ICGA WCCC prospects ...

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
guyhaw

ICGA WCCC prospects ...

Post by guyhaw » Fri Jan 02, 2009 3:29 pm

I am concerned to hear that David Levy is likely to impose constraints on the WCCC at short notice, especially as the likelihood is that consultation will not make much of a difference, given the 'power of the first draft' of his intentions.

In yachting, we have 'Line Honours' prizes and 'Handicap' prizes, the first to encourage innovation and the second to give a level playing field regardless of innovation.

Why not allow as many cores as people like, equating 4x the number of cores with 1 extra ply of search, and ELO-handicapping accordingly against some concept of 'extra ply = N more ELO at depth of search D'.

Guy

kiroje

Re: ICGA WCCC prospects ...

Post by kiroje » Fri Jan 02, 2009 6:59 pm

guyhaw wrote:Why not allow as many cores as people like, equating 4x the number of cores with 1 extra ply of search, and ELO-handicapping accordingly against some concept of 'extra ply = N more ELO at depth of search D'.
I think the problem is that not all count ply as the same

diep
Posts: 1780
Joined: Thu Mar 09, 2006 10:54 pm
Location: The Netherlands
Contact:

Re: ICGA WCCC prospects ...

Post by diep » Sun Jan 04, 2009 6:21 pm

guyhaw wrote:I am concerned to hear that David Levy is likely to impose constraints on the WCCC at short notice, especially as the likelihood is that consultation will not make much of a difference, given the 'power of the first draft' of his intentions.

In yachting, we have 'Line Honours' prizes and 'Handicap' prizes, the first to encourage innovation and the second to give a level playing field regardless of innovation.

Why not allow as many cores as people like, equating 4x the number of cores with 1 extra ply of search, and ELO-handicapping accordingly against some concept of 'extra ply = N more ELO at depth of search D'.

Guy
I spoke with Marc Uniacke online and it seems to me that the years 90 fear for hardware is big. No matter whether that destroys computerchess forever. The thing in computerchess now is that the biggest progress of the past years is difficult to understand. Took me also years to understand, as it was an area where in past i didn't need to invest in.

It is easy then to look at the only clear difference, which is hardware.

Yet if you look back in history, in the 90s, hardware mattered more, and the micro world championships from the past won't get back. Besides that i remember a world championship 1997 where money really brought bigger speed. Remember Bruce's Ferret run on 767Mhz alpha 64 bits chip together with the Peter Gillgasch alpha assembler programmed Darkthought? That against P2-300Mhz (which also were not officially released yet by intel, yet everyone got one for free from generous sponsors).

We live now in a time that there is no longer 30 commercial chessprograms, there is no serious money to make with computerchess engines. A few GUI and 'course/video' delivering companies such as chessbase can still sell somehow at far too big prices for their products, but that's about it.

So the big interest of programmers to show up at a world championship has definitely gone.

If others than me show up with big hardware, that is one of the few things that makes the world champs interesting. First of all it gives you a rather good excuse in case you lose. Yet if you win, and my hardware most likely will be inferior to what hiarcs has, it is even more clear how little hardware says nowadays. As long as you have some serious time control that is...

At blitz of course it is obvious.

This world championship seems to get shaped after a single voice out of UK, who hides nowadays under the name of a reporter (why is that?) here. I find that personally very bad.

Especially because in the 2002 world champs the vaste majority, and that majority has not changed, of all programmers, wanted to get rid of the microworldtitle and open up hardware limits every year, instead of once each 3 years.

The thing is, that same UK program so in favour for this limit, also doesn't show up for years in any tournament when it played a tad less.

I remember very well Uniacke. Aegon tournament. One of the last big events out of the 90s. Anand gave a simultaneously exhibition. Hiarcs managed to get a drawn rook endgame against Anand. It was total dead draw and in a total easy manner. Clearly annoyed, as it was the last board, Anand played on. Entire audience saw that Anand was annoyed, just not the hiarcs operator. He played on fanatically.

Then for years it didn't show up. Now for 1 year it is strong, yet not the best. By world champs 2009 other engines will have gained a lot in playstrength as well. Hiarcs coming years won't get much stronger, maybe in 2011 it will get a boost again (certain programmers have proven over time of course that each few years they can improve). Others will make up for a lot though. It is again in the same position like it was a few years ago when it was real strong, in a local maximum.

Logically that means it joins now and then for a few years it won't.

Now based upon the opinion of 1 person, Levy is taking this decision?

Reality learns that in 2010 Hiarcs might not join anymore when other engines manage to improve their chessknowledge a tad, whereas in 2009 the majority of the programmers get overruled by just 1 proposal.

If i counted well, Stefan MK doesn't mind a 8 core limit in 2009,
even though he would have a major hardware advantage in 2009 world champs, if he manages to improve his parallel code a tad.

Rybka has a very bad parallel algorithm, so i wonder why they fear it. Maybe because they trust tips from Anthony Cozzie might improve it a lot, as he already did do a good parallel search on a supercomputer?

I'd say the more dubious you search the less your profit towards bigger hardware.

Yet the proposal came from Marc, or whatever name he has now.

This is however not about a 8 core limit. This is about the manner the political decision taking takes place. I wonder how many will show up now in that nation.

My proposal to ICGA is very clear however. Don't pay for operators.
I don't want people who hide that much to put such a big stamp onto a tournament where obviously hardware has become too complex to impose limits out of the past which are impossible to control, given the manner in how the ICGA normally operates.

It makes the temptation to cheat just too big, that is the second problem it creates.

In other sports at toplevel, about 100% of all sportmen and sportwomen use dope. Just they try to hide it, or go to the maximum 'allowed' and swallow everything to get everywhere to the maximum.

Famous is the cyclists in the tour the france who had to get awakened every 2 hours by their coach as otherwise they would possible die, because their blood was too thick. They all got that call. That's what EPO is doing.

In chess obvious is each yearly blitz event at ICC. Dos Hermanas. As i speak with several GM's, i can assure you that from the past years, the entire top30 used a computer. No exceptions. Just some are more clumsy in using it than others, that's all. If you make it too easy to cheat, consider it done.

In all this the computerchess nerds are quite honest guys in comparision. In other sports where social mature people play, there is no such thing as honesty.

Now the question is whether it is clever in todays computerchess to impose such childish limits knowing that it is going to be very hard to control it.

The only solution to that problem would be local machines in the tournament hall that have no connection to the internet at all. One has to check for wireless connections regurarly at the machines, yes during tournament play type nerd commands like that on the machine.

None of the ICGA officials knows how to do that.

They run far more behind than the dope control of the olympic organisations.

There is no other solutions than to put machines locally in the playing hall.
Now a few years ago Bruce was already pissed he had to pay 500 dollar extra in USA to transport his computer by air. He was amazed in Netherlands he didn't need to pay that fee. The dutch were far more generous to him there. This cheapness all airlines have now.

A few years ago interests were much bigger than today to pay for those extra costs than today.

Is it clever/smart/wise to impose more costs upon the amateurs of today?

Vincent

Spock

Re: ICGA WCCC prospects ...

Post by Spock » Sun Jan 04, 2009 7:19 pm

diep wrote: This world championship seems to get shaped after a single voice out of UK, who hides nowadays under the name of a reporter (why is that?) here. I find that personally very bad.
diep wrote: Now based upon the opinion of 1 person, Levy is taking this decision?

Do you mean Harvey of the Hiarcs team ?

diep wrote: This is however not about a 8 core limit. This is about the manner the political decision taking takes place. I wonder how many will show up now in that nation.
Politics impacts almost all areas of our daily lives unfortunately.

User avatar
Mike S.
Posts: 1480
Joined: Thu Mar 09, 2006 4:33 am

Re: ICGA WCCC prospects ...

Post by Mike S. » Sun Jan 04, 2009 7:48 pm

guyhaw wrote: Why not allow as many cores as people like, equating 4x the number of cores with 1 extra ply of search, and ELO-handicapping accordingly against some concept of 'extra ply = N more ELO at depth of search D'.
That seems not possible in practice. World Champion is the one who wins the tournament, is is on place #1 in the end. Also, your proposal would need to involve very rough estimations. Should such approximative math decide about a WCh. title? I don't think so.

I support Levy's decision (which I hope it is, in fact) but as always, I see things from the user & fan viewpoint. It would not a bad decision though, if programmers would share exactly that viewpoint, except somebody works only for (or against) the other programmers, or for academic fame in a small circle. - But I know that there are many super technology freaks who have very much trouble now, due to this 8 cores limit.

Other people have explained that 8 cores are still too many, because a good 8 core computer is MUCH more expensive than a good quad, which means that money is still a big factor of success in that competition.

To put it short, money (or insider's relations to hardware labs with latest CPUs and other privileges like that) should be banned, and the hardware limit is a big step forward in this direction. Bravo Levy!

Note, that at the same time he announced that at the computer olympiads, participation of unlimited systems will be possible. But not at the WCCC which is about computer chess, not about (all purpose) hardware, and not about money. Otherwise, maybe you would not even need to play chess. Just bring all your cash, and the one who has the biggest pile, is champion.

Also, I would to mention that uniform platform was not a problem at all, at the Chess960 computer world championships in Mainz, 2005-2008. Did anyone NOT join them because the hardware was limited?! I don't think so. It's a prove that it can be done, and the software competition is more meaningful under such circumstances.

I would even standardize and/or limit the books, but I am afraid that is a too radical proposal. Nevertheless, I wonder why some seem to mistrust their chess software creations so much, that they need to put their hopes in super hardware and super tuned opening books, instead.
Regards, Mike

Spock

Re: ICGA WCCC prospects ...

Post by Spock » Sun Jan 04, 2009 8:13 pm

Mike S. wrote: Also, I would to mention that uniform platform was not a problem at all, at the Chess960 computer world championships in Mainz, 2005-2008. Did anyone NOT join them because the hardware was limited?! I don't think so. It's a prove that it can be done, and the software competition is more meaningful under such circumstances.
There is a difference between equal hardware for all and a hardware *limit*

With this new hardware limit, some contestants will still be disadvantaged. It achieves nothing. Those that come with an 8-core will still have an advantage over those that come with a dual. I know Harvey said that when he had to explain to people that the winner was not necessarily the best software, due to hardware differences, people became confused and lost interest, and his push for publicity came to nothing. Well, nothing has changed in that regard, the new rule is useless, his push for publicity will still come to nothing if that is true.

Either unlimited hardware, or equal hardware for all. And as some have pointed out, equal hardware is of no interest to a lot of people because it doesn't add much over and above a tournament they can play themselves.

But the decision is made, I feel weary about talking about it.

pijl

Re: ICGA WCCC prospects ...

Post by pijl » Sun Jan 04, 2009 9:18 pm

Mike S. wrote: Also, I would to mention that uniform platform was not a problem at all, at the Chess960 computer world championships in Mainz, 2005-2008. Did anyone NOT join them because the hardware was limited?! I don't think so. It's a prove that it can be done, and the software competition is more meaningful under such circumstances.
In 2005 and 2006 the hardware was not limited. I joined with an 8 core machine, remote in 2005, and brought my own quad in 2006.
Since 2007 the competition was limited to 4 programs and only then it was uniform hardware.
In 2008, the qualifying step on ICC was unlimited hardware again. Just the finals in Mainz were uniform platform.
Richard.

Nick C

Re: ICGA WCCC prospects ...

Post by Nick C » Sun Jan 04, 2009 9:37 pm

I would even standardize and/or limit the books, but I am afraid that is a too radical proposal. Nevertheless, I wonder why some seem to mistrust their chess software creations so much, that they need to put their hopes in super hardware and super tuned opening books, instead.

I don't think that limiting books will happen. David Levy wrote that the decision to limit cores was to try to pull back Rybka. Limiting books would give Rybka an even bigger advantage over the competition. Right now, the best way for an opponent to beat or draw with Rybka is to use a wide and deep book created at home using Rybka with their own engine at the WCCC.

The argument for limiting cores has been some nebulous connection to what some individuals think that "the public" want. Ie., that they want to see a chess engine win on hardware that they can purchase. The public are not going to get the private books that make the difference in these events, so that is a null argument anyway.

The whole thing is just plain silliness.

User avatar
Harvey Williamson
Posts: 1820
Joined: Sun May 25, 2008 9:12 pm
Location: Media City, UK
Contact:

Re: ICGA WCCC prospects ...

Post by Harvey Williamson » Sun Jan 04, 2009 9:56 pm

Nick C wrote:
I would even standardize and/or limit the books, but I am afraid that is a too radical proposal. Nevertheless, I wonder why some seem to mistrust their chess software creations so much, that they need to put their hopes in super hardware and super tuned opening books, instead.

I don't think that limiting books will happen. David Levy wrote that the decision to limit cores was to try to pull back Rybka. Limiting books would give Rybka an even bigger advantage over the competition. Right now, the best way for an opponent to beat or draw with Rybka is to use a wide and deep book created at home using Rybka with their own engine at the WCCC.

The argument for limiting cores has been some nebulous connection to what some individuals think that "the public" want. Ie., that they want to see a chess engine win on hardware that they can purchase. The public are not going to get the private books that make the difference in these events, so that is a null argument anyway.

The whole thing is just plain silliness.
David Levy wrote that the decision to limit cores was to try to pull back Rybka.
Where did he write that?

User avatar
Dr.Wael Deeb
Posts: 9635
Joined: Wed Mar 08, 2006 7:44 pm
Location: Amman,Jordan

Re: ICGA WCCC prospects ...

Post by Dr.Wael Deeb » Sun Jan 04, 2009 10:00 pm

Mike S. wrote:
guyhaw wrote: Why not allow as many cores as people like, equating 4x the number of cores with 1 extra ply of search, and ELO-handicapping accordingly against some concept of 'extra ply = N more ELO at depth of search D'.
That seems not possible in practice. World Champion is the one who wins the tournament, is is on place #1 in the end. Also, your proposal would need to involve very rough estimations. Should such approximative math decide about a WCh. title? I don't think so.

I support Levy's decision (which I hope it is, in fact) but as always, I see things from the user & fan viewpoint. It would not a bad decision though, if programmers would share exactly that viewpoint, except somebody works only for (or against) the other programmers, or for academic fame in a small circle. - But I know that there are many super technology freaks who have very much trouble now, due to this 8 cores limit.

Other people have explained that 8 cores are still too many, because a good 8 core computer is MUCH more expensive than a good quad, which means that money is still a big factor of success in that competition.

To put it short, money (or insider's relations to hardware labs with latest CPUs and other privileges like that) should be banned, and the hardware limit is a big step forward in this direction. Bravo Levy!

Note, that at the same time he announced that at the computer olympiads, participation of unlimited systems will be possible. But not at the WCCC which is about computer chess, not about (all purpose) hardware, and not about money. Otherwise, maybe you would not even need to play chess. Just bring all your cash, and the one who has the biggest pile, is champion.

Also, I would to mention that uniform platform was not a problem at all, at the Chess960 computer world championships in Mainz, 2005-2008. Did anyone NOT join them because the hardware was limited?! I don't think so. It's a prove that it can be done, and the software competition is more meaningful under such circumstances.

I would even standardize and/or limit the books, but I am afraid that is a too radical proposal. Nevertheless, I wonder why some seem to mistrust their chess software creations so much, that they need to put their hopes in super hardware and super tuned opening books, instead.
How about limiting/equaling the source code of the programs,I mean all the programs to have the same number of code lines,the same compiler,etc.,I would even go farther....all the participants must wear standard uniforms and have the same hair cuts....
Sorry for the irony Mike,but it's getting a trivial issue and you'll see how much interest the new WCCC will get with these brainless limitations....
Last edited by Dr.Wael Deeb on Sun Jan 04, 2009 10:04 pm, edited 1 time in total.
_No one can hit as hard as life.But it ain’t about how hard you can hit.It’s about how hard you can get hit and keep moving forward.How much you can take and keep moving forward….

Post Reply