OpenChess

Here are my (current) opinions about R3/IPPOLIT. It is abundantly clear that IPPOLIT's creator(s) reverse-engineered R3. Whether or not IPPOLIT "started out life" as R3 is hard to say, though my guess would be that the R3 principal functioning (w/o contempt, SMP, and other "extras") was replicated on top of their own bitboard implementation [which might have been taken from Crafty?!], and then additions/subtractions were progressively made to it. One annoyance for comparing R3/IPPOLIT is that the former has many things (such as entire search functions, move generators, and evaluation features) that the IPPOLIT designer(s) chose to ignore/exclude.

I would say that the evaluation of IPPOLIT infringes the copyright of R3, both with features and numerology [either of these could be debated] -- I'd have to analyse the situation more to determine whether the "degree of infringement" exceeds that of my opinion of Fruit/R1. I do not think IPPOLIT would pass the ICGA "originality" test. Comparing searches seems much more difficult to me, and I haven't really convinced myself of a good methodology for that, particularly to separate "ideas" from "specifics" (most notably for pruning, but also other aspects).

Statements like "IPPOLIT is essentially R3, with some additions and (a lot of) subtractions" are perfectly reasonable to me, as long as they are not taken too strongly [one needs to make a list of R3/IPPOLIT similarities/differences, and quantify the value of each component therein -- and then prepare for the inevitable dismissal of one's work ]. Similarly, "IPPOLIT differs in almost every facet from R3" is again reasonable, but be warned that some facets differ a little, and some differ a lot, some facets are important, and some are not.

I've never been happy with trying to compare the R1/Fruit "overall similarity" to that of R3/IPPOLIT, in part because the situations are so different. But perhaps I should revisit this at some future point. ["I'll address that more at a later time."]

I haven't looked much at R4, but I agree that the eval is more lightweight than R3.

by your standerd a world chess competition may have crafty , critter,maybe stpkfish and what other "legal " engine in it

Ford , Honda .BMW

by your standerd a world chess competition may have crafty , critter,maybe stpkfish and what other "legal " engine in it

I'm not sure why you think I'd find the listed engines to be OK, without more investigation.

BB+ wrote:

by your standerd a world chess competition may have crafty , critter,maybe stpkfish and what other "legal " engine in it

I'm not sure why you think I'd find the listed engines to be OK, without more investigation.

perhaps thats why i put a question mark , or was being ironic , sarcastic even

but amazed you didnt say ....but wait , im sure (at moment) honda and BMW are by my standard clones of model T Ford , and perhaps i had best compare that with stevensons rocket


still by your standard now at least cloning of a clone but not making said clone commercial , ( instead giving to world freely its code ) is a whole lot better than that scumbag thief profiting financially while those he stole from went without

abd they did fix the bishop bug for all as well


world eagerly await your investigations and posts on crafty , critter and stokfish now it hates jobs half done

BB+ wrote:Here are my (current) opinions about R3/IPPOLIT. It is abundantly clear that IPPOLIT's creator(s) reverse-engineered R3. Whether or not IPPOLIT "started out life" as R3 is hard to say, though my guess would be that the R3 principal functioning (w/o contempt, SMP, and other "extras") was replicated on top of their own bitboard implementation [which might have been taken from Crafty?!], and then additions/subtractions were progressively made to it. One annoyance for comparing R3/IPPOLIT is that the former has many things (such as entire search functions, move generators, and evaluation features) that the IPPOLIT designer(s) chose to ignore/exclude.

I would say that the evaluation of IPPOLIT infringes the copyright of R3, both with features and numerology [either of these could be debated] -- I'd have to analyse the situation more to determine whether the "degree of infringement" exceeds that of my opinion of Fruit/R1. I do not think IPPOLIT would pass the ICGA "originality" test. Comparing searches seems much more difficult to me, and I haven't really convinced myself of a good methodology for that, particularly to separate "ideas" from "specifics" (most notably for pruning, but also other aspects).

Statements like "IPPOLIT is essentially R3, with some additions and (a lot of) subtractions" are perfectly reasonable to me, as long as they are not taken too strongly [one needs to make a list of R3/IPPOLIT similarities/differences, and quantify the value of each component therein -- and then prepare for the inevitable dismissal of one's work ]. Similarly, "IPPOLIT differs in almost every facet from R3" is again reasonable, but be warned that some facets differ a little, and some differ a lot, some facets are important, and some are not.

I've never been happy with trying to compare the R1/Fruit "overall similarity" to that of R3/IPPOLIT, in part because the situations are so different. But perhaps I should revisit this at some future point. ["I'll address that more at a later time."]

I haven't looked much at R4, but I agree that the eval is more lightweight than R3.

I dare to oppose on this. I was quiet about R3/IPPO issue because I am quite familiar with R4 internals (and R4/ippo indeed have too much in common) and so far I never bothered to dig into R3. Right now I am analysing it just because I found an eval term in R4 that I thought was exclusively my own idea. Since R3 predates my implementation I am curious whether it is present there too. It is too soon to draw any conclusions but to my surprise R4/IPPO similarity is far greater than R3/IPPO. R4 was released quite a bit after IPPO, so one my draw (very wild) conclusions.

some side notes:
1) I fear that I had mistreated Ippolit for a long time without looking into R3 myself. My 'prejudice' was mostly based on your R3/IPPO comparison paper. Now I am almost convinced that Ippo was indeed written from scratch - and is completely legal. I still think that most of ideas were gained through RE, but this is not illegal in any way as long as no 'dumb copying' took place. From engineering point of view Ippo is far superior to R3. (as opposed to R4) R3 is very bloated with stuff that does not seem to add ELO and takes time to execute = unnecessary slowdown. I think after Ippo release Vasik realized that too and cut out a lot of stuff in R4.

2) I have asked to join ICGA investigation panel just to make sure evidence is objective. Indeed it was objective and you made a very good job. However, my only concern was about the evidence itself and I am indifferent about the verdict. You have looked at R3 internals and you are one of the very few persons in this world who could say that it is (almost) clean from fruit origins, yet you remained silent... strange...

Richard

veritas wrote: world eagerly await your investigations and posts on crafty , critter and stokfish now it hates jobs half done

Maybe you are just sarcastic, but I will respond anyways:

Crafty and SF are open source - anyone can compare them against each other, or against fruit, or whatever. No problem here.

As for Critter - there is no need to do any reverse engineering (unless it is a hobby) - for people like Mark it is enough to ask me for the sources and I will send them.

Richard

Richard Vida wrote:

veritas wrote: world eagerly await your investigations and posts on crafty , critter and stokfish now it hates jobs half done

Maybe you are just sarcastic, but I will respond anyways:

Crafty and SF are open source - anyone can compare them against each other, or against fruit, or whatever. No problem here.

As for Critter - there is no need to do any reverse engineering (unless it is a hobby) - for people like Mark it is enough to ask me for the sources and I will send them.

Richard

well i am surprised you only quoted one sentence that without rest is completely out of context

yes i was being "mildly" sarcastic towards Mark , mildly as i have had ," have had" much respect for him prior to his attempt to crucify ippolit

as for the three engines i metioned ( NOT Ford Honda and BMW ) its pretty obvious to anyone with half a brain that critter crafty and stokfish are " clean "

im disgusted that Ahmed / tester poseing in one of his fake accounts" Peter " on his wc of a forum insinuated critter was "dirty "
but only disgusted not at all surprised as hes a pathetic attention seeker

Richard Vida wrote:It is too soon to draw any conclusions but to my surprise R4/IPPO similarity is far greater than R3/IPPO.

Very interesting, thanks for giving your input. Would you care to give any details? I don't have any of the non-free Rybkas.

zwegner wrote:

Richard Vida wrote:It is too soon to draw any conclusions but to my surprise R4/IPPO similarity is far greater than R3/IPPO.

Very interesting, thanks for giving your input. Would you care to give any details? I don't have any of the non-free Rybkas.

Well knowing Vas hed have certainly investigated Ippolits / Ivanhoes source codes and taken anything he saw as useful , after his r3 was so easily surpassed by them , shame he didnt see fit to take the bug fix's as well

still , he has shown incompetence in his plagiarizing of crafty so....

2) I have asked to join ICGA investigation panel just to make sure evidence is objective. Indeed it was objective and you made a very good job. However, my only concern was about the evidence itself and I am indifferent about the verdict. You have looked at R3 internals and you are one of the very few persons in this world who could say that it is (almost) clean from fruit origins, yet you remained silent... strange...

I mentioned briefly in B.2 of RYBKA_FRUIT that R3 had a re-written evaluation function by Larry Kaufman. Beyond that, I don't think anyone really brought up R3, except for Bob saying that R3 might have Crafty bitboard code. There was an occasional comment as to whether R3 should be investigated, and the general consensus was no, as Rybka 2.3.2a sufficed for showing that a world championship was won with tainted code [and most Panel members probably didn't want another reading assignment ], at which point the ball would be in Rajlich's court. The fact that R3 was/is not freely available also played a minor role.

If I had thought that the majority of the Panel was mentally assuming that R3 was also overly Fruit-like, I might have said otherwise [though I'm not sure I would have been willing to invest a lot of time to write a convincing document about it] -- but given the (well-known) LK/evaluation situation and (say) Don Dailey's opinion in the Report [also Gerd's, though I think he got the date of Rybka 3 wrong], I didn't think this was the case. Also, I can't say that I agree with (say) Mark Uniacke's opinion as quoted in the Report regarding later Rybkas, but then I'll let him speak for himself.

1) I fear that I had mistreated Ippolit for a long time without looking into R3 myself. My 'prejudice' was mostly based on your R3/IPPO comparison paper. Now I am almost convinced that Ippo was indeed written from scratch - and is completely legal. I still think that most of ideas were gained through RE, but this is not illegal in any way as long as no 'dumb copying' took place.

I guess my involvement in the Rybka cases has led me to conclude that "dumb copying" is merely one way [and indeed, the most obvious] that copyright can be infringed. In another thread, I mentioned that one reason GCP favoured legal action by FSF/Fabien was precisely that many programmers have a "code is everything" sense of copyright [and a lawsuit would be a good way to publicise the contrary into the collective consciousness], while my opinion is now that there is much "creative expression" in the internals of a chess program beyond just the code. Chris Whittington is disputing some aspects of this.

For the specific R3/IPPOLIT comparison, I won't be able to get my R3 notes until I get back to Sydney in a week, but my recollection is that there were "too many" specific evaluation elements that would go beyond what I would expect from an "idea", at least in my book. The fact (as you've noted) that IPPOLIT's eval is almost a subset of R3 with many terms thrown out also makes a comparison difficult.

EDIT: When I spoke with GCP yesterday, he seemed to opine that IPPOLIT/R3 was "clearly" from reverse-engineering, that there was likely a copyright question to be considered [he can say more about his opinion there -- I don't want to misrepresent his position], and that there seemed to be great care that no "code" strictu sensu was copied.