryu-O Champion > > >> > > >> > > >> > > > >> > Talking about handicap systems to level the playing field. If > > >> > Bonanza is maybe a 6 dan why is it playing a 9 dan? 9 dan Ryu-O > > >> > Champion > > >> > Watanabe said, "I don't think I'll lose". Then he said, "A loss ca= n > > >> > never be permitted". Does anyone have any doubt what will be the > > >> > result? > > >> > Bonanza will be slaughter. Can we learn anything from this? I like= =20 > > >> > to > > >> > see more computer vs man matches. But make them balanced. Shouldn'= t > > >> > Bonanza be playing a 6 or maybe 7 dan? > > >> > Mark > > >> > > > >> > Larry Kaufman wrote: > > >> >> By a curious twist of fate, just as computer shogi programs= =20 > > >> >> are > > >> >> reaching > > >> >> the level where they can compete with pros without a handica= p > > >> >> (as > > >> >> planned for Bonanza shortly), in chess the opposite has just= > > >> >> occurred. A > > >> >> match is now taking place in which computer chess program > > >> >> "Rybka" is > > >> >> giving a pawn handicap to a renowned grandmaster (Jaan=20 > > >> >> Ehlvest) > > >> >> in > > >> >> an > > >> >> eight game match, and the computer is leading 3-0! > > >> >> You can either take this as evidence of the superiority of > > >> >> shogi > > >> >> over > > >> >> chess, or as a forecast of things to come in shogi. I think > > >> >> it's not > > >> >> so > > >> >> many years away that a shogi program will spot the Meijin a > > >> >> lance > > >> >> and > > >> >> defeat him. Few could have imagined this would happen in=20 > > >> >> chess, > > >> >> but > > >> >> it > > >> >> has, and shogi will be next. > > >> >> > > >> >> Larry Kaufman, chess International Master and Shogi amateur = 5 > > >> >> Dan. > > > > > Mark 2 --^---------------------------------------------------------------- This email was sent to: = shogi-l shogi net EASY UNSUBSCRIBE click here: http://topica.com/u/?a2i6Ys.= aB5TiY.= c2hvZ2kt Or send an email to: shogi-unsubscribe topica com For Topica's complete suite of email marketing solutions visit: http://www.topica.com/?p=3DTEXFOOTER --^----------------------------------------------------------------