WCCC 2011 - Junior is the 2011 World Champion

[marcelk]

Taking the top program and finding 50 ELO is ORDERS OF MAGNITUDE easier than engineering your own program to be the same strength, even given the richness of ideas that are freely available from looking at chessprogramming wiki and reading the sources of other programs. THIS is why legitimate authors are upset.

+1

Let me clarify that there is a reason I threw in the 150 elo number as minimum required 'added value' in such a hypothetical updated rule set and not 50 elo. (And 150 is arguable but not way off but substitute 200 if you wish). The reason being that a 50 can still be achieved by cosmetic changes and optimizations, yet a 150 is very unlikely so and this 1. demonstrates skill and 2. will make an observable difference in a limited-rounds tournament. I also don't believe in copy-pasting that much elo from one engine to another.

[Don]

The Houdofish team might prefer to be listed in more than one entry (like Nalimov). And if it is really really good, why not let them increase their chances this way by letting their genes spread...

I'm sure they would. Which is exactly why it has to be forbidden. Because others could of course also release a 150-Elo detuned version of their engine, let a few dozen 'developers' undo the detuning and add somemeaning less changes, and enter them all as 'cooperators' to multiply their chances.

Exactly! A simple thought experiment:

You go to a tournament and to make this simple there are 9 programs all of exactly the same strength. To make an even number of pairings you offer to run a second copy of your program, one with some minor change but is essentially within 5 or 10 ELO of everyone else. What are the chances that you will win the tournament with one of your programs? What is everyone else's chancing of winning? Were you being gracious by offering the fill that extra slot?

There are different ways to describe the same elephant. Another is unwillingness to be transparent towards the TD regarding origins, and yet another is the difficulty of verifying originality even with source code provided.

But the important point is that the elephant in question is a naughty animal and intentionally wants to sabotage the rules and wreck the system. It would be plain stupid to accomodate them, because however you will change the rules, it won't change the elephant, and he will still find ways to wreck the new system. It is like saying: "My neighbors hate me so much they throw stones through the glass panes of my front door. So next time I am going to leave for work I will leave it open, so they won't be able to do that anymore". Bad idea...

The main reason the tournament is so devaluated is absence of Stockfish, Komodo, Critter... Had they all been there, I don't think there would have been many complaints other than from incurable fanboys,which should not be taken seriously. Changing the rules in the direction you propose does not do the slightest to cure that problem. And the engines it is intended to open the road for would most likely not come either.

[marcelk]

The Houdofish team might prefer to be listed in more than one entry (like Nalimov). And if it is really really good, why not let them increase their chances this way by letting their genes spread... :-)

I'm sure they would. Which is exactly why it has to be forbidden. Because others could of course also release a 150-Elo detuned version of their engine, let a few dozen 'developers' undo the detuning and add somemeaning less changes, and enter them all as 'cooperators' to multiply their chances.

I fear there are far simpler ways to cheat than through such conspiracy scheme that would be easy to spot as well (eg. play remotely with a Houdolitto+GM centaur for example, or let the operator convey information about the opponent's pondering and influence the move selection or time allocation that way...)

There are different ways to describe the same elephant. Another is unwillingness to be transparent towards the TD regarding origins, and yet another is the difficulty of verifying originality even with source code provided.

But the important point is that the elephant in question is a naughty animal and intentionally wants to sabotage the rules and wreck the system. It would be plain stupid to accomodate them, because however you will change the rules, it won't change the elephant, and he will still find ways to wreck the new system. It is like saying: "My neighbors hate me so much they throw stones through the glass panes of my front door. So next time I am going to leave for work I will leave it open, so they won't be able to do that anymore". Bad idea...

The main reason the tournament is so devaluated is absence of Stockfish, Komodo, Critter... Had they all been there, I don't think there would have been many complaints other than from incurable fanboys,which should not be taken seriously. Changing the rules in the direction you propose does not do the slightest to cure that problem. And the engines it is intended to open the road for would most likely not come either.

The assumption there is that forcing everyone to write their own move generator, SEE, SMP before they can add value is a good thing. You can also see it as a barrier that keeps new entrants out.

[bob]

But most "authors" want a fair competition without clones/derivatives, where everyone writes their own program and then we compete head-to-head with each other on as level a playing field as we can define...

This position is not an obvious majority opinion anymore from the tri-ennial ICGA meeting this week where this was a lengthy agenda point. A fair group of participating programmers present have expressed they want the rules to be updated. One line of thinking is that attribution plus added value should be sufficient to compete, instead of 100% originality.

Even I have proposed updating the rules. But I have not seen anyone that would say "copy and attribute and you are OK."

That could be because you were not present, not because these opinions are not discussed among the participating programmers.
'Attribution' can for example to be understood as getting recognized as co-author on equal footing and sharing the accolades. If you use open-sourced Houdofish as your base line and your entry wins, the Houdofish team is co-winner. If you want to be sole winner, then just don't re-use Houdofish's code but write your own. If another team wants to use Houdofish also, that is fine as long as they have added >150 elo by themselves as well as you did.

So teams are co-mingled? That sounds reasonable to have the same person on multiple teams, just because he wrote the original? Exactly how many were proposing that? I'd bet "very few".

Provided the copying is legal ofcourse. If the Houdofish team doesn't want this to happen with their code, but still wishes to publish their sources, they can apply a more restrictive "Crafty-like" license and the code is off-limits in tournaments.

Easy to copy, hard to create.

If we would want to reward effort, we should also ban the use of compilers. Or ask hand-punched paper tape entries only.
Some think it should not be about effort but about what it brings on the board.

Why the max hyperbole? Nobody has said it should be "max work". That is NOT the same thing as "max originality".

The "added value" clause is completely hopeless. Why add a rule that can't be enforced? What is "added value"? Elo? "How is it measured?" ICC? Obviously can't be in real tournaments since you have to show added value before competing, you have to compete to show added value. "How much added value"? 50 Elo? 100 Elo? Again, how will this be measured. You can't use ICC/FICS. It is trivial to manipulate ratings there to show whatever you want...

I'm not an expert but I've heard of a program called 'bayeselo', many rating lists seem to use it.

What is the INPUT for BayesElo? (BTW I use it daily in my cluster testing). The answer is "raw PGN". From WHAT games? Under WHAT conditions? Under WHAT time controls? Etc. Too easy to manipulate.

If that doesn't work they could always throw in the 'expert opinion' clause popularized by rule 2.

As I said, unenforceable rules are worse than no rules...

Yes indeed, and that would be another fine reason to update them.

The core of the question is if the current generation of programmers/members want a tournament that calls itself the WCCC to be one with the strongest legal programs or one with the strongest 100.0%-original ones. In the 1970s this distinction didn't exist but now it is there and it appears to stay. The question how to make understandable/enforceable/etc rules for the desired format is only a secondary question, not a primary one.

In case the ICGA wants remain the place for original programs only (which is a possible outcome), I agree that "Computer Chess Olympiad" is a way more appropriate label for this tournament than "WCCC".

A "world championship tournament" has NEVER been able to reliably choose "the strongest program." That's why the human WCC is not a tournament, but a match (or series of matches)...

[bob]

1. The Ippo's claim to be fresh and original.
2. Vas claims to be fresh and original.
3 Houdart claims to be fresh and original.

Answer is quite simple actually:
1. Ippo is as original as Komodo, R3 is RE, ideas are taken, new program is written.
2. R1 and R2 are not, R3 almost certainly clean code wise, therefore according to your "free taking of ideas is ok" motto, is also original.
3. Houdini 1 and 1.5 in smaller matter are certainly not original, but are completely legal. There is chance the Hounding 2.0 is clean code wise, but I seriously doubt it.

Regarding 1. and 2. and Bob's rotated bitboards, table coefficients (not acquired by tuning) are certainly not something that violates copyright, therefore doesn't violate originality assumption either.

If you copy the CODE, you absolutely violate copyright... There is a difference between "ideas" and "source code" in copyright law as it specifically addresses computer software...

[Terry McCracken]

The Ippo people, or at least Milos seems to be jealous and contemptuous of Junior's title and thinks that Junior should not have been shown any honor for the win. They don't respect anyone but themselves and they have the over-inflated EGO's.
If you look at the "real" authors you will see there is no such infighting among them.

I get your 'spin' Don...but it won't work.

The 'Ippo people' are 'real' authors, every bit as much as you...
very methodical and documented development, innovations galore, a complete table-base solution, java GUIs, etc.
(the list goes on and on, i won't waste space here)

and they're publishing a program that's stronger than Komodo (source code included!) for free...
while remaining anonymous... i.e. taking no individual credit for anything.

taking note of your endless anti-ippolit propaganda,
the question begs to be asked:
who's jealous, contemptuous (and/or over-inflated ego) here...you or them?

Norman, in the first place- I have looked at most all tests and matches run by them ag. the capitalist engines in the last 2 years, and Junior has never even been run by them, if it was, it was so far down i did not even notice it in the list.

In the 2nd place, Junior is not even in the top 10 in any "capitalist" rating lists I have seen.

Dailey said they are jealous and show no respect. They are jealous of no one, not like too many that are on this forum. They are not EVEN JEALOUS OF HOUDINI. Suspect, but not jealous. And they have great respect for Richard and Critter as well as Stockfish and Tord and his guys.

My advice to Dailey would be if HE wants their respect, get his foot out of his mouth, shut up and work on his program.

Do you see Richard and Tord running around bitching about everything. NO. They are too busy with their programs- as Dailey is shopping his ego around.


Best Regards Norman,

gts

Why don't they ban you and other useless tolls?


[size=0]Fu cking Asshole![/size]

[marcelk]

So teams are co-mingled? That sounds reasonable to have the same person on multiple teams, just because he wrote the original? Exactly how many were proposing that? I'd bet "very few".

Teams have been co-mingled from the time that Nalimov was allowed to provide chess code for multiple entrants. Nalimov's code was the first to become a commodity and many see no reason it should be the last.

I'm not an expert but I've heard of a program called 'bayeselo', many rating lists seem to use it.

What is the INPUT for BayesElo? (BTW I use it daily in my cluster testing). The answer is "raw PGN". From WHAT games? Under WHAT conditions? Under WHAT time controls? Etc. Too easy to manipulate.

Conditions set by the TD, or prior winners, whatever, its implementation is solvable.

A "world championship tournament" has NEVER been able to reliably choose "the strongest program." That's why the human WCC is not a tournament, but a match (or series of matches)...

According to wikipedia it better be like that:

A world championship(s) is the top achievement for any sport or contest. The title is usually awarded by contests, ranking systems, stature, ability, etc. This determines the best nation, team, individual (or other entity) in the world in a particular field.

If if here it doesn't (and apparently never has?) it only highlights the problem and not justifies the format.

[hgm]

I fear there are far simpler ways to cheat than through such conspiracy scheme that would be easy to spot as well ...

So you spot it (easily). Now what good would that do you when you constructed rules that explicitly allow it....???

In fact this kind of 'conspiracy' is exactly what we are facing today, and you seem to fall for it with open eyes. You think it is a coincidence that an author or group of authors play sillybuggers and throw a buggy template of a superstrog engine in the public domain?

The assumption there is that forcing everyone to write their own move generator, SEE, SMP before they can add value is a good thing. You can also see it as a barrier that keeps new entrants out.

I think you are missing the point. None of the things you mention is what kept Stockfish, Critter, Komodo etc. out.

The only novel aspect of your proposal that creates the remote impression it could be a good idea is that it allows participation of public code without the author's explicit consent. Take that way, and we will be exactly in the same situation as we are now: Ippo derivatives won't be able to run, because the Ippo authors are unavailable for granting pemission.

But if you allow that, wouldn't it be far simpler to just allow publicly available engines to participate without author consent, without additional rule changes? Just appoint operators for Stockfish, Komodo, Critter, Ippolit....That would be 4 times as effective as what you propose, as the latter would not do anything to encourage participation of the first three mentioned.

An unrelated remark:

In professional cycling (one of the world's major sports!) the UCI world championship is usually not won by the world's strongest cyclist either.

[marcelk]

But if you allow that, wouldn't it be far simpler to just allow publicly available engines to participate without author consent, without additional rule changes? Just appoint operators for Stockfish, Komodo, Critter, Ippolit....That would be 4 times as effective as what you propose, as the latter would not do anything to encourage participation of the first three mentioned.

That is also a nice idea. (It requires changes: one that 'an' author needs to enter and two that source code must be available at TD's request.) There are many variations, up to awarding the world title to the leading entry on the rating lists.

[hgm]

Indeed, availability of source code could be a problem. I think it would be unwise to drop that requirement, so that would make it impossible to enter closed-source engines like Komodo by third parties. But for engines like Stockfish and Ippolit it would still work.

But the essetial element of my scheme is that in case of multiple third-party entry the TD decides who to accept. If the aim is to raise the status of the tourney maximally, it means they should be able to go for hightest Elo. So even if Tord would wants to enter Stockfish, but I would want to enter my own closed-source private derivative of it that is 250 Elo stronger, it means the original Stockfish will be excluded. But of course Tord (and other Stockfish authors) would still be able to share the honor of a World-title when my derivative wins. (That is what they brought upon themselves by opening their sources.)