Author: Robert Hyatt
Date: 08:39:19 12/03/03
Go up one level in this thread
On December 02, 2003 at 11:29:21, Vincent Diepeveen wrote: >On December 01, 2003 at 13:30:53, Robert Hyatt wrote: > >>On December 01, 2003 at 13:01:21, Dann Corbit wrote: >> >>>I sent a letter to David Levy >> >> >>I doubt it to, based on output. However I have not been asked >>in any official query, as was done in the past with other programs. > >???????????????????????????????????????????????????????????????????????? What part of "official query" do you not understand? "official" -> "someone from the ICGA or the tournament committee". Care to make _any_ guess how many such "clone" questions I get in a single year? The evidence in the email was hardly conclusive. I was not given an executable to examine, which would have been _much_ more revealing. I'm not willing to accuse someone of copying my code until I can do so beyond a reasonable doubt. IE with bionic and voyager and le petite, there was _zero_ doubt. Same function names. Same strings in the program. Same binary bit patterns that I use for evaluation, move generation, etc. Same binary files produced on execution (book.bin, position.bin, etc) even though the author did take the time to change the names. But the contents were _identical_. There were other things I looked at that I won't go into here. In this case, someone chose the same names. So, suppose they _started_ with Crafty's source, and rewrote it piece by piece? Nothing wrong with that at all, and it might cause them to use similar names. But the ICGA _never_ asked me anything at all, and, as a result, I never mentioned this. I did respond to the person that sent me the email, but that email I won't post here. > >I was forwarded by the person who raised the protest (no that wasn't me) to the >ICGA a month or 2 ago: I received that email. I probably still have it somewhere. I am _not_ the ICGA. I am barely a member and that is going to end soon. > >----- Original Message ----- >From: <censored> >To: "Robert M. Hyatt" <hyatt@cis.uab.edu> >Sent: Monday, March 31, 2003 12:05 AM >Subject: Potential Crafty clone > >> Hello Robert, >> >> I have reason to believe a rather well-known program >> is actually a Crafty clone. It's rather new, the first >> versions already being extremely strong, the lastest being >> one of the top free engines. >> >> My suspicion started after observing that it had some holes >> in the evaluation which are also present in Crafty. It has >> a slightly higher speed than Crafty, uses the exact same >> extensions. The major difference seems to be that it uses >> (as far as I can determine) a crude form of tactical forward >> pruning, or at the least it searches a bit deeper but is >> positionally weaker. >> >> I have been able to determine of what source files the >> program consists (it's only available as an .exe): >> >> attacks.c >> board.c >> captures.c >> evasions.c >> hash.c >> init.c >> interface.c >> io.c >> programname.c >> make.c >> moves.c >> next.c <----- >> pawns.c >> phase.c <----- >> root.c <----- >> score.c >> search.c >> speed.c >> timer.c >> uint64.c >> undo.c >> utility.c <----- >> >> Most of these are rather generic so it's not suspicious >> that they are also in Crafty, with the exception of those >> indicated, as far as I know they are more or less unusual AND >> also present in crafty. There are some other things, like the >> seperation of make/unmake (here called undo), which as far as >> I know is also unusual. >> >> A previous version of the program consisted of: >> >> attacks.c <====== >> bench.c <------ >> captures.c >> chessboard.c >> hash.c <====== >> init.c <====== >> io.c >> programname.c >> make.c <====== >> moves.c >> next.c <====== >> pawnhash.c >> phase.c <====== >> ponder.c <------ >> recog.c >> score.c >> search.c <====== >> searchr.c <------ !!! >> sort.c >> test.c <------ >> timer.c <====== (time.c) >> undo.c <====== (unmake.c) >> utility.c <====== >> >> There are a lot of additional hits with Crafty now, the most >> alarming being searchr.c. I indicated all sources which are >> also present in Crafty, there are a lot of them. >> >> Even more alarming is this: >> >> material_score >> pawn_structure_score >> passed_pawn_score >> king_safety_score >> interactive_score >> total_score >> >> His evaluation debugger has the exact same structure as the one >> in Crafty. Especially nice is the 'interactive score', I've never >> seen this anywhere but in Crafty either, and it's certainly something >> that is unlikely to be arrived at independently. Combining this >> with the rest of the hits makes me very very very suspicious of this >> program. >> >> What is your opinion? Am I paranoid or are we onto something here? >> Is there any way to track down more information? > >> --
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.