The last world computer chess champion, Zappa, has been released as by ChessBase in the Fritz 9 interface as "Zap!Chess." (What, is the name in Khoi? Is it pronounced "Zap{CLICK}Chess"?) I interviewed the programmer, Anthony Cozzie, last year after he won the title in a surprising upset over usual suspects Junior and Shredder.
This continues the commendable ChessBase practice of picking up the strongest newcomer engines and putting them into their best-of-breed interface. With so many strong new engines out there (Toga is a Fruit flavor and both are very strong) and more appearing all the time, it makes sense for ChessBase to focus on the interface and Playchess.com features, which require far more resources. As I mentioned here not too long ago, new engines get the hobbyists excited, but they have all been so strong for years now that there is nil difference to 99.9% of the potential customers. You can't tell if you're being pummelled by a 2800-level computer or a 2750. The useful advances will come in the area of making them play more strategic, instructive chess so they can be more interesting opponents and better trainers.
This Zappa announcement's thunder was stolen awhile ago by the engine Rybka, which is the current computer-computer top dog by a wide margin and the darling of the computer-chess fans. (Also in the Ninja message boards.) The still rough-around-the-edges program by IM Vasik Rajlich has dominated recent events and is ubiquitous on the servers.
Wonder if Kramnik has (again?) missed the boat with his planned match against Fritz, considering that there is now Rybka, or for that matter, Hydra in its various incarnations since quite a while.
Kramnik's "last chance of humanity to beat the computer" would be a flash in the pan if he plays against the second best computer.
Nah, nobody really pays that much attention. It's about sponsorship. If the Rybka people could put up a pile of money, great. But "strongest computer" has little meaning when it comes to PC programs playing humans. A gap of even 100 points on a comp-comp list would not be reflected against humans in the same way. Some comps give humans more trouble because of style, but we've known for a while it's more about opening books and blunders than anything else when it comes to humans vs comps. They are all very very good at the same thing, tactics.
The differences between programs come out in the hard-and-fast world of comp-comp chess, but humans don't lose (or win) to computers because of evaluation differences or a single ply here and there. It's about controlling the type of position out of the opening and avoiding tactical messes.
Rybka's huge lead on the comp-comp list isn't meaningless of course. It might indeed perform better against humans, but not to the same degree. Hydra is probably out of this conversation to a degree. The Adams debacle aside, I don't see why it wouldn't be significantly stronger than the micros, enough to be considered stronger against humans.
Hey Mig, remeber that the last Freestyle tournament was won by an unaided Rybka engine. That is, a single Rybka engine was stronger that a lot of GM's playing with engines (that means there aren't blunders).
That kind of trows your theory to the ground.
What theory, and why? Chess is still a sport and this event was just a few games. Many of these GMs were using Rybka themselves. Rybka also beat Rybka in that event... Hardware differences, opening books, etc. all are relevant.
Human assistance can also be human error. This event was played a faster time control so it was much harder for humans using computers to coordinate their efforts.
ceteris paribus, I don't believe that Kramnik's incremental higher rating of 9 ELO points over Adams will compensate for a 5.5-0.5 debacle, it may perhaps result in a 5-1 debacle against the best computer (not Fritz)
John, actually Hydra won this year's Freestyle tournament, a full point ahead of the field. Chessbase simply hasn't reported the finals yet.
I'm still trying to parse "Toga is a fruit flavor and both are very strong." Sounds like Brecht. I believe I will tell this to people I meet today, while holding a placard with the word "Fish" printed on it. Whatever it means it was more fun to read than slash-dot.
So the real question is, when are we going to see Hydra in a world computer championship? Now I doubt it will lose to crafty, but how about against Junior , Shredder, Rybka and now Zappa, running on even half the hardware it takes to run Hydra? That is when the real fun begins.
After much introspection, Toga is indeed seen to be a fruit flavor, and once again the chess world is indebted to Mig for his profundities.
99.9% do not use engines to play against, but to assist their analysis and opening investigations.
I am interested in buying either Rybka or Zappa. Which one would you suggest. They both cost about the same. I will be using it mainly as an aid in analyzing my games I play and when studying the recent GM games to understand what is going on. Presently I am using Hiarcs on PDA (Palm OS Tungsten T3), Deep Fritz 8 and Chess Assistant 7.1 which has Tiger.
Alkelele, that's not true. A vast majority of chess software is purchased, or gifted, for playing. Not by strong players, but most casual players don't even know what analysis and training mean. Those who download relatively unknown programs and collect engines are a tiny minority. Those who are reading this are relatively serious players. But most people who have ChessMaster or Fritz are playing the handicap modes. Software might lead them to become more serious. Fritz and the playing programs outsell ChessBase - the real tool for analysis and opening prep - by a huge margin. And even though my old ChessBase Cafe column was read by people we can assume are fairly serious aficionados, most of the questions were fairly basic ones about playing.
Many strong players would use engines more for playing if they played better; i.e., more like GMs and less like tactical monsters that will happily take on pawn weaknesses and bad minors only to crush you anyway. It's decent training for your tactics but it's depressing and it makes you gun shy. Just to survive you start playing anti-computer stuff and avoiding all open positions, hardly a good training habit.
As for what to buy, getting the Fritz 9 interface with Playchess.com access and all the training and analysis features is hard to pass up. If you already had another Fritz 9 generation program you could just get Rybka and plug it into that. From informal results on the servers, I don't doubt that Rybka is an overall stronger engine in a statistical sense, though no one has much info about Zappa right now. I also use engines to analyze and will be interested to run Zappa through my personal test suite to examine its evaluations. But as I think I've made clear, I think interface is far more important for just about everyone.
To the guy who posted using the handle "gg" and "tgg" in various recent comp-related threads: Are you listening?
Not that I expect Mig's opinion will carry any more weight with you than did mine, or the author of a published book I quoted. You have already stated your disdain both for IMs' analysis, and for all "received opinions" (i.e. of the experts we all study in hope of imbibing their knowledge and skills, and/or benefiting from their experience).
Is it possible that Rybka is actually better at evaluating most positions than human Grandmasters? It seems like it could. Perhaps it really can determine the type of positions where activity outbalances bad pawns or optically bad minors with a higher level of sensitivity than humans. If programs like Rybka can demonstrate so many exceptions to standard rules of positional play, then human play could develop in a more Watsonian rule-independent and creative way in the future. Every elite GM a Topalov or Morozevich :)
Jon,
Oh, now you have Mig to cite, rather than ever having an opinion of your own. Although what Mig says here is not relevant.
IM commentary is confused and contradictory on ICC - I do not need a book to tell me that. I read what they say. Before you excused them by saying they are looking at many games or don't have time or whatever. Well, Fritz does not need excuses.
I never posted as tgg, maybe other people posted as gg that's not my business.
Andrew Dimond is worth quoting though. I like his opinion. Computer evaluation is better because they figure out how to win. Humans underestimate tactics because they can't do it - that is a weakness in human evaluation. Not a moral weakness in computers because they are mean enough to calculate variations, like that was cheating.
Does the Computer Chess Championships not regulate the hardware platform in any way? (The relevant links are broken on their home page.) That seems to be a big inequalizer between competitors. Does anyone here know what a "typical" competing system looks like, and what proportion of teams run on a laptop versus a desktop versus remotely on some giant server grid?
Thanks.
Mike Parsons, actually Hydra has been in several computer championships. It won the Paderborn Computer Chess Tournament in 2004 and 2005 with much ease and undefeated. The Hydra team did not play in this year's Paderborn, which was won by Rybka, but considering that Hydra beat won this year's PAL/CSS Freestyle Tournament, it's safe to say Hydra is still top dog, although it must be said that Rybka has indeed made a very strong introduction and is perhaps the best 'engine' ever created.
Hydra won first place, allright, but Rybka won 4 places out of the top 8, wow
There have been posted lots of unpleasant remarks about Zappa!Chess at the 'talkchess' forum. They say the new product after having been converted from the UCI format to the internal chessbase format produce lots of bugs like sudden system crashes, checking when there is no actual legal check, lots of illegal moves, etc. In other words, Chessbase made hasty to set the product on the market in the face of Rybka dominance without much re-engineering, thorough testing, etc.
For those thinking about buying Zap!Chess, I'd wait a while. I got it three days ago, and it works halfway on my XP computers, and not at all on my Windows ME machine. It will install, open, and sometimes play a bit, then I get a memory leak (I think) problem, where it says Chessprogram 9 has to close, if you keep having problems restart. Of course, I restarted about 20 times, and it still won't play, in fact I've not managed to play a whole game. I had similar problems on my XP machines, though not as frequent, and I at least managed to play a game on it.
I tried plucking engines from the folder and dropping them straight into the engine folder. This didn't work either, as it asks for the CD then doesn't recognize it. Of course I installed the update from Chessbase (which by the way isn't automatic on the "Query Upgrade" feature, but has to be downloaded separately from Chessbase), but nothing has worked. I'm trying installing minimal options too, and I doubt it will work either.
I've not had this problem with either Shredder 9, or Fritz 8. I don't know if I've got a bad CD, and I'd love to hear if others have had this problem.