Reinhard, SMIRF in Gothic Chess Championship?

[Michael Yee]

Hello Reinhard,

I noticed that the gothic computer chess tournament is coming up soon, and was curious how your preparation is coming and what you think your chances are...

But don't feel pressured to give out info if it would help the competition!

Thanks,
Michael

[Reinhard Scharnagl]

Well Michael,

I am still waiting for some licensing details from Ed Trice. I have made a lot of postings on that point but still I am not better informed than before.

So I am slowly proceeding with Smirf, hoping to be finished with it in time. But each day I get less motivated.

Reinhard.

PS.: If you would like to read here some topics from "Programming", may be you get some news about Smirf.

[Michael Yee]

Hello Reinhard,

I remember those long discussions in CCC about the (unfortunate) patent... I guess I thought the entrance fee would "cover" participants from potential legal trouble.

In any case, if you do compete, good luck! I've been following your project for a while since I'm interested in static eval from both a machine learning perspective (learning weights through self-play or from human games) and from an abstract/mathematical perspective (namely, your approach of computing value based on board geometry, etc.).

Michael

[Reinhard Scharnagl]

Hello Michael,

thank you for watching my slowly proceeding Smirf project. There is no big number of watching people. You become interesting to others if you would have success. But my approch is not necessarily ending with the creation af a top engine. Having selected a very disliked chess extension - the Fischer Random Chess - you do not make yourself a lot of friends.

But I have been very enjoyed by some wake chess engine programmers starting to enable their program also for FRC. That might be a result of the Arena GUI, which is trying to enable playing FRC with UCI or some Winboard protocols, though FRC still is not really supported by those. After having written a German language book on Fischer Random Chess / Chess960 I hoped for to increase its popularity, but only very few people care about that or even buy it.

It is my goal to create a chess program without an implementation of expert knowledge. Because of that my independent efforts to have a neutral and naive approach to chess might be better to understand. I hate huge look up "knowlegde" implemented like big opening libraries or table bases, because so the development of really intelligent processes is nearly disabled. But the main stream is favoring engines with persistant data always endless growing, whereas the use of limited ressources for me seems to be essential in the game of chess.

To avoid conflicts with the "patented" GothicChess I have invented a free own 10x8 variant called Capablanca Random Chess, CRC. One problem is that there is no GUI supporting sumultaneously 8x8 and 10x8 boards including the capablanca piece set (to support my so called FullChess idea). Also there is still no unifying protocol. Therefore I have written a GUI during the last weeks, covering both geometries and widly supporting PGN notation compatibilitly, using a still developing new DLL based protocol with only one single interface function. It is named TMCI: "third millenium chess interface" and it should be finally published for free usage but be moderated in questions of coming additions or changes.

It is not sure, that Smirf will be ready next month, but I hope it will be able to participate, even when playing at a probably very low level. May be things could change e.g. by being motivated by a license agreement according GC, but I doubt about that. The chances are that Smirf will be published first without any automated or program supported GC facility. But if people would be interested in CRC or FRC, Smirf would in no way become less interesting.

Regards, Reinhard.

[Brian Richardson]

I for one found the SMIRF approach to determining piece values based only on the number of squares that pieces can move to to be very interesting, along with the passed pawn notion of the 7th root. I use a similar approach in Tinker, but based on the fact that there are only 5 moves for a pawn to queen.

Brian

[Reinhard Scharnagl]

Hi all,

what always is astonishing me is that there are a lot more of reactions to my FRC/CRC approaches from the USA and far away from me than from Germany (my home country) itself. I would not worry much about that, if English would be my native language, but it is not. So it is futil to think over the possibility, that I may have written my FRC book in English language, it would be always to be recognized as an elaborate of a German writer.

What I really do not understand actually is that only so few people are interested in the (as far as I know) only basic book on FRC in our German language. Do you have an idea or some tricks for me to know, how I could get a better resonance in German speaking countries? You would hardly imagine how many books I have sent to may be multiplicator individuals, but with very very poor results.

Overmore I am still looking for a sponsor to support me e.g. with competitive hardware etc.. It would be easier if I would do some nonsense task like painting pictures of the White House with printed illegal chess moves, but being serious nobody takes any notice of Smirf. Well that is enough moaned. Ok then, I go on and continue working on Smirf ...

Reinhard.

[EdTrice]

Hello everyone, especially Reinhard,

A few things have been said that I would like to address.

1. Reinhard (and all other participants in the GCWC 2004) have been issued a license to create Gothic Chess sofware engines for this tournament. Like all of our intellectual property licenses, they have a one year term that is renewable. Reinhard is the only individual who keeps asking "for more", but I do not know what else he needs. He has several email confirmations of the license, he has several public posts made by me regarding this, but I cannot give him an invitation engraved in gold.

Reinhard: you have all of the necessary licensing you need to enter this tournament.

2. The tournament was originally scheduled for August, and it was moved to November so that SMIRF would be able to participate. This was not a popular decision, but I made it believing the contest would be better with more entrants rather then fewer. Other programmers were following an aggressive programming schedule to make it in time, and now many of them felt that they have been "rushed" for no reason. I apologized to them, and now it is my hope can participate.

3. People do not have to like the fact that Gothic Chess was patented. It was done so to "protect" the investment of our firm, which is making money on selling this game (boards, pieces, program, tournaments). If someone else would come along and so the same thing, thereby "stealing" all of our research, that would not be fair.

The patent allows us to charge a royalty fee if others also which to sell Gothic Chess items. This right to charge expires in 2019 or thereabouts. After that time, anybody can make and sell Gothic Chess stuff.

Just so everyone knows, all programmers only paid $1 for their license. I cannot possibly be any more generous than that.

4. Any further discussions of the patent should be directed to me at GothicChessInfo@aol.com as is now our company policy.

I wish everyone the best of luck in this contest!

If SMIRF participates, we will have 8 entrants.

It will certainly be exciting!

[Reinhard Scharnagl]

Hi Ed,

Reinhard: you have all of the necessary licensing you need to enter this tournament.

this is the first time, that I could read this such clear. Maybe (because English is not my genuine language) I had not got that message.

... and it was moved to November so that SMIRF would be able to participate

That being a reason for that also is quite new for me. But may be as it is, I have tried to keep you informed on Smirf's progresses mostly combined with questions on licensing details, which (in my view) seemed to be ignored. It should have been obvious to you that I have been very engaged in the whole thing, when deciding even to write an own GUI for Smirf covering 8x8 and 10x8 boards, which is nearly completed. But without getting an adequate resonance I have been really frustrated concerning Gothic Chess.

People do not have to like the fact that Gothic Chess was patented.

May be, but this is irrelevant for me. I do not feel good with patenting such derivations of traditional games, but its your decision. And I would have no reason to think on that at all, if some questions could have been answered instead of being repeatedly delayed (as I felt that it was). But again, I have not been able to understand, why it has been impossible for me to view any licensing form or proposal to that. But that has been the secondary problem, primarily I have been very unsure whether participating with Smirf without having ever seen a licensing contract would be a prejustice in some aspects. I am not a lawyer and not familiar neither with US rights nor with patenting material at all. And I was feeling very unsure reading about some conflicts inviting your opponent to some very expensive adventures from Europe to USA to meet your lawyer.

Any further discussions of the patent should be directed to me ...

Ok, read some of my postings again. Answering here has not been done to talk on any details, but to reflect that I obviously have another view of our communication results. I always have tried to proceed our relation even when others would have preferred to terminate their contacts.

I am still working on Smirf. But Smirf for sure now would not be that strong as it possibly could have been.

Reinhard.

[EdTrice]

Hello Reinhard,

My hope is that all is clear now. If you had sent communications to me that were not received, this is the first I am hearing about it and I offer any apology that you will accept.

There is no point in mentioning the past. I intend to live the rest of my life in the future

I will hope to hear from you soon by email if you have any more patent concerns. Everyone else is ready to go.

[Reinhard Scharnagl]

Hello Ed,

I will think it all over. It is late here in Germany and I will take some sleep now.

Regards, Reinhard.

[Reinhard Scharnagl]

Hi all,

I just want to give the information to you, that a Smirf prototype has been completed to participate in Ed Trice's championship. It seems not to be that strong as I have wished it finally would be, but in the end I succeeded in implementing a permanent brain facility. It still will need a lot of work to make Smirf a really completed program.

But nevertheless it supports 8x8 and 10x8 chess, combined with Fischer castlings FRC / CRC, can read, write and comment PGN files opened for those variants downwards compatible to traditional game notations.

I wish all participants of that tournament a fair and attractive event.

Reinhard.

[Uri Blass]

Hi all,

I just want to give the information to you, that a Smirf prototype has been completed to participate in Ed Trice's championship. It seems not to be that strong as I have wished it finally would be, but in the end I succeeded in implementing a permanent brain facility. It still will need a lot of work to make Smirf a really completed program.

But nevertheless it supports 8x8 and 10x8 chess, combined with Fischer castlings FRC / CRC, can read, write and comment PGN files opened for those variants downwards compatible to traditional game notations.

I wish all participants of that tournament a fair and attractive event.

Reinhard.

Hi Reinhard,

I think that in games you never complete your program and there is always a room for improvement.
The professional programs get better every year and we do not see signs to the end of it.
It may be interesting to compare the playing strength of smirf on 8*8 board with other chess programs.

Did you test it in games against other programs or you still had no time to do it because you concentrated on preperation for the gothic chess championship and it is better to test it in gothic chess and not in chess?

Do you plan to release a version of smirf that can play normal chess after the tournament?

I also think that releasing your tool to calculate perft can be productive
espacially if you extend it to calculate perft on pgn files(I mean calculating things like the sum of perft 4 for every position in the pgn).

Uri

[Reinhard Scharnagl]

Hi Uri,

Did you test it in games against other programs or you still had no time to do it because you concentrated on preperation for the gothic chess championship and it is better to test it in gothic chess and not in chess?

most I used special test runs, which mean not much, but simply to secure to create a legal move generating and searching engine. Sometimes I played with Smirf against Shredder, but was not that successful. I have not had the time to do it the way it should be, but the date has been fixed ...

Do you plan to release a version of smirf that can play normal chess after the tournament?

There has nothing to be done for that. It is an essential feature of Smirf that its engine (only one) can play 8x8 chess, 10x8 chess, Fischer castlings ... Because of that Smirf is already playing traditional chess or FRC, too.

extend it to calculate perft

There are so much things left to do... but indeed, why not have such a perft tool built in?

Regards, Reinhard.

[Uri Blass]

What is going on in this championship?

There is no update about the results for some days in
http://www.gothicchess.org/world_championship.html

there were supposed to be played more rounds.

Uri

[Reinhard Scharnagl]

Hi Uri,

there is room for speculations. Well there has been a two days break, and I hope the event will be continued now. But of course there must have been some difficulties, whose nature are not really clear.

First there has been a problem with Chess V, which insisted to crash in a special situation. To the programmer the chance has been given to fix that behaviour, but I do not know, whether he would produce a new release. Then there should have been troubles with the operators, which were not present at each day of the event. And I have no idea on the reasons of this strange situation.

Finally after the two days break there obviously will be a moving of the whole event to another place (or sponsor?) with significantly weaker hardware, e.g. with RAM limited to 256MB at each computer. The participiants will have to change their software or settings appropriately in time ... so wait and see! :-)

Reinhard.

[Uri Blass]

Hi Reinhard,

I see that smirf got a draw against tscp Gothic when practically both flags fell.

This seems not to be a serious tournament because it seems that things are dependent on the speed of the operator and there are too many bugs in programs thanks to the lack of interface like winboard that allow programmers to test their program.

Tscp is known as relatively weak chess program because the it is a simple program that the author did not write in order to compete(he wrote a clearly faster and better program with the name stobor).

My opinion is that the fact that a program that is based on tscp has such good result only shows that there is not much interest in programming
gothic chess.

I expect good programmers to be able to do something that is clearly better than tscpgothic in a few months.

I guess that the patent about gothic chess does not encourage the real good programmers.

In your case you may be a good programmer based on the speed of your move generator but I guess that you did the mistake of optimizing too early.

I think that a better strategy to get good result in the tournament is to work less on improving the speed and concentrate more on testing.
I read that you had hash bug(and maybe you have more bugs that you did not find) and I guess that smirf could probably score better with slower implementation that has no bugs and with testing that things that you believe to be better are really better.

Uri

[Reinhard Scharnagl]

Hi Uri,

yes Smirf gained 1/2 point! But I am not sure whether this is correct or not.

TSCP Gothic is not that weak as it may seem. But you are right when stating, that the development of GC programs is lacking because of missing a Winboard like GUI.

One point, why Smirf still is weak playing, is that I have spent a lot of time in developing an own GUI and protocol. May be other programmers are interested in that protocol. But as long as I am the only one, it would not make sense to open the SMIRF GUI for more engines. Overmore I would rather target 10x8 CRC instead of the "patented" GC.

Still I think, there is a big error somewhere around refetching cached values in my engine, but I have not detected it yet. I hope I can fix it soon. The "completing" of my engine has been done in a hurry, and the result thus is far away from there where Smirf's goal might be. But it has had priority to finish it into a real playing program.

So we will wait and see ...

Reinhard.

[Dan Honeycutt]

yes Smirf gained 1/2 point! But I am not sure whether this is correct or not.

Hi Reinhard:
Is this a real tournament? Looks more like a cluster .... I best not finish that.

I've been trying to follow smirf's results without much luck. It's early am Saturday here in Atlanta which is in the same time zone as King of Prussia, Pennsylvania. I just checked the tournament site looking for the rest of Thursdays results and maybe some Friday results. Zip. Nada. The babe in the T-shirt is nice but I was hoping for some chess.

As a participant maybe you have access to information and can give us an update. 1.5/5 (if that's really the score) is not a stellar result but I know you were pressed for time. I'm not really much of a fan of Capa/FRC but I do admire your efforts and I wish smirf the best.

Dan H.

[Uri Blass]

Hi Dan,
Smirf has 1.5/6 based on the results.

The table is simply wrong and the people who constructed it forgot one loss of smirf.

Uri

[Dan Honeycutt]

Hi Uri

You are correct. Looks like they fixed the table but still no new results.

Dan H.