Author: Dann Corbit
Date: 12:07:01 03/17/05
Go up one level in this thread
On March 17, 2005 at 14:46:53, Thomas Gaksch wrote: >On March 17, 2005 at 14:14:45, Dann Corbit wrote: > >>On March 17, 2005 at 14:05:18, Thomas Gaksch wrote: >> >>>On March 17, 2005 at 13:13:23, Dann Corbit wrote: >>> >>>>On March 17, 2005 at 03:22:06, Gabor Szots wrote: >>>> >>>>>Do you agree that after giving you due credit Toga's "author" can claim that >>>>>"his" engine is his own intellectual product and can be called a new engine? >>>> >>>>Let me say it plainly: >>>> >>>>Toga II is fruit. >>>> >>>>There are a few tiny tweaks. >>>> >>>>If there was ever a model for the definition of a clone, it should say: >>>>"See Toga II" >>> >>>you are absolutely right. i have never said anything else. i am very frustrated >>>that everybody is criticising me. i made a big mistake. thats true. but i can´t >>>turn back the time. i only could apologize me for my exorbitancy (i hope it is >>>the right word). >>> >>>what is bad on it, that the little changes seems to improve the playing >>>strength. why shouldn´t it be tested. are not all open sources projects clones? >>>i dont know. >>> >>>thomas >>>a frustrated "co-author" or user of open source software >> >>You made some very important and interesting changes to an existing chess >>engine. >> >>The criticism you received is well deserved, because you misrepresented to truth >>to get some free testing. >> >>On the other hand, you did come clean and tell the truth when put on the spot, >>and so I think you deserve credit for that. >> >>This is all behind us now. I think that fact that you were honest about it in >>the end and your apology is enough. >> >>How about turning it into something very positive. Why not talk about your >>ideas for the changes and tell us why you decided to try them? > >hi dann, >of course i can do that. but it is absolutely nothing new what i have >implemented. >the first thing was the king saftey. i noticed that fruit likes to advance the >king pawns after casteling. so ein tried to implement an penalty score which >depends on the value of the enemy pieces. >then i implemented the lazy evaluation in the middle game (also well known) >then the fractional plys with the recapture extension, push pawn extension and >null move threat extension. these extensions wasn´t implemented in fruit. so i >decided to test them and it seemed to work well with fruit. >afterwards i noticed that the program often exchanged 3 pawns for 1 officer. in >most cases this was bad for fruit. so i implemented an penalty for this sort of >exchanges. >i like the history pruning in fruit. i implemented a more agressive depth >reduction for history moves and also implemented a research for moves which fail >high after reduction. >i hope i have nothing forgot. but i think that where the main changes. >you see there was nothing new. but ist seemed to work well with fruit. Even at that, certainly interesting and valuable. In addition, the effectiveness of these simple cures makes them worth study. If I can labor for a week and add 3000 lines of code for 5 Elo, or if I can labor for 30 minutes and add 30 lines of code and add 100 Elo, it is a no-brainer which idea to try first.
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.