Computer Chess Grandmaster Title?

[Steve Maughan]

As I write Maverick it's apparent that to create an engine from scratch which plays at genuine Grandmaster level is not trivial. It's more than just bolting together a set of known heuristics. Which made me think - it would be great if there was a "Computer Chess Grandmaster" tItle. It would give our hobby a target to aim for. Maybe it would also attract new talent.

Here's what I propose. The title would be governed by a body e.g. CCRL. If anyone wants to submit their engine there is a fee of $xxx which goes to the testers. The engine would first need to pass a test to ensure it wasn't a clone. Possibly also submit source code. It would then be tested by someone like CCRL. If it scored >2500 ELO in a x game match againt other engines then the author would be awarded the title of computer chess Grandmaster.

I could see some honorary title awarded up front to those individuals which have written engines which are clearly of GM strength.

What do others think?

Steve

[jdart]

2500 is an arbitrary number. The various ratings lists for engines have rating numbers that are internally meaningful but comparing them with human (FIDE) ratings is problematic. "Grandmaster" is whatever you want it to be, for computers.

--Jon

[lucasart]

Here's what I propose. The title would be governed by a body e.g. CCRL.

In principle, it sounds like a good idea. In practice I can already see all the politics and BS that will arise from it.

If anyone wants to submit their engine there is a fee of $xxx which goes to the testers.

That's where I draw the line. This is completely against the whole spirit of free software. There are people who donate their brain time to write free software, and people who donate their CPU time to test these. If you introduce money in there, the only effect is that you eradicate free software and only commercial developpers or usurpers (cloners) will participate: I certainly won't.

The engine would first need to pass a test to ensure it wasn't a clone. Possibly also submit source code.

Same as above:
In principle, it sounds like a good idea. In practice I can already see all the politics and BS that will arise from it. The truth is that it's very hard to determine if and to what extent an engine is plagiarized work, especially if the source code is not available. The only way to solve the problem IMO is that the event would be limited to open source engines ONLY. And the source should be available to everyone, not just a judge from CCRL (only experienced engine developpers would be able to make the right decisions, people running external tools not understand what they are doing should never be in a position to judge what they do not know).

It would then be tested by someone like CCRL. If it scored >2500 ELO in a x game match againt other engines then the author would be awarded the title of computer chess Grandmaster.

2500 CCRL is arbitrary, but what difference is there between receiving an official computer GM titel (whatever that is/means) and knowing that your engine is > 2500 on CCRL. I'm happy to know that my engine is ~ 2750 on CCRL, and I don't need anything else.

I could see some honorary title awarded up front to those individuals which have written engines which are clearly of GM strength.

I don't understand this business of honorary title. Seriously, is that so important to flatter one's ego? Would you look at yourself differently in the mirror in the morning knowing that you have this title? Would you put it in a golden frame on the wall in your living room? This is ridiculous...

Conclusion: I think the current system is perfect the way it is, and all your suggestions would only pervert it. CCRL is great and so is CEGT, and IPON (for top engines only).

Besides, it's not about the ELO it's about the achievement. Clearly between:
- the usurpers who take the source code of IvanHoe, put their name on it, butcher the source code a little bit (without testing of course) to pretend they have done some genuine work, call it another name and compete in CCRL and CEGT.
- and the genuine engine developpers

WE know who is who and what the difference is. It really doesn't matter what clueless end users think. Not to me anyway.

Despite it's weak ELO, I certainly consider your engine Maverick a much greater achievement than Firenzina or Vitrivius.

[kinderchocolate]

As I write Maverick it's apparent that to create an engine from scratch which plays at genuine Grandmaster level is not trivial. It's more than just bolting together a set of known heuristics. Which made me think - it would be great if there was a "Computer Chess Grandmaster" tItle. It would give our hobby a target to aim for. Maybe it would also attract new talent.

Here's what I propose. The title would be governed by a body e.g. CCRL. If anyone wants to submit their engine there is a fee of $xxx which goes to the testers. The engine would first need to pass a test to ensure it wasn't a clone. Possibly also submit source code. It would then be tested by someone like CCRL. If it scored >2500 ELO in a x game match againt other engines then the author would be awarded the title of computer chess Grandmaster.

I could see some honorary title awarded up front to those individuals which have written engines which are clearly of GM strength.

What do others think?

Steve

There is zero probability that this would work. The title is only meaningful if people actually care about it. Everybody knows Houdini, but almost nobody cares or remembers the current world computer champion. My friends still think Houdini is the current world computer chess champion because it's the strongest. If the world title is practically useless, how important a grandmaster would be?

[Steve Maughan]

From the initial reaction it sounds as if there is zero appetite for my suggestion.

I'm a little surprised but no problem. There are clearly some issues about who would do the testing, how one would test for clones, where the cutoff is etc - but I don't think they are insurmountable if the will is there - but it would seem it simply isn't.

I'll drop the idea,

Best,

Steve

[JarvisXerxes]

From the initial reaction it sounds as if there is zero appetite for my suggestion.

I'm a little surprised but no problem. There are clearly some issues about who would do the testing, how one would test for clones, where the cutoff is etc - but I don't think they are insurmountable if the will is there - but it would seem it simply isn't.

I'll drop the idea,

Best,

Steve

I second the idea. The same logic applies to human - why do we assign GM titles to humans? To boost their egos?
Surely those titles can be commercialized as well, as is the case with those professional players. For non-pro players, that title is a great incentive, and more importantly, an honor for one's intellectual achievement. If you ask me, why do we need "honor" at all? Well ... that's a question too philosophical for me to answer.

I believe writing an original and exceptionally strong engine is no less challenging than actually being a grandmaster.

[lucasart]

Actually these kind of "titles" exist in a way. Some people create some fake tournaments, delivering fake trophees, like the CSVN folks for example. This year there were about 6 participants, and the winner of their "International Chess Tournament" was ... Rookie !

This kind of thing really makes me laugh, and I can only conclude that these people are desperate, and create such events as a way to draw attention and bring their work to a level where it does not belong. Everyone knows what the strongest programs are, and who is (or is not) a genuine developper of such programs.

Comparing the hard work and skills required to write a GM strength program and becoming a GM yourself is a very strange idea to say the least:
- it is true that most GMs do not have the skills to write a strong chess program, and some of them would never be able to do it (simply because they do not have a "technical background")
- for someone like me, it "only" took me a year to make a 2500 ELO program from scratch, but I could spend the rest of my life studying chess, I would *never* be a GM.

So these tasks require very different skills sets, and are not comparable.

[Gerd Isenberg]

Actually these kind of "titles" exist in a way. Some people create some fake tournaments, delivering fake trophees, like the CSVN folks for example. This year there were about 6 participants, and the winner of their "International Chess Tournament" was ... Rookie !

This kind of thing really makes me laugh, and I can only conclude that these people are desperate, and create such events as a way to draw attention and bring their work to a level where it does not belong. Everyone knows what the strongest programs are, and who is (or is not) a genuine developper of such programs.

Please stop your ignorance and bashing on the motives of other programmers to play competitive tournaments.

It is not the fault of the CSVN apparently stronger programs did not participate or are not original. Do you think Rookie, The Baron and The King et al. are weak programs? Can you tell me what is original on DiscoCheck?

[lucasart]

It is not the fault of the CSVN apparently stronger programs did not participate or are not original.

My point is that such titles are meaningless. It may not be the fault of CSVN, but at the end of the day, neither, nor Komodo, nor Critter, nor Stockfish, not dozens of other genuine programs that are stronger than Rookie have shown any interest in this tournament. And it's not against CSVN or Rookie that I'm saying that. It's *in general*, and I think that ICGA is as much of a useless dinosaure as the CSVN.

Do you think Rookie, The Baron and The King et al. are weak programs?

It depends what you call weak. In absolute terms they are strong, but so are lots of other programs. As you can on this list (which is probably not complete) there are at least 62 different programs that are stronger than The Baron (I couldn't find Rookie on there unfortunately)
http://www.computerchess.org.uk/ccrl/4040/

Can you tell me what is original on DiscoCheck?

Perhaps the fact that I wrote it from scratch over the course of 3 years, starting from int main() onwards. Would it make you happy if I submitted it to the signature test of Richard Pili ? After all why not. Let's see the results, and we will know if I would have qualified for their meaningless competition. I can run it and send you the results if you PM me your email (or you can do it yourself if you do not trust me, since my program and source code is freely available, see my signature).

[velmarin]

You wrote int main, then copy kurt eg, stock, ect, You only get to International Master, being generous.