Author: Robert Hyatt
Date: 06:12:31 08/20/05
Go up one level in this thread
On August 20, 2005 at 06:03:03, Arturo Ochoa wrote: >On August 19, 2005 at 21:29:27, Robert Hyatt wrote: > >>On August 19, 2005 at 20:49:56, Arturo Ochoa wrote: >> >>>On August 19, 2005 at 20:36:45, Uri Blass wrote: >>> >>>>On August 19, 2005 at 19:50:08, Richard Pijl wrote: >>>> >>>>>On August 19, 2005 at 15:43:01, Thomas Lagershausen wrote: >>>>> >>>>>>On August 19, 2005 at 15:26:55, A. Cozzie wrote: >>>>>> >>>>>>>Here I screwed my book creator, because the horrible, horrible Fritz interface >>>>>>>resets the book options every time you touch the mouse, and played with >>>>>>>incorrect options. But somehow Zappa managed to slip into the win anyway; >>>>>>>perhaps it was a bit lucky. >>>>>>> >>>>>>>anthony >>>>>> >>>>>>Never play with the cb-gui. It had cost a lot of programs points in the history >>>>>>of wccc. >>>>>> >>>>>>I hope Zappa 2.0 is still uci and can be used in Arena. >>>>>> >>>>>>TL >>>>> >>>>>Zappa's book requires the CB-GUI >>>>>Richard. >>>> >>>>I think that it is unfair to use books that require the CB-GUI >>>>If I understand correctly it means that the GUI choose the book moves and not >>>>the engine so the author is using something external program to help him to >>>>select moves. >>>> >>>>I think that the engine needs to choose all the moves(otherwise the playing >>>>thing is not original work of the author and the authors of the chessbase gui >>>>should be mentioned as part of the team). >>>> >>>>Uri >>> >>>Uri Blass, the king of the absurd arguments. >>> >>>1) For CCT7 and the Elhvest Match, I used the native format for Zappa. >>> >>>2) For WCCC2005, there are several engines that has been using the ChessBase >>>GUI. However, it doesn't mean that the Book was made by ChessBase. The Book >>>Responsible for Zappa in the WCCC2005 has been Erdogan whose book is in the >>>ChessBase format. It is his original work performed by several years. >>> >>>The engine is a complete original work of Anthony Cozzie. >>> >>>If you are going to begin your post-WCCC2005 nonsense before the Tournament is >>>over, I suggest you find other hobby according to your absurd world, instead of >>>writing craps every day of the year. >> >> >> >>This argument comes up every year, and Uri has a valid point. > >To begin such argument where the Tournament is quite over is just absurd. >Erdogan's Book is in CB Format and he had only a few weeks to test Zappa. > I don't see the point. A book can be converted trivially. To PGN first, for example... But the point of the discussion is that the current approach is flawed, in a very basic way... > >> >>Although the ICGA doesn't seem to quite grasp the problem that many of us have >>pointed out. The opening book is a _significant_ part of a chess engine. Which >>means the code to select moves from that book based on some sort of algorithm is >>going to possibly play a significant number of moves, if not the majority of the >>moves in the game. Allowing someone else to write this code and then share it >>among multiple engines is simply wrong. Writing custom books is fine, but the >>chess engine author should be responsible for any code that makes chess playing >>decisions. > >While the ICGA doesnt care about format rule to forbid this, the discussion is >just a waste of time. Nobody can say that I have ever used a CB Book for a >specific engine. Erdogan has developed the book in several years and I had to >retire from the Zappa's behalf for personal reasons. Jeroen has done a book used by more than one engine in the same tournament. And it has been discussed both at previous player's meetings and here. And the ICGA seems to take the path of "let's not piss off the commercial companies as they contribute money with their 'professional' entry fees." > > >> >>For example, we could have the following issues: >> >>1. If the GUI chooses book moves, and handles book learning, is it reasonable >>that the _same_ code be used in multiple engines? I tend to say "no". > >Invalid: Not time for learning, since Erdogan only had 3 weeks to tune his Book >for Zappa. He has already an incredible work. There can be learning _during_ the tournament to not repeat an opening that was not as good as planned. The engine should be responsible. > > >> >>2. If the GUI does the time allocation, tells the engine how long to search, >>when to search longer, when to search faster, handles multiple time controls, >>and so forth, isn't that a major function that a chess engine has to manage? If >>so, is it fair that multiple programs share this code since they share a common >>GUI? Again, I would say "no". >> >>3. If the GUI handles endgame tables, should the GUI be able to either >>instantly play a table move, or say "let's search, this is a draw, and we want >>to give the opponent a chance to make an error." I coded "swindle mode" into my >>program, along with code to handle missing tables (you have kpk but not kqk so >>your program might never promote without a fix.) Is it reasonable for a single >>author to write code to do all of that, and then have multiple programs share it >>in a tournament? Again, "no". > >Ir doesn't apply for Zappa because it access the EGTBs via Zappa and not by CB. I am talking about GUIs in general, if you are reading carefully. Not what happens for one specific program. But unless someone has turned off the GUI endgame database probes, the GUI will most definitely play those moves... > >> >>4. Should a book author be able to write a book for multiple engines? Can he >>physically separate the two projects so there is _nothing_ in common? Of course >>he can't, and this is simply a bad idea. This has come up multiple times, and >>yet it never gets addressed properly, because it might "offend" a commercial >>company that sells programs that could share the book. > >It doesnt apply either for this case, since Zappa is using the help of Erdogan >and I was contacted by Diepeveen. I'm talking about the "general problem". Bruce Moreland raised the issue at the Paris WMCCC event, where there were cases of two programs using the same book author, which ought to be viewed as a no-no... > > >> >>I have no problem with a GUI doing "GUI tasks". Recognizing moves, displaying >>the board, displaying the clock, etc. But the GUI has no place going farther >>and actually influencing which moves are played in a game. Because that crosses >>over into what the engine is supposed to be doing. GUI means "Graphical User >>Interface". Not "Graphical User Interface and front-end to make basic chess >>move decisions before letting the engine do anything." > >Zappa required a last hour booker and Erdogan was a great solution. On fact, >Zappa is winning the Tournament for now. And he deserves to win. But that has nothing to do with whether someone should be allowed to use someone else's code to play their book moves. That's the point here... > > >>In the Zappa case, I don't have any problem with it using a "book" written by >>anybody (so long as it is a one-user book and not shared.) But the current GUI >>is taking over too much of the game's complexity... Winboard/Xboard is an >>example of a reasonable "GUI". Just does GUI tasks. No book, no timing >>decisions, no endgame table probes, no nothing but relaying information between >>the user and the engine, exactly what a "user interface" should be doing... > >I don't really care if you have problems or not. Winboard is not a reasonable >GUI as far as it can't manage a simple book structure. Yes, I admire how Peter >Beger can stand the work with simple txt files, but he has had more thanb 3 >weeks to prepare the Crafty's Book. Aha. That is _THE_ point. The GUI should _NOT_ manage any kind of book structure at all. That is part of playing the game, and the engine should be doing the playing, not someone else's GUI...
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.