Author: Peter Fendrich
Date: 04:50:31 11/28/03
Go up one level in this thread
On November 28, 2003 at 06:12:43, Odd Gunnar Malin wrote: >On November 28, 2003 at 05:00:49, martin fierz wrote: > >>just a short question: >> >>if i rip assembler stuff like popcount and firstone from the crafty source, but >>the rest of my program is entirely different, am i doing something wrong? >> >>if i use the kogge-stone floodfill algorithms posted here by steffan westcott, >>am i doing something wrong? >> > >As soon as you start copying code from another source I think it is cheating if >you want to participate in a competition with your program. > >>how much foreign code is allowed? > >None. That is impossible to prove and not a practical standpoint in this issue. How can you ever know if the statement: for (i=0; i<64; i++) is a copy or not? In some sence it is a copy from a lecture where I just changed from 10 to 64. /Peter >> >>cheers >> martin >> >>PS: i use a table-based popcount, not crafty's assembly code. i use a modified >>version of crafty's lastone. i don't use steffan's floodfill code, but i plan to >>give it a try. > >This is not cheating because here you have learned an idea you want to test it >yourself. > >There is a difference in what you want to do with your program. >- If it is a private app. you can do what you want (no one can check). >- If you want your app. to be public you have to get permision from the author. >- If you want to participate in a public tournament you have to write it all >yourself or add the orgin to your team (with permission). > >Some borderlines: >- Fritz's (or others) booklearning code. In a tournament this isn't of any help >but as soon as the competition is match play then the author of the code is also >part of the team and the tournament rule decide if more than one engine can use >it. ( I have to add that CB probably learned some of the learningcode from Hyatt >(I think it was mention here a year back), but this is an article in a paper >(ICCA) and not copying of codes.) >- Nalimov's egtb code. Here too is he part of the team so if the tourney only >allow an author to participate with one entry, only one entry should use his >code. >- Opening book, same as for the egtb code. > >The extreme case that someone seems to not think of as a clone is to start with >tscp or gerbil and modify it. >I think even when you have modified all the code you should still not be allowed >in a tournament without the orgin's permission and added to the team. >Basicly here you add a startrating of 1600-2000 for your engine in contrast with >one who start from scratch where the engine would be below 1000 the first tries. > >BUT it is allowed to look into other code when you are learning, or hunting for >ideas. This is probably the cause that the code is open. >If you draw similarities with normal chess, you study theory books (papers) >before a tournament and in some variants you want to look into other sourcecode >(games) to see how this works in a real game. What you aren't allowed to is to >bring with you these books and games when you write your own code (play your >game). > >I would have wanted that authors that release the sourcecode had spend a litle >time to write some pseudo code instead to show the unique idea and publish this >instead of the code. This would have forced people to learn the code and test it >througout before they can use it. > >Bruce Morland's site is an excellent sample of this. If you still don't >understand his ideas after reading it several times you can take a peep at a >sourcecode where the ideas are implemented. > >Another sample could be the three-four line Hyatt added after the booklearning >stuff in ICCA 1, 1999 about positional learning. These few lines of words are >all you need to implement positional learning into your engine. > >Odd Gunnar
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.