World Computer Chess Championship ?

[IanO]

And yet, it is a competition that has been held 19 times, starting in 1974. And at one time, the term "World Computer Chess Championship" was applicable. I do not find it pretending on the part of the organizers to call it the World Championship.

However, I do think they need to see the trend (lower participation) and adjust to the times. The field of authors is much larger and diverse than it once was. Many can not afford to travel to China or to Japan (were these legitimate locations for WCCC tournaments or vacations?).

Completely legitimate! Remember that the chess events are only one small part of the Computer Olympiad, an academic conference with a dozen game championships. Arguably, the Go, Shogi, and XiangQi events are more important in from an academic standpoint. There is more active research in Monte Carlo Search these days than the played-out chess arena. (Remember, the common belief is that chess is solved because of Deep Blue's victory fifteen years ago!)

Even holding the event in the Netherlands, while great for Dutch participants, is not a great locale for competitors who live much farther away. Instead of expecting the authors to come to the ICGA, the ICGA needs to come to the authors to increase interest/participation. By that, I mean entice more authors to interact with the ICGA. In terms of the WCCC, I think that there should be satellite locations in addition to the main tournament location. For example (not to be American-centric), the possibility of a satellite location in the US may have enticed several American authors to participate. Also, having two divisions, one for original programs and one open to all programs, may increase interest.

My suggestions may have little merit, but it is obvious that the title "World Computer Chess Champion" has much less import in the aftermath of the Rybka affair. I have to sort of agree with you, Marco. The title is baseless at the moment. The ICGA needs to do some work to increase the interest of computer chess programmers. It would be sad to see it become totally irrelevant in the computer chess community, given its origins and its role in the history of computer chess.

A modern world championship really has to be an online event in order to attract the top competitors, amateurs without a travel budget.

[Adam Hair]

And yet, it is a competition that has been held 19 times, starting in 1974. And at one time, the term "World Computer Chess Championship" was applicable. I do not find it pretending on the part of the organizers to call it the World Championship.

However, I do think they need to see the trend (lower participation) and adjust to the times. The field of authors is much larger and diverse than it once was. Many can not afford to travel to China or to Japan (were these legitimate locations for WCCC tournaments or vacations?).

Completely legitimate! Remember that the chess events are only one small part of the Computer Olympiad, an academic conference with a dozen game championships. Arguably, the Go, Shogi, and XiangQi events are more important in from an academic standpoint. There is more active research in Monte Carlo Search these days than the played-out chess arena. (Remember, the common belief is that chess is solved because of Deep Blue's victory fifteen years ago!)

It was not clear after an admittedly cursory search that the WCCC is held during the Computer Olympiad.

[Peter Skinner]

I was not involved in this illegal RE job you mention. All the information I used for creating Houdini is publicly and freely available to anyone looking for it. The reason Houdini is #1 since 18 months is entirely through my own ideas and improvements, not via any RE of other engines.

Have you ever considered that publishing the result of an illegal RE job (which is how many people think about the creation of Ippolit) might also imply that such information is not "publicly and freely available to anyone" since the original rights owner (which would be Vasik Rajlich in this case) does not lose his rights through the illegal action?

These sources have now effectively become public domain, ALL top engine authors use them: Houdini, Critter, Komodo and even Stockfish (see the latest pawn shelter/storm code)!
There's nothing I can do to revert this situation.

With your reply to Ed you show the same behaviour as many times before: you avoid to stay on topic when someone mentions a critical issue about your engine, and switch to standard replies instead. The reader of Ed's post, including yourself, saw his statement "the source code you started with is the result of an illegal and punishable RE job", and usually one would have expected a reply close to that specific point but you replied that you were "not involved in this illegal RE job" which is of course a non-topic and turns your whole reply into an empty statement.

Sven

I really don't see what it is exactly that you want me to say.
Again:
- The reason Houdini is #1 is because of original work, not because of RE of other engines.
- The allegedly illegal actions that resulted in these sources do not constitute any justification for the RE of Houdini.

Robert

I believe what most are trying to get across here is Houdini wouldn't exist without the initial RE of Rybka. That source was taken, improved, and we now have Houdini. That still doesn't exclude the main issue of Houdini starting from said RE source. Without that single event, you would have had nothing to improve upon. Regardless of the now "public domain" consideration, the base code was not "public domain" at the time.

Do I believe Houdini has a place in future tournaments? No, but because of it's origins. That doesn't mean I don't appreciate the level of competition it has created. As one would say: "You can polish a turd as much as you want, but it is still a turd."

You have polished a turd to an amazing degree, and deserve credit in making it the best turd out there currently, but it is still a turd.

Peter

[Adam Hair]

I would also prefer that the ICGA call the current event the "Computer Chess Olympiad", as it is not really a world championship any longer. In its current quaint format (moves made by people on a physical board!) it is suitable as a social event for academics, just like the other Computer Olympiad events, and the oldies tournaments for dedicated computer enthusiasts.

However, the state of the art in computer chess has moved from commercial and academia to the amateur, and the championship format should reflect that. I propose the formation of a new world championship organization, run by the programmers and tasked with organizing an annual online tournament, similar in format to the other online tournaments. This would require neither an expensive venue, nor travel from the participants, so costs should be low.

Any criteria placed on the participants? The difference in opinion concerning the inclusion of engines based on Ippolit/Strelka/Fruit needs to be ironed out if a new world championship organization is to have the support of all authors.

[Laskos]

I was not involved in this illegal RE job you mention. All the information I used for creating Houdini is publicly and freely available to anyone looking for it. The reason Houdini is #1 since 18 months is entirely through my own ideas and improvements, not via any RE of other engines.

Have you ever considered that publishing the result of an illegal RE job (which is how many people think about the creation of Ippolit) might also imply that such information is not "publicly and freely available to anyone" since the original rights owner (which would be Vasik Rajlich in this case) does not lose his rights through the illegal action?

These sources have now effectively become public domain, ALL top engine authors use them: Houdini, Critter, Komodo and even Stockfish (see the latest pawn shelter/storm code)!
There's nothing I can do to revert this situation.

With your reply to Ed you show the same behaviour as many times before: you avoid to stay on topic when someone mentions a critical issue about your engine, and switch to standard replies instead. The reader of Ed's post, including yourself, saw his statement "the source code you started with is the result of an illegal and punishable RE job", and usually one would have expected a reply close to that specific point but you replied that you were "not involved in this illegal RE job" which is of course a non-topic and turns your whole reply into an empty statement.

Sven

I really don't see what it is exactly that you want me to say.
Again:
- The reason Houdini is #1 is because of original work, not because of RE of other engines.
- The allegedly illegal actions that resulted in these sources do not constitute any justification for the RE of Houdini.

Robert

I believe what most are trying to get across here is Houdini wouldn't exist without the initial RE of Rybka. That source was taken, improved, and we now have Houdini. That still doesn't exclude the main issue of Houdini starting from said RE source. Without that single event, you would have had nothing to improve upon. Regardless of the now "public domain" consideration, the base code was not "public domain" at the time.

Do I believe Houdini has a place in future tournaments? No, but because of it's origins. That doesn't mean I don't appreciate the level of competition it has created. As one would say: "You can polish a turd as much as you want, but it is still a turd."

You have polished a turd to an amazing degree, and deserve credit in making it the best turd out there currently, but it is still a turd.

Peter

These arguments really have to end. An utterly crappy "WCCC" and similar tourneys which, in the long run, have to be put into their obscure place, that is, thousands bucks spent per author for some, mostly junk engines to play 10 games, for authors to have some coffees, and a marketing issue which has to be solved. Just ruin the WCCC & Co. brand to market crappy "results" as who knows what World Champion whatever crap, and all is set.

Kai

[Houdini]

I believe what most are trying to get across here is Houdini wouldn't exist without the initial RE of Rybka. That source was taken, improved, and we now have Houdini. That still doesn't exclude the main issue of Houdini starting from said RE source. Without that single event, you would have had nothing to improve upon. Regardless of the now "public domain" consideration, the base code was not "public domain" at the time.

Your speculations about Houdini are incorrect.

Do I believe Houdini has a place in future tournaments? No, but because of it's origins. That doesn't mean I don't appreciate the level of competition it has created. As one would say: "You can polish a turd as much as you want, but it is still a turd."

You have polished a turd to an amazing degree, and deserve credit in making it the best turd out there currently, but it is still a turd.

Peter

I appreciate your juicy imagery while celebrating that Houdini 1.5a has now been downloaded over 150,000 times from my web site.
Enjoy your future, more or less irrelevant tournaments with the engines of your choice .

Robert

[rvida]

... about RE:

Curiously enough, the authors of Critter and Komodo have no problem with that and actively pursue this course.

Please don't put my name nor my engine into your mouth. I am in no way an advocate of reverse engineering a closed-source engine. In fact, it is partially your fault that I decided to look into H1.5. It was because of your repeated denial (or avoidance of answering) to the question of Houdini origins despite the almost identical output of the pre-1.5 versions (1.0 & 1.03). I would have given up after a brief look if the underlying framework were much different from Ippolit, but alas it was not. Almost all of move generators / make / undo etc. were identical. And I never published any Ippo/Robbo sources with your innovations included. Apart from a very brief and incomplete description on this forum I kept them confidential.

Want to know why I think my Critter is more honest work than your Houdini? Because I wrote all the code myself. I got through the pains of porting my (then weak) engine from pascal to C. I went from 0x88 to bitboards. I tried almost all known board representations and choose one that gave the highest NPS. I tried and tested almost all ideas presented on the old CCC forum (btw. have you ever tried the botvinnik-markoff extension? - it is very powerful but very fragile). After the publication of Ippolit (due to heavy censoring on fora I was aware of it much later than most others) I was amazed by the elegance of it despite the horrible code readability. I took many things from it and tested idea after idea in _my own_ framework, and kept all the ones which worked. Sometime after version 0.90 (which was not exactly a weak engine) it started to behave in some positions like the engines from Ippolit family.

Can you see the difference?

Btw. my curiosity was entirely satisfied with H1.5. I never touched the later version.

[rvida]

I believe what most are trying to get across here is Houdini wouldn't exist without the initial RE of Rybka.

I assume you are implying here the Ippolit->Rybka connection.

This is in a way correct, but only indirectly.

Some people here are worshiping a mantra (Ed Schroeder comes to mind) that Ippolit is no more than a leaked/hacked source code of Rybka3. Things would be very simple if that would be the case. Alas, this is not entirely true. While I am pretty sure that the author(s) of Ippolit did a "comprehensive read" of the R3 binary, their engine was indeed written from scratch and the differences are too many to call it a clone. Especially, if someone is in doubt about the ICGA verdict in Fruit->Rybka case, the Rybka->Ippolit case looks quite innocent in comparison.

Unfortunately much of the CC people are doomed to stuck to their beliefs and hype, because not everyone can read disassembled code

[Houdini]

Please don't put my name nor my engine into your mouth. I am in no way an advocate of reverse engineering a closed-source engine.

On this forum you have acknowledged RE'ing Rybka 3 and 4, Komodo and Houdini. You have commented in the past about the inner workings of all these engines (for example on their evaluation functions).
You have found great pleasure in disclosing a lot of RE information about Houdini in various forums, it's a bit late to claim innocence. Even quite recently you were publishing information on the Immortal forum to help someone else RE Houdini 1.5a (if moderators allow I will post the link).

And I never published any Ippo/Robbo sources with your innovations included. Apart from a very brief and incomplete description on this forum I kept them confidential.

That is factually correct, you never published Ippo/Robbo source with my innovations included. You used a different label.

Btw. my curiosity was entirely satisfied with H1.5. I never touched the later version.

You have already made RE comments about Houdini 2.
For example, that Houdini 2 uses 1/3200 pawns in its internal evaluations (whereas Houdini 1.5 uses 1/200 pawns), see http://www.talkchess.com/forum/viewtopi ... 945#444945
How would you know this without disassembling Houdini 2?

Robert

[rvida]

Please don't put my name nor my engine into your mouth. I am in no way an advocate of reverse engineering a closed-source engine.

On this forum you have acknowledged RE'ing Rybka 3 and 4, Komodo and Houdini. You have commented in the past about the inner workings of all these engines (for example on their evaluation functions).

Rybka3 - I never looked into the binary. I have R3 equivalent source code from Yuri O.
Rybka4 - Yes, but I did not disclose anything specific besides the fact that it has much lighter eval than R3 (which is much more similar to Ippolit than R3 eval was)
Komodo - Only partially. Didn't touch the eval (which the authors are proud of), I was curious only about board representation and search. Didn't publish any Komodo specific secret. In our private correspondence I was actually helping them with some hints to improve speed.

You have found great pleasure in disclosing a lot of RE information about Houdini in various forums, it's a bit late to claim innocence. Even quite recently you were publishing information on the Immortal forum to help someone else RE Houdini 1.5a (if moderators allow I will post the link).

It is not a pleasure anymore. In the past I was quite angry with you because of your attitude (denial of starting from public source code). Now I have some respect for your talent, moreso because some of your innovations were almost same as mine (honestly).

And the russians? The smart ones won't post source code on that forum. The incompetent ones (maybe they are not russians after all?) will never succeed in RE. On the immortal forum there were published (not by me) the piece-square table initialisation routines and the material table intialization... Most of this is publicly available anyway. I only corrected some mistakes they made.

Btw. my curiosity was entirely satisfied with H1.5. I never touched the later version.

You have already made RE comments about Houdini 2.
For example, that Houdini 2 uses 1/3200 pawns in its internal evaluations (whereas Houdini 1.5 uses 1/200 pawns), see http://www.talkchess.com/forum/viewtopi ... 945#444945
How would you know this without disassembling Houdini 2?

I don't have Houdini2. But I do have RDP access to a friend's (beta tester) computer which has H2.0 installed. He asked me if I could patch his binary to show "true" evaluations and i did it. It was about a 10min job. I did not made a full disassembly (relevant tools were not installed on his computer), just searching for some patterns using only "hiew". The 1/3200 thing was a random finding - all constants in eval I saw were about *16 (+/- some tuning) compared to H1.5.