Loop 2007 / Fruit 2.1

Code, algorithms, languages, construction...
BB+
Posts: 1484
Joined: Thu Jun 10, 2010 4:26 am

Loop 2007 / Fruit 2.1

Post by BB+ » Mon Apr 18, 2011 8:15 am

I made a cursory glance at Loop 2007 (the 64-bit version). It uses exactly the same PST as Fruit 2.1. The pawn eval uses the same scores. It shifts the mobility count as with Fruit (by 4,6,7,13) before multiplying, and then the arithmetic is the same (4 for N, 5 for B). I'm not that interested in going further.

Prima
Posts: 328
Joined: Tue Dec 14, 2010 6:12 am

Re: Loop 2007 / Fruit 2.1

Post by Prima » Mon Apr 18, 2011 11:48 am

Thanks BB+ for this enlightenment.

This is all the more reason the proposal by Bob Hyatt or the "chess regulating body" requiring all closed-sourced chess engine authors partaking in tournaments should submit a copy of their source code, makes sense.

It's unfortunate that some honest closed-source authors may be affected by this. But considering how some greedy & selfish people stand to benefit "commercially" from works that was provided freely & openly or through GPL, closed-source verification is a must.

When these copycats can no longer find codes to copy, from strong open-source programs that may be too apparent to support their commercial chess engines, they opt out of the commercial market with people's money in their pockets and call it quits. How convenient.

Ancalagon
Posts: 15
Joined: Sat Sep 11, 2010 1:24 pm

Re: Loop 2007 / Fruit 2.1

Post by Ancalagon » Mon Apr 18, 2011 12:39 pm

BB+ wrote:I made a cursory glance at Loop 2007 (the 64-bit version). It uses exactly the same PST as Fruit 2.1. The pawn eval uses the same scores. It shifts the mobility count as with Fruit (by 4,6,7,13) before multiplying, and then the arithmetic is the same (4 for N, 5 for B). I'm not that interested in going further.
Well, I am sure this is not the place to discuss this BB+, but in case Fabien Letouzy is considering legal action, it could be that the author of Loop earned more money from his program than any other programmer that may have copied code from Fruit. I believe Loop was converted to a chessprogram for the Nintendo somewhere in that timeframe (2007). It could have been another gameconsole but that is what I remember of Loop. So Fabien might be interested in looking at this further. In case he does not read this messageboard, I think it would be better if you told him of your findings? Otherwise this might drag on another six years or so.

Eelco

ThinkingALot
Posts: 144
Joined: Sun Jun 13, 2010 7:32 am
Contact:

Re: Loop 2007 / Fruit 2.1

Post by ThinkingALot » Mon Apr 18, 2011 2:23 pm

BB+ wrote:I made a cursory glance at Loop 2007 (the 64-bit version). It uses exactly the same PST as Fruit 2.1. The pawn eval uses the same scores. It shifts the mobility count as with Fruit (by 4,6,7,13) before multiplying, and then the arithmetic is the same (4 for N, 5 for B). I'm not that interested in going further.
That's not surprising. I recall Loop's ponderhit against some Fruit/Toga versions to be well over 70% :!: .
Prima wrote:requiring all closed-sourced chess engine authors partaking in tournaments should submit a copy of their source code, makes sense.
IMHO, that's not a solution. The code of a top 1 engine may cost a hundred thousand so sharing it is quite unsafe 8-).
I think all those who release their sources should just get accustomed to the idea of their code being used in commercical programs. Don't agree - nobody forces you to release your code. There's absolutely no way to prevent its further unlicensed use.

Vempele
Posts: 4
Joined: Sun Apr 03, 2011 6:54 am

Re: Loop 2007 / Fruit 2.1

Post by Vempele » Mon Apr 18, 2011 3:42 pm

BB+ wrote:I made a cursory glance at Loop 2007 (the 64-bit version). It uses exactly the same PST as Fruit 2.1. The pawn eval uses the same scores. It shifts the mobility count as with Fruit (by 4,6,7,13) before multiplying, and then the arithmetic is the same (4 for N, 5 for B). I'm not that interested in going further.
You might want to ask Tord Romstad. http://www.open-aurec.com/wbforum/viewt ... 875#p29875

(I recall Tord later mentioning something about some unspecified closed source engine being very similar to Fruit/Toga but I can't find that post)

BB+
Posts: 1484
Joined: Thu Jun 10, 2010 4:26 am

Re: Loop 2007 / Fruit 2.1

Post by BB+ » Tue Apr 19, 2011 5:25 am

I believe Loop was converted to a chessprogram for the Nintendo somewhere in that timeframe (2007)
http://en.wikipedia.org/wiki/Wii_Chess

It seems that the Loop 2007 engine was the one that finished 3rd in Amsterdam, as the Loop Chess page lists the date June 18, 2007 (the ending day of the WCCC that year).

Tord
Posts: 35
Joined: Mon Jun 14, 2010 9:08 pm
Real Name: Tord Romstad

Re: Loop 2007 / Fruit 2.1

Post by Tord » Wed Apr 20, 2011 10:33 am

Vempele wrote:You might want to ask Tord Romstad. http://www.open-aurec.com/wbforum/viewt ... 875#p29875
I've seen some snippets of Loop's code, but never the entire program. There is nothing in those snippets I have seen that is interesting for this discussion.
(I recall Tord later mentioning something about some unspecified closed source engine being very similar to Fruit/Toga but I can't find that post)
Your recollection is correct, but I was not talking about Loop. When I looked at the engine in question again recently, I noticed that the similarities to Fruit were even more striking than I remembered. This puts me in a somewhat uncomfortable position. On the one hand, it seems clear that the community should be told about this case of plagiarism. On the other hand, the source code was sent to me confidentially by the author, and going public would be breaking his trust. In the end, I have chosen to keep the name of the engine private (I've mentioned it to Fabien, though), because it is no longer in development (since a long time), was never commercial or very close to the top, and never won any major tournaments.

I probably will go public if the author starts working on the engine again, and/or enters a public tournament. But I have very good reasons to believe that the project is completely dead, and that we'll never see another version.

BB+
Posts: 1484
Joined: Thu Jun 10, 2010 4:26 am

Re: Loop 2007 / Fruit 2.1

Post by BB+ » Wed Apr 27, 2011 4:34 pm

With the ICGA Investigation for Rybka winding down, it looks like Loop 2007 will be next on the agendum. I attach a preliminary version of an annotation of its evaluation function. The differences to Fruit 2.1 are few, and some might be in a Toga version (such as the Tempo bonus). Unlike some other derivatives, it seems that Reul did not even bother to change the numerology.
Attachments
LOOP.eval.txt
ASM dump of Loop 2007 eval function, annotated, 1315 lines
(54.71 KiB) Downloaded 676 times

User avatar
marcelk
Posts: 52
Joined: Fri Jan 28, 2011 10:27 pm
Real Name: Marcel van Kervinck

Re: Loop 2007 / Fruit 2.1

Post by marcelk » Sat May 07, 2011 3:29 pm

That is indeed the Fruit evaluation. After going through all the Rybka investigations is easy to recognize.

Isn't "Loop" the follow-up program of List, which was accused to be a derivative in the 2003 WCCC? At some
point in time was investigated by the ICGA and declared to be legit: http://ticc.uvt.nl/icga/journal/pdf/toc28-1.pdf

That means one of these:
1a. the investigation overlooked the copying,
1b. the code that entered the tournaments was not the code that was examined; or
2. the copying happened after the investigation.

Each of these scenarios is "intriguing" at least..

BB+
Posts: 1484
Joined: Thu Jun 10, 2010 4:26 am

Re: Loop 2007 / Fruit 2.1

Post by BB+ » Mon Aug 22, 2011 9:01 am

Back in 2008, there was a commentary on the LIST case, as:
Chris Whittington wrote:Reading the article it sounds like kangaroo court.

Someone complains. The bigwigs demand to see the source. The programmer says no, my source is mine and secret. They ban him.

Obviously what went wrong was the pre-decision. To ask for source they need strong evidence already. Otherwise anybody can, in Stasi manner, wreck any competitors tournament for them.

Stinks.
I happen (now) to know more details, and there was indeed rather strong evidence that LIST, at least in some versions, was derivative of Crafty. This was presented to the ICGA with said complaint.

Again one can relate this to the "delay" with the Rybka situation, as the ICGA was unwilling to move in that case w/o more evidence, or at the very least a specific accusation. As CW put it in a different post from that era: Source is highly sensitive. In order to be able to insist on seeing source code a very strong case must be made first, with substantial evidence and the opportunity for refutation by the accused programmer.

In any event, I think the ICGA process regarding Loop (the investigative part of which was requested by Fabien) is currently stalled, as no one has been able to contact Fritz Reul.

Post Reply