POLL: Whether Houdini Is A Clone?

[Dr.Wael Deeb]

Of course it is....a 3300 Elo program from a scratch,come on

BTW<I have A FireBird 1.1 personality that is stronger than Houdini 1.03a by 30-40 Elo and I am not a programmer....
Will post results soon....
Dr.D

Just wondering .... have you done tests in both long and short time controls ? and this personality wins in all matches?
-- this is interesting.

I play 20 minutes + 20 seconds increment time control and I've played around 300 games with these personalities so far on two overclocked Q9950 quads....the latest running tournament engage 33 engine playing each other 4 games in a RR....a total of 128 games for each engine and personality and FireBird 1.1 Dr.Deeb I is leading and will probably win the tournament....
I'll post the results,the cross table and the games in .cbv format....
Dr.D

[Mark Young]

Who Cares, Stop the witch hunt. I guess Houdini is now the target because it is #1 on more and more rating lists. If Vas does not care, why should anyone else care. For now I will take Robert's word. He has a better track record then Vas.

[Dr.Wael Deeb]

Who Cares, Stop the witch hunt. I guess Houdini is now the target because it is #1 on more and more rating lists. If Vas does not care, why should anyone else care. For now I will take Robert's word. He has a better track record then Vas.

True,he's way more polite and cooperative with his fans and he charges no penny

My theory is that Robert is an extremely talented programmer who had a +2700 chess engine and after the Ippo/Robbo breakthrough he got several ideas from the source and created a 3300 engine....nothing wrong with that I guess....

As for Houdini being number one,no Sir,it isn't....there are at least 2 chess entities which are stronger than Rybka 4 and Houdini 1.03a....
Dr.D

[kranium]

Of course it is....a 3300 Elo program from a scratch,come on

BTW<I have A FireBird 1.1 personality that is stronger than Houdini 1.03a by 30-40 Elo and I am not a programmer....
Will post results soon....
Dr.D

Just wondering .... have you done tests in both long and short time controls ? and this personality wins in all matches?
-- this is interesting.

I play 20 minutes + 20 seconds increment time control and I've played around 300 games with these personalities so far on two overclocked Q9950 quads....the latest running tournament engage 33 engine playing each other 4 games in a RR....a total of 128 games for each engine and personality and FireBird 1.1 Dr.Deeb I is leading and will probably win the tournament....
I'll post the results,the cross table and the games in .cbv format....
Dr.D

You are a genius!
my own tests confirm the improvement without doubt...

i can't express how much i appreciate the testing, 'tweaking', and optimization of settings you have done here...
this is not easy work...takes enormous discipline, diligence and patience...

your effort here very similar to the valuable 'tuning' work that Joona Kiiski is doing for Stockfish...and he's now listed as one of the authors.

i have a version almost ready now Fire 1.4...based on 1.1, with Dr.D settings...
it's very strong.

i would like to release it soon as Fire 1.4 DD (or something similar)
(if agreeable to Dr. D of course)



Norm

[hyatt]

First, terminology.

"Clone" is an exact copy of a program. Usually produced by taking a .exe file (windows) and using a debugger to change key strings such as the program name, so that its origin will remain hidden. This has been done many times, particularly on chess servers.

"Derivative" is a modified copy, but starts from source rather than from executable. This makes it possible to actually change parts. Some change a lot. Some change so little that it might actually qualify as a "clone".

However, in today's discussions about the ip* and family of programs, most seem to intermingle those two terms and use then interchangably, which is not that unreasonable. Either term suggests the truth, which is that it is not an "original work" by any reasonable definition, and can't participate in the usual computer chess events...

But then Rybka is derivative of Fruit, and nobody cares.

Can't help that. I was just explaining the terms. If the shoe fits _any_ program, it should wear it.

[Chan Rasjid]

First, terminology.

"Clone" is an exact copy of a program. Usually produced by taking a .exe file (windows) and using a debugger to change key strings such as the program name, so that its origin will remain hidden. This has been done many times, particularly on chess servers.

"Derivative" is a modified copy, but starts from source rather than from executable. This makes it possible to actually change parts. Some change a lot. Some change so little that it might actually qualify as a "clone".

However, in today's discussions about the ip* and family of programs, most seem to intermingle those two terms and use then interchangably, which is not that unreasonable. Either term suggests the truth, which is that it is not an "original work" by any reasonable definition, and can't participate in the usual computer chess events...

Words should be used with its usual meaning if possible otherwise there would be chaos. Cloning in chess engine need not be confused at all with derivatives - all the confusion comes from non-programmers. Programmers could easily distinguish the two and need not mixed them together.

The usual meaning of a clone in genetics is the clone has a direct material link (genes) traceable to an ancestor. In chess engines, it means a form of copying that is not acceptable to the computer chess community - basically copying from the executable/binaries or from cut_and_paste from the source codes. As long as an author writes the codes himself, whether in the same programming language or in another language, it is not a clone - there is no direct material link traceable to another as an ancestor.

Derivative implies a substantial copying of ideas to the point that the derivative behaves closely to that of the other engine it imitates. Robert Houdart seems to have admitted Houdini is a form of Ippo* derivative.

Houdini might not be much different from other close source programs; if he did not "tell", it is just like any other unless we have proofs through disassembly and the disassembly must also be applied to all and sundry including Hiarc, Shredder, Rybka, Komodo, etc... not just singling out Houdini to bash. If Robert Houdart were to make enough of changes, its behavior might just be that of another "original" strong engine. Tournament organisers are not supposed to know Houdini is a derivative unless official complaints are made against it.

Rasjid.

[Mark Young]

Who Cares, Stop the witch hunt. I guess Houdini is now the target because it is #1 on more and more rating lists. If Vas does not care, why should anyone else care. For now I will take Robert's word. He has a better track record then Vas.

True,he's way more polite and cooperative with his fans and he charges no penny

My theory is that Robert is an extremely talented programmer who had a +2700 chess engine and after the Ippo/Robbo breakthrough he got several ideas from the source and created a 3300 engine....nothing wrong with that I guess....

As for Houdini being number one,no Sir,it isn't....there are at least 2 chess entities which are stronger than Rybka 4 and Houdini 1.03a....
Dr.D

I see Norm is going to release Fire 1.4, Based on your settings.

I test default settings, You could go crazy trying to test all the program settings. I have to draw the line somewhere. I can't wait to see what you guys have come up with, and I hope it holds up at long time controls. (3 mins a move average)

[lmader]

"Clone" is an exact copy of a program. Usually produced by taking a .exe file (windows) and using a debugger to change key strings such as the program name, so that its origin will remain hidden. This has been done many times, particularly on chess servers.

"Derivative" is a modified copy, but starts from source rather than from executable. This makes it possible to actually change parts. Some change a lot. Some change so little that it might actually qualify as a "clone".

The usual meaning of a clone in genetics is the clone has a direct material link (genes) traceable to an ancestor. In chess engines, it means a form of copying that is not acceptable to the computer chess community - basically copying from the executable/binaries or from cut_and_paste from the source codes. As long as an author writes the codes himself, whether in the same programming language or in another language, it is not a clone - there is no direct material link traceable to another as an ancestor.

Derivative implies a substantial copying of ideas to the point that the derivative behaves closely to that of the other engine it imitates. Robert Houdart seems to have admitted Houdini is a form of Ippo* derivative.

Rasjid.

If I have read this correctly, Rasjid's definition of derivative is different than Dr. Hyatt's: Rasjid's says derivatives copy only ideas, Dr. Hyatt's says derivatives copy actual code. I think Dr.Hyatt's definition is correct. I would say that copying only ideas, no code, does not make for a derivative. Anyone care to weigh in?

[BB+]

How is a POLL a way of determining this? Fools agree on many things. Various persons throughout history have even exploited this behaviour. The question of whether Houdini 1.X is a "clone" of Y can be investigated with definite techniques, and rumours concerning the situation are unlikely to clear up the issue.

Programmers could easily distinguish the two and need not mixed them together.

Could and do don't seem to be the same. Especially when there is reason to conflate the matter. For instance, I think the word clone is still the word of choice for IPPOLIT and friends in some locales.

I would say that copying only ideas, no code, does not make for a derivative. Anyone care to weigh in?

I think this has been the historical standard in computer chess. As ZW and others have pointed out, the more recent events in the last years have made this less clear. The concept of Intellectual Property can also play a part.

Here is a scenario: Engine A has a material imbalance table of 419904 entries, computed from 8 parameters. Engine X has a similar table, also computed from 8 parameters (knight/pawn affinity, rook/pawn affinity, redundant major pieces...), which differ only slightly (all plus or minus 1 or 2 millipawns). Engine Y has a slightly simplified table, using only 6 of these 8 parameters, and rounds all the numbers to the nearest multiple of 5, after re-scaling from millipawns to centipawns. Engine Z has a material imbalance table with 419904 entries, using 5 parameters, 4 of which are used in Engine A, and has numbers that are kinda-sorta similar. Engine W has a material imbalance table with 236196 entries (not doing opposite-colour bishops)... It becomes not easy to make a cut-and-dried rule after a bit of this. I would say that W is clearly just the idea, probably Z is also, X is essentially a clone, and with Y more information would be needed. Now repeat this for other aspects of the code...

[Dr.Wael Deeb]

Of course it is....a 3300 Elo program from a scratch,come on

BTW<I have A FireBird 1.1 personality that is stronger than Houdini 1.03a by 30-40 Elo and I am not a programmer....
Will post results soon....
Dr.D

Just wondering .... have you done tests in both long and short time controls ? and this personality wins in all matches?
-- this is interesting.

I play 20 minutes + 20 seconds increment time control and I've played around 300 games with these personalities so far on two overclocked Q9950 quads....the latest running tournament engage 33 engine playing each other 4 games in a RR....a total of 128 games for each engine and personality and FireBird 1.1 Dr.Deeb I is leading and will probably win the tournament....
I'll post the results,the cross table and the games in .cbv format....
Dr.D

You are a genius!
my own tests confirm the improvement without doubt...

i can't express how much i appreciate the testing, 'tweaking', and optimization of settings you have done here...
this is not easy work...takes enormous discipline, diligence and patience...

your effort here very similar to the valuable 'tuning' work that Joona Kiiski is doing for Stockfish...and he's now listed as one of the authors.

i have a version almost ready now Fire 1.4...based on 1.1, with Dr.D settings...
it's very strong.

i would like to release it soon as Fire 1.4 DD (or something similar)
(if agreeable to Dr. D of course)



Norm

Thanks a lot Norman and of course the major credit goes for you for creating such a marvellous chess engine like FireBird 1.1....I spotted the excellent work at once and that's why I decided to make it even stronger
It will be an honour to release the next Fire DD or whatever you prefer
Cheers,
Dr.D