Chess960 in Mainz

[Graham Banks]

No need to worry - I am more interested in computer shogi than in computer chess at the moment, and I don't expect to spend much time on my chess program in the future.
Tord

That is a sad scenario for us computer chess folk.

Having said that, roll on the new Glaurung!

[pichy]

I still haven't heard anything about a computer tournament in Mainz this year, and I can't find any information about it on http://www.chesstigers.com. Does anyone know whether there will be a computer tournament this year?

Tord

This is the only news that I read about Mainz 2007:

http://www.chessbase.com/newsdetail.asp?newsid=3827

[Tony Thomas]

Tord, you better watch out, Vasik might come and try to steal your brain to avoid competition.

No need to worry - I am more interested in computer shogi than in computer chess at the moment, and I don't expect to spend much time on my chess program in the future.

Besides, I don't really consider myself a competitor to Vasik, nor to any other chess programmers. I am by nature not a very competitive person, and I am therefore not very interested in the competitive side of computer chess. Like everybody else, I like to see my own program improve, but I am also happy to see other programs improve more rapidly.

Tord

I was talking about a free competition. Provided that your program with no chess knowledge play as good or better than the current version, you can get another 140 points just by adding good knowledge, then by improving the SMP scaling, you can probably get an engine close in strength to rybka. We would love to have a competitive Tord. We dont care if you take the commercial route, we want someone else (Anthony was close) who can destroy Rybka's domination.

[Tord Romstad]

Hi Marc!

May I ask you a quick question about Glaurung 1.2.1?

You may, but in this particular case I'm afraid I don't have a very interesting answer.

I remember reading you on this forum stating that one can make Glaurung stronger in tactical test suite but with the side-effect of making it weaker in eng-eng matches. Could you tell me how to? (I'm interested in solving tactical test suite).

I don't recall exactly what I wrote, but I think I was speaking in very general terms, and not just about my own program. Everybody who has been doing chess programming for a while has experienced that the parameter settings which maximizes the scores on tactical test suites are rarely optimal for practical playing strength. Therefore, I and most other programmers never tune our programs for tactical test suites, and consequently we don't know what parameter settings perform best in test suites.

It is therefore quite likely that you know more than me about how to maximize Glaurung's scores in tactical test suites.

I'm asking because I realized that when I increase "style" parameters like aggressiness, developpment etc. results are better at quick time control, but for tougher problems requiring long searches (like 10min or more on a PIV 2.0), the boosted Glaurung doesn't do as well, the default settings are doing better.

This is interesting and somewhat surprising, and I am not sure why it happens.

So I'm looking for parameters that would do good tactically at long time control ( I don't expect you to have an answer for that).

Unfortunately, I don't.

Good luck with the new version,

Thanks!

Tord

[Tord Romstad]

I was talking about a free competition. Provided that your program with no chess knowledge play as good or better than the current version, you can get another 140 points just by adding good knowledge, then by improving the SMP scaling, you can probably get an engine close in strength to rybka.

140 points will require a lot of effort, I think: Much more than I expect to be able to do in the next year or two. My new program does have some chess knowledge, of course, just not as much as I would like. I expect that adding all the knowledge from Glaurung 1.x will not give an tremendous improvement in strength, but mainly a more attractive style.

We would love to have a competitive Tord. We dont care if you take the commercial route, we want someone else (Anthony was close) who can destroy Rybka's domination.

That won't be me, I'm afraid.

Tord

[Tord Romstad]

No need to worry - I am more interested in computer shogi than in computer chess at the moment, and I don't expect to spend much time on my chess program in the future.
Tord

That is a sad scenario for us computer chess folk.

Not necessarily. You can just learn to play shogi, start the CSRL, and have twice as much fun as before.

Of course, Glaurung will still be developed, just more slowly than in the past.

Tord

[Dariusz Orzechowski]

Of course, Glaurung will still be developed, just more slowly than in the past.

Ah, I breathed a sigh of relief. I'm a fan of your engine since times of good old Gothmog and I follow all your endeavours with great interest. I wish you good luck with your shogi engine! I would be happy if you sometimes lift the veil of secrecy and write something about your shogi ups and downs here.