A poll to gauge interest in possible tournament

[Roger Brown]

Why Gaviota, Scorpio, Booot etc. cannot use their own system?

Miguel

Hello Miguel,

I think I can almost spot the Winboard protocol engine authors and the UCI protocol engine authors.

My opinion is that a chess engine should be able to manage all phases of the game.

An opening book should be allowed, as should bitbases, tablebases and anything else that you, the author were able to wire into the thing.

The limiting factor would be single core versus multi-core engines (I can see Robert Hyatt cringing) but again, if the tournament is being played on a mutli-core machine and the setup allows use of two or more cores per engine, why should an author who coded smp be penalised by not being able to use all the available cores?

Each author should walk with everything fully loaded. UCI engines could access a generic, reputable book. I doubt that engine authors possess the skill set of book makers such as Sandro and Arturo so own books are not likely to be deep monsters, exiting in the end-game.

My belief is that book handling, pondering, knowledge of the game state all belong with the engine. Code as you like and put whichever aspect of that equation you want with the gui but then you should not insist that a level playing field requires that author A cannot use his own book etc.

Less than two cents...

Later.

Ps. Adam, please do as you like, it is your time and resources. Ignore my rantings if needed...

[Adam Hair]

There are far too many improvised "rating lists". Have a look at the tournament forum: almost every week a new one is created...

It's quality and consistency we need, not quantity. A proper website that attracts a large audience, and maintains a serious and well conceived rating list.

There is already CCRL, and Graham's awesome amateur divisions tournament. If you want to do another similar thing, it needs something extra to be relevant and/or interesting. For example:
* A website
* Strict and fair testing procedure, the likes of IPON.
* No clones. In fact only open source engines, that way it is easy to enforce and spot the cheaters. Closing the source is what allows clones to thrive.

CEGT has become rather messy and dispersed (too many lists at different tc with or without pondering).

I probably would not bother to compute any ratings. My focus would be much more on creating something that would be fun for authors and spectators and much less on controlled Elo measurement. I have been measuring engine attributes for 5 years, and I want to take a break from that.

As for the other points:

*There will be a website
*It will be conducted in a fair, consistent, and transparent manner
*Open source only would be too restrictive, IMO. Closed source is not synonymous with cloning. Besides, there are methods for spotting clear-cut clones. Engines that blur the line between original and derivative are much harder to detect, but, since the community has not decided on how to define that line, I am not going to insist on code purity. I will just require that engines exhibit a requisite amount of uniqueness in their move selection.

[Adam Hair]

I voted no. Apart from a quick glance now and then I did not follow WBEC, ChessWar and TCEC either. I run my own tournaments and that satisfies me as far as engines are concerned, also I do not have spare time to follow these tournaments. Anyway, I prefer to watch human tournaments where I have a little bit more chance to understand the ideas behind moves, also there are blunders that make them more interesting. And there is the rooting factor, I can root for a human but never for an engine.

Cheers,
Gábor

One reason why I test the also-ran engines is because I understand their moves, including the blunders. So, I understand what you are saying.

[Adam Hair]

When WBEC and ChessWar both ended in 2012, IMO the computer chess community lost two important proving grounds for chess authors. To be sure, there still exist some great tournaments such as TCEC, Graham's Amateur series, Geert's Country Chess, and Gsei's Web tournament to name a few. However, of those mentioned only Gsei's Web tournament uses own books, learning, multiple cores, and pondering. However, at the present time the Gsei tournament has a limited field. WBEC and ChessWar were expansive tournaments, running over 100 engines with each edition. Each of these tournaments allowed many engines to be run in competitive conditions and provided valuable feedback and enjoyment to the authors.

I do not think that I can replace either one of these tournaments. And I am not trying to compete with any existing tournament. However, I am interested in conducting competitive tournaments involving engines of all levels of strength, using everything built into each engine to play chess. I would like to gauge how interested the community is in my proposal. So please vote in the poll, and feel free to express your opinion in this thread.

A tour with ponder on, own book, learning, everything the engine is capable. No update of book and learning are allowed from the author and other persons, once the book is used the engine should update itself same with learnings. The engine is allowed some time to update itself after every game. A medium TC where spectators will not get bored , probably TC 20 minutes for whole game, no inc, no move control, there is excitement to this TC as engines are somewhat pressured to deliver a consistent move accuracy without being time pressured, everybody will improve their sloppy time management . Learning will be encouraged, an important subject that has been forgotten.

Choosing the time control will be the toughest choice. It does not help that my computer is dated (made in 2007).

[Adam Hair]

IMO ChessWar was by far the best engine tournament ever held, and it is a real pity it was discontinued. Olivier really knew how to get the best from any engine.

I am not sure that I could run a tournament as well as Oliver.

The live broadcasting and the chat were always very entertaining. It would be great if something similar could be revived.

I do think that live broadcasting and a chat is something I should attempt.

[Adam Hair]

Ps. Adam, please do as you like, it is your time and resources. Ignore my rantings if needed...

Done.

Why Gaviota, Scorpio, Booot etc. cannot use their own system?

Miguel

Hello Miguel,

I think I can almost spot the Winboard protocol engine authors and the UCI protocol engine authors.

My opinion is that a chess engine should be able to manage all phases of the game.

An opening book should be allowed, as should bitbases, tablebases and anything else that you, the author were able to wire into the thing.

The limiting factor would be single core versus multi-core engines (I can see Robert Hyatt cringing) but again, if the tournament is being played on a mutli-core machine and the setup allows use of two or more cores per engine, why should an author who coded smp be penalised by not being able to use all the available cores?

Each author should walk with everything fully loaded. UCI engines could access a generic, reputable book. I doubt that engine authors possess the skill set of book makers such as Sandro and Arturo so own books are not likely to be deep monsters, exiting in the end-game.

My belief is that book handling, pondering, knowledge of the game state all belong with the engine. Code as you like and put whichever aspect of that equation you want with the gui but then you should not insist that a level playing field requires that author A cannot use his own book etc.

Less than two cents...

Later.

I consider the tournament I am proposing to be competitive, so I think that anything an author has built into his/her engine for the purpose of playing chess should be included. Thomas mentioned not allowing anyone other than the tablebase creators to use particular tablesbases. However, I think that any author who has permission to use a particular tablebase format (and I believe everybody is allowed to use Gaviota, Scorpio, Syzygy, and Robbo bases) and has coded that ability into their engine should be able to use those tablebases.

[Richard Allbert]

A yes from me, good luck.

[Ferdy]

When WBEC and ChessWar both ended in 2012, IMO the computer chess community lost two important proving grounds for chess authors. To be sure, there still exist some great tournaments such as TCEC, Graham's Amateur series, Geert's Country Chess, and Gsei's Web tournament to name a few. However, of those mentioned only Gsei's Web tournament uses own books, learning, multiple cores, and pondering. However, at the present time the Gsei tournament has a limited field. WBEC and ChessWar were expansive tournaments, running over 100 engines with each edition. Each of these tournaments allowed many engines to be run in competitive conditions and provided valuable feedback and enjoyment to the authors.

I do not think that I can replace either one of these tournaments. And I am not trying to compete with any existing tournament. However, I am interested in conducting competitive tournaments involving engines of all levels of strength, using everything built into each engine to play chess. I would like to gauge how interested the community is in my proposal. So please vote in the poll, and feel free to express your opinion in this thread.

A tour with ponder on, own book, learning, everything the engine is capable. No update of book and learning are allowed from the author and other persons, once the book is used the engine should update itself same with learnings. The engine is allowed some time to update itself after every game. A medium TC where spectators will not get bored , probably TC 20 minutes for whole game, no inc, no move control, there is excitement to this TC as engines are somewhat pressured to deliver a consistent move accuracy without being time pressured, everybody will improve their sloppy time management . Learning will be encouraged, an important subject that has been forgotten.

Choosing the time control will be the toughest choice. It does not help that my computer is dated (made in 2007).

Provide a paypal donation link or something I am sure some will share.

[tpetzke]

Hi Adam,

your tournament, your rules. So this is just a personal opinion.

As an engine author I've done both. Created my own table bases and included the table base access code from Miguel into my engine, so I can compare the efforts related with both activities.

Including the probing code is rather trivial, there might be some tuning when to do a hard and when a soft probe, but this is nothing compared to writing your own tb system or coding chess endgame knowledge via rules into an engine.

If all engines access the tb code the diversity in engine play for that part of the game is gone. Those authors that invested the huge effort in creating their own system should get some reward.

Thomas...

[Adam Hair]

Hi Adam,

your tournament, your rules. So this is just a personal opinion.

As an engine author I've done both. Created my own table bases and included the table base access code from Miguel into my engine, so I can compare the efforts related with both activities.

Including the probing code is rather trivial, there might be some tuning when to do a hard and when a soft probe, but this is nothing compared to writing your own tb system or coding chess endgame knowledge via rules into an engine.

If all engines access the tb code the diversity in engine play for that part of the game is gone. Those authors that invested the huge effort in creating their own system should get some reward.

Thomas...

That is a compelling argument. Though, as a beta tester for Gaviota, I could be accused of being biased if I follow your recommendation