<teamrybka> interesting comments but I am not sure you read what I actually wrote. I did not criticize Aquarium trees, which are fine. I discussed the limited datbase functions like not being able to sort lists or have CA-style classifiers. It is CA 10 that broke the trees, not Aquarium. I hope CA 11 has the bug fixed.

As for tests, I disagree. Tests measure knowledge, not understanding. The problem with computer analysis is that it explains nothing, just indicates superior moves. So it can't make you a better player. My Ph.D. had little to do with exams. It was all about my dissertation.

Finally, being published has little to do with connections. Might help the first time, but publishers only come back for more if the books sell. The greatest validation an author can have is to be asked to write another book. If the readers are not happy, careers will be vary short indeed. And money only helps with vanity self-publishing. Publishers are not easily bribed.

Having spoken too or emailed both eric and tim, I can speak here. Tim said Rybka 4 beat Toby-tal but not Rybka 3. Eric said chess assistant 10 sucks and aquarium sucks in the opening. I must have a special version, as mine works fine and aquarium trees work everytime. I will say I use all the databases and so do top GM's.

Personally, I feel ratings mean very little in regards to expertise, as I favor standardized tests to measure knowledge. I like the solataire by pandolfini and the convekta testing software. Because someone is rated high does not prove experience or why you should listen to them. As far as ability goes, I feel test are easier to measure results with.  Eric should know this, having gotten a PHD !

I can get a draw with rybka 3 on human or dynamic and full strength, but feel humble enough to want to keep improving. I have a consistent 2590 rating and beat fritz 11. I also know I get a precious 25 provisional games to secure a high rating with, so prepare now.

It is better to test and simulate tournament conditions. Many players lost rating chances by getting rated too early. You can take the same test, and pinpoint what to work on. Asking some GM a question is no guarantee to get a correct answer. what if he never played that opening line, position, ending, etc; how then can he be "the fountain of misinformation"? LOL

I bet if he wins in the opening or middlegame, he knows little about endings, as he never needs it. You can break down players by "taking them out of their element". We call it "transposition". On the GM circuit, most knew Kasparov was a change up artist and could steer clear of pet variations. He put work into openings, with the idea of the middlegame or ending in mind.

I have beaten strong players and masters too, just by knowing how to transpose to lines or even openings that exploit certain skillsets.

Chessbase and aquarium are the two best for database work. Eric does seem to have a ghost writer.

I personally trained under his former boss, jeremy's boss too ! FM Ken Smith, who I used to order from at chess digest. Ken said eric and jeremy were both Ghost writers of his and wrote for him. I think people change over time. Most of you don't know in russia it is a saying "chess champions are MADE , not born". It should say PROGRAMMED/trained.

Botvinnik felt the PC was the best tool to train champions and for analysis work, along with using blitz for training games.

I feel like some FM or IM are better at teaching or coaching than a GM. I want to see chess teaching/coaching change. How much of a teacher education degree does eric have? Wasn't it linguistics? It is common knowledge that anyone with a masters or above can be faculty and yet never have taken specialized teaching courses. The head of the psych dept where my girlfriend got her psych degree from, had a woman with a zoology degree teach psych classes and openly admitted she only had gen psych on her resume; as all her training was in ZOOLOGY. Yeah really,lol, go figure on that one (blind leading the blind).

what I'm saying is anyone can write in america and with connections or money... get published, yet not prove expertise. I think eric and jeremy are good teachers, as I own a ton of their books and tell students to read it ! I'm just saying don't over rely on ratings, but search out "hard work" for people who can prove things and show evidence. When we all take the same test under the same test conditions, it shows who is more knowledgeable. There are thousands of ways to win or lose a game, and eric or jeremy should "man-up" and say so. Winning a tournament game is not too hard, if you know how. It is a matter of knowing how to apply knowledge and adapt to changes.

I would tell you that some GM's use FM,IM, or lower to be seconds and do database work, engine work,etc, what we insiders call "grunt" work, the "gopher" job. I can hire someone to operate a PC if I tell/show what I want done. They don't have to be expert or even know much chess. Seconds are usually doing testing and prep/research work for a GM. They sort patterns in chessbase or use classifiers in chess assistant, or run IDeA in aquarium and include/exclude spec moves the GM prefers or... the engines selected, or...do "the daily", aka the daily list of assignments for ongoing projects a GM wants done. S/he may want an opening TN made or checked, or chinks in an opponents game exploited or just found ! Most of you don't know this.

This is why you can use "presets" on engine work, and save settings. It is because a GM will run a players game,moves,ideas,etc. through the engine until the engine comes close to that choice/style. Now he knows which engine and settings to use to anticipate opponents moves. He can do engine matches to find engines or settings that trump it, thus you have so much variety. A GM said once...that just seeing the analysis can reveal the engine and settings! That is an insider secret for the real reason informants are sought by GM's, to see what ideas the competition has, or the engine work. lol

Today, it is about money, plain and simple, and huge amounts of time for chess. Money for books,videos,computer cd/dvd (software),etc. Then you need time to study,practice,drill,etc in addition to memory work. The trophy loses it's meaning. In other countries, they train and study first, then do tournaments. Here it is a money making gimmic, as well as mislabeling players too low. It is better to test/prep first.Think of a tournament as final exams. How can you pass if you never took the class? Caught in an opening and position goes downhill. Screw up the endgame and lose. unfamiliar with the middlegame or specs of tactics,combo,positional play, you let it slip away. Chessbase can pinpoint patterns and show what goes together. It can tell you the endings,strategy,etc. if you save openings in a seperate "database"( aka chessbase folder).

The old people may have done it by hand and believed you had to play lots of games...but we use high tech methods now, and take years off the learning curve. Even SCID has the chessbase feature "time on square", so you can quickly come up to speed on any phase of the game in any unfamiliar opening/midlegame/ending. You can use the "move to" feature to see where they go most often. Both of these skills used to take years to develop.

You can select a group of endings and see where certain pieces go as well as how long they stay put. Eric or jeremy may remember years ago this was done by hand with pen and paper or a stack of index cards, as well as a filing cabinet in russia; where games were sorted and indexed by hand.

Now anyone with time,money,intelligence,motivation,desire... can be champion; if you have enough resources and study/train/drill/understand enough chess knowledge.

Is there a free version of TobyTal?  I'm just a poor student.

ok

Interesting article, FM Schiller, thanks. But as I pointed out at least once before, your articles would benefit from closer proofreading, e.g. Ribka > Rybka, our > are.

I can't help thinking that someone is "ghostwriting" these articles you publish here. They sometimes seem hastily or perfunctorily written.

Another problem I have with TobyTal is that it has not been entered in any of the engine shootouts/tournaments.  I would think that Tobiason would long ago have entered TobyTal in such a tournament in order to validate his claims.

If I want a super-strong engine, I can get Robbolito for free.  If I want an excellent engine with a great UI and helpful study/analysis features, I can get Fritz. 

As a software developer, some of Tobiason's comments make me think he doesn't know anything about computer science:

1. Chess will be solved in 4 - 6 years by tablebases?  Since there are more possible chess positions than there are atoms in the universe, explain to me how someone is going to build that database!  Not to mention, it has taken years to expand Nalimov tables from 5 pieces to 7 piece postions.  This is because each time you add a piece, the number of possible resulting positions increases exponentially.  It has taken several years to get from 5 to 7 pieces, so how are we going to get from 7 to 32 in the same period of time?

2. He has stated that he has encrypted his code so that it is impossible to decompile.  This ignores the fact that the program must decompile itself on start up in order for an operating system to run it.  This means, in turn, that the encryption key must be embedded in the program.  It is child's play for a comp sci grad student with a debugger to find the encryption key, and then use the key to generate a non-encrypted image of the engine.

If Tobiason does not understand such basics of software development, I am very reluctant to trust anything he says about his own product or anyone else's.

Note also that whatever process is used to encrypt the executable file can also compress it.  (Think WinZip with a password.)  That may account for the small file size that Schiller has noted.  Thus a comparison between TobyTal and any other program based on file size alone is entirely invalid.  You'd have to compare the working set of TobyTal and the other engine at runtime to have a legitimate comparison of how each uses resources.  Eric, have you compared the working set of TobyTal and other engines, or are you relying on file size by itself?

I think that it is probably possible to make computer programs that would choose their openings based on a set of criteria such as time control, opponent repertoire, match situation, previous results with an opening etc. But of course they don't have subjective preferences that humans do, as they are not humans.

Also, I believe that programs have done quite well at Chess 960 ?

having studied a bit of artificial intelligence and algorithms, i once made a program that would play tic tac toe repeatedly and learn from the results. in just a few games, it was not possible to beat the computer.

i've always wondered if a computer is programmed to do this with chess, no openings or anything, just playing legal moves and learning from the games results, would "discover" the sicilian or the king's indian :)

very interesting topic. By the way, I use Crafty to analyse games and has worked well for my amateur-ish use.

the database i use to store games and analysis is SCID, a very handy tool. it has a very complete system for searching through games, and tools for tracking ratings of players over time, winning percentages of specific lines and such information.

both are open source and free. i use them on linux, i think both can be run on windows but i'm not sure.

@manonel He is not the programmer. So his rating is irrelevant. You have to look at the product features and decide.

This is an excerpt from Toby's website about tobychess

The last lines are precious...in four to six years all engines will just be obsolete? huh? Even when chess is "solved", there will still be a use for engines....

and as a last preface --Yes, there are two #7's :D  not my error....

7. Our engine is 32 bit bit but plays well vs the 64 bit versions of Robo and Rybka.

7. In the end if the Robo people wish to resolve this in court, we will happily submit our code to the court for 3rd party comparison under the courts supervision to ensure our code remains secret.

	I am selling our engine on-line and at tournaments. At the North American Open in Las Vegas, we sold 21 TobyTal's while Chess Palace told me they sold 5 Fritz 12's and 4 Rybka's. People could see the engine matches and install them on their own computer and test them themselves.

	I do not sell the engine online as a download. We only sell it with the Deep tactics interface and to prevent piracy, future versions will be tied to the interface like Fritz 12.

	The issues will be settled on the chessboard in the coming months. We will add tablebase support, new chess knowledge and make a windows XP version in the next few weeks and will hopefully fix the bug in the current version. I will send this to all of you as a free upgrade if you have already purchased TobyTal [Thanks for the business]. After that we will create multi core and 64 bit versions with opening learning and multi computer capabilities. This will be sold as Deep TobyTal for $89 or an additional $30 upgrade. 

	In 4-6 years, there ill be no more chess engines as chess will be solved and the moves will simply be looked up in tables (similar to the dinosaur methods of tablebases used today). Then all these chess engine arguments will be obsolete.

Toby

@darkphobos: Humans choose which moves to include with their repertoires, Kasparov did not always follow the moves suggested by seconds. Most famously, in his "revenge" game against Kiril Georgiev at the 1988 Olympiad, Nikitin arrived with his preparation and Garry threw it on the floor, saying he was just going to crush Georgiev, no opening special needed (Not his exact words, this is a family oriented site! I was reminded of the "We don't need no stinkin' badges" scene from Blazing Saddles. I witnessed this, as did several GMs as we were having a political meeting at the time.

Humans select opening moves, with knowledge of history and statistics. Psychological factors are involved.

That's why it is simply not true that computers have defeated top grandmasters. Turn off their opening books and they usually get clobbered. The computer engines are simply cheating by using these opening books. I await the day when the computer programs will construct their own opening books without any human input and come up with something reasonable.

The same things could be said of human grandmasters. I await the day when human GMs construct their own opening books without any outside human input and come up with something reasonable.

Humans learn openings from opening books, periodicals such as Informant, and especially by replaying the games of other players. When 2 GMs play 20 moves of theory in a popular opening variation, how many of those moves were developed by the players themselves? Typically none of them--all the moves were discovered and tested by other players and memorized by the GM.

Opening theory is the collective creation of thousands of players over many years. Cut off a player from all that information and there will be an enormous loss of effectiveness. It doesn't matter if the player is human or silicon.

The real differences here between human and computer are twofold:

1. Computers have better memories although top GMs seem to do really well in this regard. If you want to make an issue of silicon memories outperforming organic memories, it's certainly true. An unquestionable advantage that computers have. We could even that up by changing the rules to allow the human player to consult an electronic opening book. Would that help a top GM much?

2. Humans have what I will call the "library research" skill. The ability to sift through many sources and paste them together into a coherent chess repertoire--or a scholarly paper on the Crimean War. Chess computers cannot do this yet and rely on their human creators to do it for them. Human GMs can do this although some of them seem to rely heavily on their seconds. Furthermore the creation of opening books is not a cutting edge chess skill. You do not need to be a GM to do a good job of going through the literature and putting together an opening repertoire that can be used to create an opening book. You need to be thorough and organized, you need a good but not expert understanding of chess, and it really helps to have a computer (!) for blunderchecking. Many outstanding books by IMs and below prove this to be true.

Karpov has Zaitsev. Rybka has Noomen. With all due respect to Mr. Noomen, who got the better deal?

@meniscus That rumor is totally false. Robolito is bloated compared to TobyTal, the code is totally different. Rybka is multiple megs. Programmers usually claim that their work is stolen by all competitors. In software you learn to ignore such rants.

Do you know how to setup robbolito? I downloaded the files but I can't seem to do anything with them.

Very interesting read.  I was not aware that the computer programs would be horrible without the human-fed opening books.

edit- i heard tobytal was just rybka updated, but that it was dismantled illegally. could be a rumor only tho