ICGA Forum

[CThinker]

Harvey quotes a statement of Mark Uniacke in the ICGA forum:

http://www.grappa.univ-lille3.fr/icga/p ... rt=20#p253

Hi Rémi,

I am not member of the CCC group or an active reader of the CCC forum but I have heard about these negative comments. In the one post I was told about I saw Vincent Diepeveen claimed he spoke to me online and that the proposal of a hardware limit came from me - both of these claims by Vincent are a complete fabrication. I prefer to not get dragged into all the mud slinging that goes on in these forums since it drains the energy from what is important but I needed to point that out to you in case you were under the wrong impression.

I have to say that if the production of violent anger is the driving force in changing decisions then we have mob rule and intimidation. Surely these are not the attributes on which to change or make decisions.

The first I heard of this hardware limit idea was when I receieved the general email from Johanna Hellemons about the proposal. I did respond in favour of it and I remain in favour of it. Judging from what you have said perhaps that is enough to get me lynched by some?

I don't think we can base a decision change on the fact one or two programmers may want to invest huge resources into expensive hardware since were we to do that the emphasises would be on any other programmers to do the same if they to want to compete. Otherwise this could be seen as being akin to someone trying to buy the title.

What is the purpose of the WCCC?

Surely it is to have a competition between the world's top chess systems for the title of World Champion. If the disparity between the hardware of each "competitor" is so great that the chances of success for most competitors is almost nil it no longer becomes a competition but instead a procession.

Recently we have seen the long standing Paderborn tournament cancelled because there were insufficient entries. Why would this be?

I believe it is because the overwhelming majority of people see little chance of competing on a level playing field because the hardware being used by the strongest programs is now so much faster than the other potential competitors that the race is practically run before it has started.

Contrary to popular opinion selling chess software makes a very limited income and certainly not one that can allow us to compete in a hardware race. The same is true of nearly all the other commercial authors let alone the amateur ones. Perhaps those in academia might be able to harness the necessary resources but for the majority it is only a dream.

In the last two tournaments alone, in terms of hardware, even though we were on faster h/w than many others we were about a factor of 5 behind some other contestants. Eventually we like apparently everyone else has to ask if it is worth the cost and effort of trying to maintain a presence in an expensive losing "hardware arms race".

We enter each competition with the target to win it, if that no longer becomes a possibility then we have to ask what is the point of entering. Apparently the same is true of the vast majority of other programmers since they are voting with their feet and not entering these tournaments.

We are willing to give up our h/w advantage to see a more level playing field since I believe that makes for a more exciting and competitive competition. Some disagree with me and prefer no limits but in the knowledge that competition is stifled.

I think Formula-1 racing is a good example with commonality to our situation. Each team can select their own driver, engine, chassis and other components within a framework of rules. That framework is reviewed on an on-going basis just as the ICGA has undertaken to review this decision. The purpose of the framework in F1 is to make the competition more competitive and allow more teams to compete than would be the case if there were no limits. Surely this is a good thing?

Best wishes,
Mark

This use of F1 as an example is wrong.

The F1 is not limited to even out the playing field. It is limited for safey reasons. All the teams are allowed to put in the most expensive car that meets the safey rules of the car.

http://www.atlasf1.com/2000/bra/preview/ward.html

Code: Select all

As each area of car design reaches a peak or plateau of performance, this performance is limited and controlled - even banned completely - by the F1 technical regulations to ensure that the cars do not become so fast and powerful that they become overly dangerous or even undriveable. The fundamental principle is that performance should not exceed safety standards and the limits of drivers' cognition and response times. 

Just look at the budget of each team in F1. Do you see a limit?
http://www.f1fanatic.co.uk/2008/09/22/t ... get-4456m/

Code: Select all

Toyota: $445.6m
McLaren: $433.3m
Ferrari: $414.9m
Honda: $398.1m
Renault: $393.8m
BMW Sauber: $366.8m
Red Bull Racing: $164.7m
Williams: $160.6m
Toro Rosso: $128.2m
Force India: $121.85m
Super Aguri: $45.6m

The fact that there are only 11 teams in F1 tell you that this is not for everyone. Certainly not for the common home car.

In contrast, they want to limit the WCCC to the home PC.

OK, now if they really want a limit, and they are using F1 as an example, then the limit should be like "the machine should not consume 1GW of power - or it will be unsafe to the operators".

[bob]

On the topic of hand interfacing, I can understand it a bit.

The problem is three fold: some support UCI, others support
xboard/winboard and others have proprietary interfaces.
The UCI to Winboard converters are insufficient and
so are the reverse. There would be a "home court" advantage if one
interface is picked. It seems that the best meeting point is hand
interfacing for engines supporting differing protocols. However,
people could use servers when opponents support the same protocols,
but this requires obtaining support from two servers (say ICC and
playchess).

Hand interfacing must remain an option to allow people to develop
on any platform they wish such as cell phones....

The lack of sufficient protocol converting may be the reason
many UCI engines don't support CCT and ACCA.

I don't see the advantage you mention. Chess servers are "protocol independent" from the perspective of WB vs UCI vs private. I am not aware of any commercial program today that can't connect to ICC, which makes everything work just fine with their native GUIs...

I also disagree about hand interfacing being necessary. We don't want cell phone programs in a WCCC any more than NASCAR wants a go-cart in their races.

As far as CCT and ACCA, who has been unable to participate? We have had every commercial program available enter at some point in time... the amateurs that can't deal with winboard or UCI simply ought not be allowed to play. That's impetus for them to fix this. You want to connect to the internet, you have to use an accepted protocol or else do without. I'd view this _major_ chess event the same way.

[bob]

When I looked, they didn't play in the last 3 WCCC events and I decided to quit looking at that point. So they take advice from non-participants, over objections of past participants?

Hiarcs participated in Bejing 2008 and finished second, 7.0/9. Also, Hiarcs was (micro-) world champion in 1993.

My point was that it did not compete in the three WCCCs prior to that year, which hardly makes satisfying their desire for restricted hardware a justifiable action...

http://www.grappa.univ-lille3.fr/icga/program.php?id=53

Harvey quotes a statement of Mark Uniacke in the ICGA forum:

http://www.grappa.univ-lille3.fr/icga/p ... rt=20#p253

I have read that and about 1/2 of it is drivel. The WCCC has _always_ been "the best of the best". Now it is "the best of the mediocre"... a far cry from the intent when _we_ formed the ICCA in 1977 in Toronto.

[bob]

When I looked, they didn't play in the last 3 WCCC events and I decided to quit looking at that point. So they take advice from non-participants, over objections of past participants?

Hiarcs participated in Bejing 2008 and finished second, 7.0/9. Also, Hiarcs was (micro-) world champion in 1993.

http://www.grappa.univ-lille3.fr/icga/program.php?id=53

Harvey quotes a statement of Mark Uniacke in the ICGA forum:

http://www.grappa.univ-lille3.fr/icga/p ... rt=20#p253

In fact over the last few years we have played in almost everything. Leiden, Paderborn, CCT, WCCC Beijing, Thüringen etc.... Winning at Paderborn and Thüringen and always 2nd or 3rd in 5 consecutive leiden tournaments . So I guess Bob did not look that hard. We had entered the recently cancelled Paderborn tournament.

No I didn't because we were talking about the WCCC _explicitly_, not any of the other events. Because this rule change _only_ affects the WCCC event.

We are also currently entered in the CSS Masters and King of Engines tournaments on Playchess. The only other engine playing latest betas in these is Sjeng.

Again, who cares? We are discussing what is supposed to be _the premier event_. Not just another tournament.

[Mike S.]

I also disagree about hand interfacing being necessary. We don't want cell phone programs in a WCCC any more than NASCAR wants a go-cart in their races.

If you are convinced that an upper limit is not suitable, it is difficult to understand why you would accept a rule which introduces a lower limit. - Although, there are mobile phone chess applets (or at least one) which can access ICC or FICS, basically:

http://www.luthman.nu/Mobi/Mobi.html

As for ICC access, the commercial interfaces of Fritz/CB. and Shredder Classic do not have it in current versions, themselves. If that software is used for ICC games automatically, I guess it must work with additional, external adapters "somehow..." (Fritz, since version 7 includes the client for CB's Playchess server, but not for other servers).

If CCT and ACCA tournaments are very popular among programmers and their cooperators, I am nevertheless afraid that they are not popular beyond this group, in a way the WCCC is. Which is a pity; I have written a couple of news entries and postings about such events in the past but got almost no response.

[Tord Romstad]

And btw, just recently, Glaurung on iPhone has started playing on ICC.

Not quite. Glaurung on the iPhone with XBoard has been playing on the ICC. The difference, to me, is quite significant. Using XBoard for some automated testing on the ICC is OK, but I wouldn't use it in a tournament, because it wasn't written by me.

I am probably more old-fashioned than most here: I find fully automated tournaments rather dull, and tournaments on chess servers extremely dull. The game loses most of its charm unless I and my opponent manually move real wodden pieces on a real chess board. The CCT is not for me.

Tord

[bob]

I also disagree about hand interfacing being necessary. We don't want cell phone programs in a WCCC any more than NASCAR wants a go-cart in their races.

If you are convinced that an upper limit is not suitable, it is difficult to understand why you would accept a rule which introduces a lower limit. - Although, there are mobile phone chess applets (or at least one) which can access ICC or FICS, basically:

This is "the world computer chess championship". We don't need ipods, iphones, and such. They have absolutely no chance, and just add clutter. There probably should be a different tournament for them, very similar to what the NHRA does for small displacement cars. There is no point in pulling your SRT-4 turbo Neon up beside a top fuel dragster that can break 300mph in the quarter.

http://www.luthman.nu/Mobi/Mobi.html

As for ICC access, the commercial interfaces of Fritz/CB. and Shredder Classic do not have it in current versions, themselves. If that software is used for ICC games automatically, I guess it must work with additional, external adapters "somehow..." (Fritz, since version 7 includes the client for CB's Playchess server, but not for other servers).

If CCT and ACCA tournaments are very popular among programmers and their cooperators, I am nevertheless afraid that they are not popular beyond this group, in a way the WCCC is. Which is a pity; I have written a couple of news entries and postings about such events in the past but got almost no response.

[CThinker]

And btw, just recently, Glaurung on iPhone has started playing on ICC.

Not quite. Glaurung on the iPhone with XBoard has been playing on the ICC. The difference, to me, is quite significant. Using XBoard for some automated testing on the ICC is OK, but I wouldn't use it in a tournament, because it wasn't written by me.

I am probably more old-fashioned than most here: I find fully automated tournaments rather dull, and tournaments on chess servers extremely dull. The game loses most of its charm unless I and my opponent manually move real wodden pieces on a real chess board. The CCT is not for me.

Tord

I remember a friend of mine commenting on the Deep Blue games, "I feel really sorry for that guy doing the moves for Deep Blue".

I too like pushing wooden pieces, but only if I am the one playing.

If the two best GMs are playing, and I would push the pieces for them, I would not do it. That would be really dull.

If I'm not playing, I should just sit back, watch, chat, analyze.

But hey, that's just me.

[bob]

And btw, just recently, Glaurung on iPhone has started playing on ICC.

Not quite. Glaurung on the iPhone with XBoard has been playing on the ICC. The difference, to me, is quite significant. Using XBoard for some automated testing on the ICC is OK, but I wouldn't use it in a tournament, because it wasn't written by me.

I am probably more old-fashioned than most here: I find fully automated tournaments rather dull, and tournaments on chess servers extremely dull. The game loses most of its charm unless I and my opponent manually move real wodden pieces on a real chess board. The CCT is not for me.

Tord

I remember a friend of mine commenting on the Deep Blue games, "I feel really sorry for that guy doing the moves for Deep Blue".

I too like pushing wooden pieces, but only if I am the one playing.

If the two best GMs are playing, and I would push the pieces for them, I would not do it. That would be really dull.

If I'm not playing, I should just sit back, watch, chat, analyze.

But hey, that's just me.

My first computer vs computer event was the 1976 ACM tournament. I though that a human, emulating an I/O device for the computer, was about the worst thing that could happen. You forget to hit the clock. your program moves while your head is turned or while you are away visiting the restroom or getting something to drink, causing you to lose time. You, or your opponent makes the wrong move on the board causing the game to be interrupted once the error is discovered. The list goes on and on.

By 1980 I had built an electronic chess board that detected the opponent's move automatically, detected when the clock was pressed by the opponent, detected when Cray Blitz made a move and sounded an alarm until I physically completed the move on the board and pressed the clock. And then Ken Thompson and I decided that this should all be replaced by a centralized "referee" machine that relayed moves back and forth automatically as well as mainained the clock accurately. In 1995 I discovered ICC (ICS at the time) and found that this had already been done despite the ACM and ICCA continuing to use manually operated conditions.

I would not want to go back under any circumstances. I have been to "up close and personal events" such as the last ACCA where several of us met in a centralized location. And we had a _far_ better time than we used to have at the manually operated events. No worries about "lost time" or "wrong moves". No worries about entering the remaining time incorrectly. We got to sit back, and watch the games, and even watch other games if our own games were not interesting, with no pressure or worry involved. I would only go back if forced to. Say by attending a dark ages WCCC event.

I can't imagine playing any other way than via automated today. Otherwise, why not dump the computer and play yourself, rather than relegating your self to being an I/O peripheral device...

[diep]

Well Levy is wrong. In case of paderborn: first professor Monien was there. Also responsible for the advances in ICCA book from 1999. There is no longer computerchess guys working in Paderborn within a few months from now, so the decision was to try to integrate the computerchess event into the chess event; let the chess organizers also organize the computerchess event.

However this chess event plays between christmas and new year. For paid GM's and a number of chessplayers that is not a problem. For the computerchess guys this is a big problem in general.

There used to be a maximum of 16 participants over there, which might not be there right now anymore.

Yet both events, CSVN events as well as Paderborn, had another big problem. That is they get announced way too late. Also the guys who took over from the previous organizer seem to have this habit. Of course they have a good excuse that 'taking over' takes some time as health of course of Theo v/d Storm is in a final stage.

In case of paderborn:

If you go search a few months before a difficult date in a total of 16 old participants, a new field, then don't get amazed if just 6 respond positive.

Paderborn was cancelled, yet we knew already that many participants from the year before had no interest in again joining at that date. So there was a potential of 8 participants, just 6 wanted to show up. They found that too little.

I would've still organized it with 6. Once you cancel an event, it hardly ever comes back.

So if he had looked a tad sooner and tried to email more programmers, it still might have continued. Just posting 1 email on a website is a good thing, but experience learns that's not enough. Programmers you give an email.

CSVN has same problem. In past at least they did do an email to old participants: "do you want to join?".

Last years they didn't even ship this anymore. Add up a total unreadable homepage whre you have to know how to click to read announcement of a tournament, which gets announced only 3 months ahead of schedule, and you will realize the situation. If you miss that announcement, usually programmers of course still get told by others, you see it say 2 months away. Now that is too late simply to plan things.

Computerchess is not only a technical sport, it also is an unpaid sport.

I"m organizing a blitz event now with prize money and its date is somewhere around march/april 2010. It is a 1 day event and it is 36 rounds 5 0 for chessplayers (not computers).

You see that YEAR?

2010

Most computerchess guys have no idea on organisation. You need to organize events years ahead, not a few months.

ICGA now has the luxury to play in Spain. Usually in big nations with a very active chess federation there is many chessprograms. France is closeby. So a few Spanish and French programs might of course show up now (which a year later no longer show up, we saw that in Italy also).

Referring to other organisations which have not only the same problem, but additional to that ICGA has a big problem that they are a commercial organisation charging 'organisation costs' (i heard from the different organizers amounts of 40000 euro to 100000 dollar) to those who want to host the world champs computerchess.

Not surprisingly a world title was commercial attractive in the past. In 90s i remember playing against tens of commercial programs that are no longer. When there is money to earn by doing well in world champs, which was the case more than 10 years ago, of course participants show up and invest a lot of money into hardware as they specified it.

That has changed. Participants no longer are on their knees begging for mercy in order to join, let alone pay $3000 entry fees. There is maybe 4 participants now that are willing to pay 500 euro entry fee (note ICGA was quick to switch from dollars to euro when that thing was worth more). Letting all decision taking for the event depend upon the minority of participants, just because they pay more, is very ugly.
Instead of being able to demand from participants all kind of things, ICGA somehow still manages to get money from organizers, and feels they are safe and still can do anything they like, as they already got that money and the rest everyone must be happy with that it happens.

That's just not reality anymore. One has to do more simply now to get participants. Welcome to the world of organizers i'd say.

Those organizers no longer are commercial companies, as there is simply no money to make for them hosting a world champs anymore. So it is not weird government organisations from nations with a bad financial control that organize nowadays the world champs computerchess.

This unlike 1997 when the organizers had a commercial interest themselves and made really good promotion.

So now what happens is that both the organizers, as well as the title awarding organisation ICGA, are doing total nothing to make some marketing for the event. Add to that, that ICGA nowadays first wants a signature of an organizer and payment in advance of money; that means that there is a vicious circle of sometimes malafide organisations we go to with the world champs and very little publicity.

That is no longer, so the ICGA should reorganize itself and revive itself and make active promotion for their event, starting by emailing everyone an invitation to join and giving all conditions that are there in a clear transparant manner and do effort to announce the world champs for 2010 a year ahead.

If there is publicity it usually is some of the chess teams that took a TV-team with them, together with usual the local TV channel that makes a broadcast.

For some weird reason it is difficult to get organizers of different events improve their slow manner of announcing/organizing events in advance.

Not only in computerchess that is the case, organizers who can organize *something* on time, they are really scarce on this planet.

Implementationdetails like how many days the event should last is less interesting. I tend to agree that it is easy to play 2 games a day of 5 hours a game. That's just 10 hours playing time.

Many chess events that run within a few days not seldom play 3 games a day having in total 12 hours of playing time.

I would never favour a time control though that is real sudden death at the end. Hardware plays a big role in blitz of a few seconds a move and sudden death time controls. Especially those who join the first time in an event and operate their own software, so not some cut'n pasted time control code from existing engines, they have big problems there always.

My experience told me that 1999 was a turning point. I used Bob's quad Xeon there. Too bad it didn't arrive that bottle of wine i had shipped you Bob! It really was a good bottle (and really expensive)!

Chessbase had 4 x 500Mhz there @ 3 minutes a move time control.
That's 2Ghz * 3 = 6Ghz minutes a move on average.

Games were a lot shorter back then usually than they are now. Many games now get above a 100 moves or so.

I'd argue a time control is needed in the last time period that allows at least 6 Ghz minute a move.

As majority of amateurs won't be able to buy an expensive 7500 euro machine (2 socket hardware a lot more expensive outside USA, in fact
for that price you can get much easier 16 core shared memory hardware from AMD, though a tad lower clocked processors), we have to assume 4 cores as the common amateur machine.

4 cores * 2.4Ghz = 9.6Ghz.
Now IPC is a tad higher of those cpu's,
say it's 2x faster.

So the current second time control of 30 minutes entire game after move 60, it is already quite tricky and the 2 minutes a move for the first 60 really is a minimum IMHO, to get above 1999 standards.

Vincent

This is the ridiculous bit:

In regard to the precise wording of the rules for 2009, the ICGA would
encourage you to discuss your thoughts on the details of the rule. This is
NOT an invitation to support or oppose the idea of 8-cores for 2009, since
that decision has alreday been made. It is invitation to help refine the
rule in a pracical and fair manner.

There is no reason AT ALL that the 2009 rules need to be set in stone NOW, there is plenty of time to act depending on the discussion. If they have time to refine and define it, they have time to ditch it and revert to the status-quo.

They want to destroy the WCCC. Let 'em have at it. We've seen absolutely horrible decisions made _during_ an event (letting an operator overrule a program, offer a draw when in a dead won position, etc) So the decision was made in a vacuum, and can't be changed? Just guarantees that I will not be playing in any future WCCC events at all, even through an operator as I have done in the past.

WCCC. May it RIP.

Bob i know how tempting it is for us all to post what you posted here. That is however not a clever decision. It is a world title. Scaring away participants just means they can keep more from the money instead of paying it out to tickets/travel.

ICGA gets now exactly what they want to and we haven't met in person yet.

Vincent

I have exchanged 6-8 emails with David in the past week about this subject. He claims that because Paderborn and then Leiden was cancelled, that "hardware" was the reason and he decided to make this change. Harvey pointed out that he planned on entering the WCCC, the time was not too long, etc. When I looked, they didn't play in the last 3 WCCC events and I decided to quit looking at that point. So they take advice from non-participants, over objections of past participants?

It was a bad decision. It is a bad decision. And that is nothing new to the ICGA in recent years either... They changed the WCCC from a 3-4 day event to a 9-10 day event (+ two more days to reach exotic locations and have time to get set up). They continue to rely on manual move entry when chess server software has been available for almost 20 years now. And yet David concludes that the lack of participation is a result of unlimited hardware when for the first 30 years or so of the WCCC it was not a problem at all, and, in fact, was a major attraction for the events...