Houdini 4 has been released

[lkaufman]

I am not claiming in general, I am assuming (not claiming again) for Houdini 4. Even though Robert stated somewhere that H dev has 0 contempt, from TCEC games in many positions I realized that there was still some contempt present and I believe the same for H4 (I don't have it myself yet to test).

I've just got a confirmation that default H4 contempt is 1, so my guess about 5-10Elo less with 60-80Elo stronger opponents is quite correct.
There you go Larry, you now have complete explanation .
P.S. I'm sure if Stefan ran H4 with contempt 0 at his list (top 10 opponents) he would get better result than with default one.

Why do you think h4 would do better on LS with contempt 0? Presumably that would be true against near-equal opponents, but not necessarily against opponents averaging quite a bit below h4.

[pohl4711]

I am not claiming in general, I am assuming (not claiming again) for Houdini 4. Even though Robert stated somewhere that H dev has 0 contempt, from TCEC games in many positions I realized that there was still some contempt present and I believe the same for H4 (I don't have it myself yet to test).

I've just got a confirmation that default H4 contempt is 1, so my guess about 5-10Elo less with 60-80Elo stronger opponents is quite correct.
There you go Larry, you now have complete explanation .
P.S. I'm sure if Stefan ran H4 with contempt 0 at his list (top 10 opponents) he would get better result than with default one.

Why do you think h4 would do better on LS with contempt 0? Presumably that would be true against near-equal opponents, but not necessarily against opponents averaging quite a bit below h4.

I did a test with some-thousand games with Houdini 3 with contempt=0 (LS-conditions) and the result was around 15 Elo weaker than default. And in Marijan's one minute ratinglist (in the Immortal-Forum) Houdini 3 is 15 Elo weaker than Houdini 3 default, too...
I believe, contempt=0 would be better, if the opponents were somewhat stronger than Houdini itself, but at the moment no engine is stronger than Houdini 4 (playing with LS-conditions).

Stefan

[A Distel]

Mogen we een 'ChessBase' versie van 'Houdini 4 (DVD)' verwachten?

ChessBase and ChessOK packages should be available around mid-December.

Cheers,
Robert

Thank you for the information.

Regards,

Alain.

[Milos]

I did a test with some-thousand games with Houdini 3 with contempt=0 (LS-conditions) and the result was around 15 Elo weaker than default. And in Marijan's one minute ratinglist (in the Immortal-Forum) Houdini 3 is 15 Elo weaker than Houdini 3 default, too...
I believe, contempt=0 would be better, if the opponents were somewhat stronger than Houdini itself, but at the moment no engine is stronger than Houdini 4 (playing with LS-conditions).

Stefan

You are problably right 10 opponents in LS-conditions are all weaker from 50-130 Elo than H4 that's the reason it's still weaker with contempt 0.
In selftesting H3 contempt 0 vs. H3 contempt 1 difference is usually 15Elo for contempt 0. When playing against recent SF H3 comtempt 0 performs 20-30Elo better than contempt 1 depending on the TC.
With 100Elo weaker opponents it is -15Elo so I guess crossing point would be something about 30-50 Elo weaker opponents where both contempt 0 and contempt 1 have same strength.

@Larry to come back to your question, I do believe that average opponent strenght difference of 100Elo can play a significant role with respect to contempt, and that you see at least 5-10 Elo worse results when opponents are noticably stronger.
Just as a direct comparison you can see Gergard's results:
http://cegt.forumieren.com/t41-testing-houdini-40-x64
where H4 is performing even better compared to H3 despite longer TC. The reason is I firmly belive that opponents there are weaker than what Robert has in his testing and substentially weaker to what Stefan has in his rating list.

[E.G.R.]

I have problems with houdini pro 4 in Aquarium as IDEA...freezes and slows down everything. Else anyone?

[hgm]

A contempt different from 0 performing better is usually an artifact of biased testing, using many more opponents on one side of the engine than on the other. One should always pit the engine against a symmetric distribution of opponents, or the tests would become unreliable, and very dependent on the Elo model used.

Note that it is always possible to create stronger opponents by giving the opponents more time. Best is to test against set of opponents that score from 25% to 75% against the engine under test, equally distributed over that range.

[Milos]

A contempt different from 0 performing better is usually an artifact of biased testing, using many more opponents on one side of the engine than on the other. One should always pit the engine against a symmetric distribution of opponents, or the tests would become unreliable, and very dependent on the Elo model used.

Note that it is always possible to create stronger opponents by giving the opponents more time. Best is to test against set of opponents that score from 25% to 75% against the engine under test, equally distributed over that range.

Good luck in trying to explain something like that to rating lists testing H, K or SF .

[hcn]

I have problems with houdini pro 4 in Aquarium as IDEA...freezes and slows down everything. Else anyone?

Yes. There is a problem. It seems it has been identified, and a patch has been promised for tomorrow.
You can read the relevant thread in the Aquarium section in the rybka forum.

[daniel71]

Does anybody yet know if Chessbase will support the 6-men Syzygy table bases using the Fritz 14 GUI? It would seem a poor choice if Chessbase neglected to support these new tablebases and Robert at this point is in the
drivers seat to have support implemented! What are the plans for 7-piece Syzygy bases? I purchased the Convekta Chess Assissiant 13 Houidini Pro 3
for the Lomonosov Endgame Tablebase support and now we have another endgame tablebase, my head is starting to hurt. Daniel

[shrapnel]

On the other hand, with the Syzygy system the improvement is clear even at 1'+1".

Robert

Its an indirect answer to my question, but I suppose that's the best I can expect.
It seems logical that if the improvement is there at even short time controls, there should be proportionately more improvement at LTCs, compared to Nalimov (hopefully).
So, Syzgy it is.