When will we see HOUDINI in official tournaments?

[IWB]

for point 2 I think that if there is a similiarity of at least 0.6 it means that Robos is not accepted because Rybka3 was before the Robos and I assume that it is possible to show more than 0.6 similiarity between Rybka3 and Rybka4.1.

Right, but then Ed's sentence come into play:

Robbolito is linked to Ippolit from an unknown author, end of story.

This means that for Ed (is there a similarity test to the CSVN? ) the Littos are an unknown start. (And we all read the sentence that there where more similarities between the Robos and R4 than between Robo and R3 ...) R4 was released later, right? When I follow that logic it means that Rybka 4/4.1 would be out of the tourney!

Bye
Ingo

PS: A similarity test with Robo 0.85 and Robo 0.9 with Rybka 3, 4 and 4.1 would be really helpfull regarding the CSVN rule!

[michiguel]

for point 2 I think that if there is a similiarity of at least 0.6 it means that Robos is not accepted because Rybka3 was before the Robos and I assume that it is possible to show more than 0.6 similiarity between Rybka3 and Rybka4.1.

Right, but then Ed's sentence come into play:

Robbolito is linked to Ippolit from an unknown author, end of story.

This means that for Ed (is there a similarity test to the CSVN? ) the Littos are an unknown start. (And we all read the sentence that there where more similarities between the Robos and R4 than between Robo and R3 ...) R4 was released later, right? When I follow that logic it means that Rybka 4/4.1 would be out of the tourney!

Bye
Ingo

PS: A similarity test with Robo 0.85 and Robo 0.9 with Rybka 3, 4 and 4.1 would be really helpfull regarding the CSVN rule!

See my reply in EO

Miguel

[Harvey Williamson]

See my reply in EO

Miguel

sigh, so at weekends you have to move 2 threads for no reason?

[BubbaTough]

1. You have a rule which you cant enforce if you allow remote engines (and to double the problem with a cluster)

Well, it seems to me like there are no set of rules that work perfectly if you allow remote engines, particularly those with custom remote hardware hard to replicate elsewhere. All you can do is set up some reasonable compromise between trust and verify. The question is not whether proposed rule sets are perfect, obviously they are not. The question is whether they are better than previous rules sets, and thus deserve to replace them. I am on the fence on the 60% rule concept myself, though I definitely see the advantages. Luckily for tournaments like yours, the whole remote engine issue is moot.

-Sam

[michiguel]

See my reply in EO

Miguel

sigh, so at weekends you have to move 2 threads for no reason?

I did not move anything, I initiated a thread in EO dealing with a spin off question from here. That was my OWN post.

Miguel

[Harvey Williamson]

See my reply in EO

Miguel

sigh, so at weekends you have to move 2 threads for no reason?

I did not move anything, I initiated a thread in EO dealing with a spin off question from here. That was my OWN post.

Miguel

Excellent! maybe there is hope.

[IWB]

1. You have a rule which you cant enforce if you allow remote engines (and to double the problem with a cluster)

Well, it seems to me like there are no set of rules that work perfectly if you allow remote engines, particularly those with custom remote hardware hard to replicate elsewhere. All you can do is set up some reasonable compromise between trust and verify. The question is not whether proposed rule sets are perfect, obviously they are not. The question is whether they are better than previous rules sets, and thus deserve to replace them. I am on the fence on the 60% rule concept myself, though I definitely see the advantages. Luckily for tournaments like yours, the whole remote engine issue is moot.

-Sam

Here I fully agree with you, I see that they want to have a "clean" tourney, just I see a bigger unjustice in testing some and let other(s) do whatever they want as it is impossible to test!

It comes down to: A rule has to apply to all or none!

Bye
Ingo

[michiguel]

1. You have a rule which you cant enforce if you allow remote engines (and to double the problem with a cluster)

Well, it seems to me like there are no set of rules that work perfectly if you allow remote engines, particularly those with custom remote hardware hard to replicate elsewhere. All you can do is set up some reasonable compromise between trust and verify. The question is not whether proposed rule sets are perfect, obviously they are not. The question is whether they are better than previous rules sets, and thus deserve to replace them. I am on the fence on the 60% rule concept myself, though I definitely see the advantages. Luckily for tournaments like yours, the whole remote engine issue is moot.

-Sam

Here I fully agree with you, I see that they want to have a "clean" tourney, just I see a bigger unjustice in testing some and let other(s) do whatever they want as it is impossible to test!

It comes down to: A rule has to apply to all or none!

Bye
Ingo

Just to clarify your position, you prefer a format that is free for all. Is that correct?

Miguel

[Rebel]

As after all the similarity software is meant to measure the similarity of the evaluation function.

But that's disturbing, because I assume - although not being a programmer - that currently much more software engineering intelligence is in the search code, not in the evaluations. I may be wrong but usually my guesses are good.

There was a document publically available (not from the original engine authors) which contained detailled analysis and descriptions of every evaluation term of the engine Toga. Insiders will be aware; I didn't keep the download link. It was comprehensive.

Does it make sense to base a similarity analysis on evaluation only, disregarding search code? To me this seems to be flawed, if being used to exclude engines from tournaments based on percentages. Eval may by 70% similar, search (not tested) may be MUCH less similar.

Maybe this needs to be reconsidered, but of course I am only talking from a spectator's viewpoint. Thanks for hearing me.

If you take a program (I randomly took Fruit 2.1) and give it 2x and 4x more time then you will see the influence of the search as demonstrated below. But it will not hide the origin.

Code: Select all

  1) Fruit 2.1 (time: 100 ms  scale: 1.0)
  2) Fruit 2.1 (time: 200 ms  scale: 1.0)
  3) Fruit 2.1 (time: 400 ms  scale: 1.0)

         1     2     3
  1.  ----- 82.64 72.68
  2.  82.64 ----- 83.31
  3.  72.68 83.31 -----

[IWB]

Just to clarify your position, you prefer a format that is free for all. Is that correct?

Miguel

NO, not at all! That is a complete missunderstanding of my position and what I wanted to say here!
It is just that if you follow the argumentation of the CSVN the rule set they implemented is useless as it proves nothing and is impossible to enforce. If one trys to follow their Rybka argumentation the only logical format for them would be to allow everything as they doubt any possible "proof" without getting willingly the original source!

So what I think is not not part of this thread but:
I think that the ICGa rule set is the only resonable. You have to support orginality, you have to forbide copying/cloning. If some resonable suspicion comes up you have to investigate exactly that. (Maybe the investigation process can be improved, but the principle is correct).

Again: Making rules knowing that you cant realy check them is complete nonsense and is just supporting unjustice even more!

Bye
Ingo