Pleading for TogaII take part in tournament

[slobo]

3) TogaII is a clone of Fruit. Therefore you cant have both Fruit and TogaII in the same tournament.

But it is possible. I did run tournaments with Fruit & Toga both on the same computer even, in the same interface. No problem Tournament rules can be pragmatically adjusted by the organizer once in a while, or exceptions can be granted if appropriate. After all, it is a computer chess tournament only, not the election of the pope.

Very good point. I sign it too.

[Guetti]

3) TogaII is a clone of Fruit. Therefore you cant have both Fruit and TogaII in the same tournament.

But it is possible. I did run tournaments with Fruit & Toga both on the same computer even, in the same interface. No problem Tournament rules can be pragmatically adjusted by the organizer once in a while, or exceptions can be granted if appropriate. After all, it is a computer chess tournament only, not the election of the pope.

Most rules are not good enough to apply them without intelligent exceptions. A situation like Fruit/Toga as is it is now, most probably wasn't in mind when the anti-clone rule (3a) was created. I am not aware of any other "derivate engine" which wasn't insignificant and/or didn't disappear quickly. As for that, Toga is unique.

(Ignoring Strelka which I consider incorrectly based on reverse engineering and breach of licenses.)

Nevetheless, it is clear to me that it will not happen anyway, for various reasons. But nobody should try to tell me that it is impossible. That's just not true.

It's not impossible, just unwanted.
Why not allow all versions of Fritz to play? Fritz5, Fritz7, Fritz8 etc? There's probably more difference between Fritz5 and Fritz10 than between Fruit and Toga. Or Rybka, all versions since Rybka1.0 beta. They are all legal, so why not include them? Maybe I would even bet some money on the top 10 rankings.

[Mike S.]

That's why I wrote (as a proposal), "...exceptions can be granted if appropriate."

I think I have also seen both The Baron and Crash Test Dummy in an official tournament, for example (if I'm not wrong, more than once). What about that? I think it is ok. I don't think that their code is 100% different

There are more such follow-up engines or parallel projects, where I would absolutely not complain about: Gothmog/Glaurung, Terra/Alaric, Joker/microMax... Maybe the 2nd engine each, is completely(?) or more or less new, but another obvious difference to Fruit/Toga is: Different people are not involved in these examples, each (or not mainly).

Another question: What if Toga registeres first, for one of the "official" tournaments like CCT, Leiden etc. Would you then say

1. Fruit cannot participate, or
2. if Fruit registers later, Toga - although legal and correct engine - has to withdraw?!

(or 3. who cares)

I wouldn't have any problem - I'd simply accept both and enjoy the tournament.

[bob]

2 points here :

1) TogaII (the derived engine) still open source.
2) TogaII is the strongest open source engine.

One more point. Engines derived almost 100% from another engine are simply not allowed in these kinds of events...

[Mike S.]

Engines derived almost 100% from another engine are simply not allowed in these kinds of events...

Simple rules are bad rules if sensible exceptions are impossible.

About "almost 100%"... 6% only (if not even less) make the difference, if we are humans or chimps. Although, that does not mean that I suggest that a chimp should participate in the FIDE world championship. He would of course need to undergo the same qualification process.

[Dann Corbit]

Engines derived almost 100% from another engine are simply not allowed in these kinds of events...

Simple rules are bad rules if sensible exceptions are impossible.

About "almost 100%"... 6% only (if not even less) make the difference, if we are humans or chimps. Although, that does not mean that I suggest that a chimp should participate in the FIDE world championship. He would of course need to undergo the same qualification process.

I think it would make sense to allow TogaII if Fruit did not want to compete.

[Ryan Benitez]

I think it would make sense to allow TogaII if Fruit did not want to compete.

At some point the GPL status may be looked at in tournaments. Who owns a GPL engine? If one engine is GPL it can use any part from another GPL engines, how much of another engine and what parts make it a clone of another GPL engine? If one GPL engine can not play should all be excluded? The intent of the GPL is to share in progress, something the computer chess community seems to sadly look down on. Tournaments are competitive, very contrary to the intent of an engine being GPL.

[Dann Corbit]

I think it would make sense to allow TogaII if Fruit did not want to compete.

At some point the GPL status may be looked at in tournaments. Who owns a GPL engine? If one engine is GPL it can use any part from another GPL engines, how much of another engine and what parts make it a clone of another GPL engine? If one GPL engine can not play should all be excluded? The intent of the GPL is to share in progress, something the computer chess community seems to sadly look down on. Tournaments are competitive, very contrary to the intent of an engine being GPL.

If it continues active development, then at some point, TogaII will diverge enough from Fruit to be a wholly different chess engine. In such an instance, I think it would be good to allow side by side competition with its progenitor. As far as two extremely similar engines in a contest -- that is a call for the tournament director. I think that the main reason they make the rule is that it is easier to legislate than think.

Suppose that ten guys from a community college somewhere all get a copy of Glaurung and make their own fork of the project. Each adds some interesting new facet so the programs are at least 50 Elo stronger than Glaurung. Now if we have ten current entries for a chess contest and we toss in 10 Glaurung clones (which would have been excellent engines even with no improvements). The big problem here is that Joe's engine now has to compete with eleven super tough Glaurung opponents instead of one. The chance to beat them all is exponentially lower than beating just one of them.

Consider that Joe's engine is 40% as strong as Glaurung. We will ignore the fact that the new clones are a bit stronger.

The chances of beating Glaurung in one game, head to head will be 40% (a bit better if white and a bit worse if black, but we will average to our 40%).

Now, he has to face 11 of them. The odds of beating them all are:
(1-.4)^11 = 0.00362797056
Translation: "It ain't gonna happen."

So we can see why such a rule makes sense, even for legitimate clones.

Now, when things that look like exceptions come along, we could just say "Heck, let's let him play!" but now all the others that get turned away can say "Why not me too!" leading to a big, convoluted discussion. Which is why it is simpler just to legislate a rule.

These sorts of rules, having to do with formal contests are well known by the chess community in advance, so I think we have to simply consider it part of the "cost" of clone writing.

On the other hand, I have not seen 10 Glaurung clones or 10 Fruit clones pop out of the woodwork recently, so we could simply waive the rule. Indeed, such choices could be made by any contest organizers, but they have to also look at the ire they may draw from non-clone competitors who may pull out of the contest in anger if clones are allowed.

So, as usual with complicated things like this, I see no easy answers.

[Guetti]

Well said, Dann.
The fascination of online tournaments like CCT10 is not that the strongest engine will come out on top. The fascination is e.g that CrashTestDummy might beat Rybka in its game. For Richard, such a victory probably brings as much (or more) personal joy as for Vas to win the tourney once again.
People tend to forget such stuff.

[PauloSoare]

I think it would make sense to allow TogaII if Fruit did not want to compete.

At some point the GPL status may be looked at in tournaments. Who owns a GPL engine? If one engine is GPL it can use any part from another GPL engines, how much of another engine and what parts make it a clone of another GPL engine? If one GPL engine can not play should all be excluded? The intent of the GPL is to share in progress, something the computer chess community seems to sadly look down on. Tournaments are competitive, very contrary to the intent of an engine being GPL.

If it continues active development, then at some point, TogaII will diverge enough from Fruit to be a wholly different chess engine. In such an instance, I think it would be good to allow side by side competition with its progenitor. As far as two extremely similar engines in a contest -- that is a call for the tournament director. I think that the main reason they make the rule is that it is easier to legislate than think.

Suppose that ten guys from a community college somewhere all get a copy of Glaurung and make their own fork of the project. Each adds some interesting new facet so the programs are at least 50 Elo stronger than Glaurung. Now if we have ten current entries for a chess contest and we toss in 10 Glaurung clones (which would have been excellent engines even with no improvements). The big problem here is that Joe's engine now has to compete with eleven super tough Glaurung opponents instead of one. The chance to beat them all is exponentially lower than beating just one of them.

Consider that Joe's engine is 40% as strong as Glaurung. We will ignore the fact that the new clones are a bit stronger.

The chances of beating Glaurung in one game, head to head will be 40% (a bit better if white and a bit worse if black, but we will average to our 40%).

Now, he has to face 11 of them. The odds of beating them all are:
(1-.4)^11 = 0.00362797056
Translation: "It ain't gonna happen."

So we can see why such a rule makes sense, even for legitimate clones.

Now, when things that look like exceptions come along, we could just say "Heck, let's let him play!" but now all the others that get turned away can say "Why not me too!" leading to a big, convoluted discussion. Which is why it is simpler just to legislate a rule.

These sorts of rules, having to do with formal contests are well known by the chess community in advance, so I think we have to simply consider it part of the "cost" of clone writing.

On the other hand, I have not seen 10 Glaurung clones or 10 Fruit clones pop out of the woodwork recently, so we could simply waive the rule. Indeed, such choices could be made by any contest organizers, but they have to also look at the ire they may draw from non-clone competitors who may pull out of the contest in anger if clones are allowed.

So, as usual with complicated things like this, I see no easy answers.

Perhaps the probability of his example happen was minor than 0.00362797056. Togga II is a unique case, this is the main point
of Mike Scheidl. It´s an excetion to the rules.

Paulo Soares