Author: Dann Corbit
Date: 18:04:59 06/19/00
Go up one level in this thread
On June 19, 2000 at 20:58:24, Adrien Regimbald wrote: >Hello, > >I've recently become curious as to how much of a gain each of you are willing to >do a non-trivial ammount of work for. Of course, if the change to code is >trivial, and provides a measurable gain, I imagine that any of us would be >willing to implement the change. > >The debate then lies with the cutoff line between value to strength of the >program and effort to implement it. > >I think it is only fair for me to state before I discuss this that with my >program Faile, I am a bit conservative about adding things to it. The code is >fairly neat, and to some extent, a major goal of my project is to keep the code >as neat and clear as possible so that it will be of use to other people. > >That being said - sometimes I am completely baffled by the lengths to which we >will go to improve our programs. Some improvements will offer at most a 1 or 2% >gain .. and it will take at _least_ 50 of these to attain any noticable >improvement in strength. Not to mention that with the addition of 50 competing >code additions, sometimes code quality deteriorates quickly. > >The commercial programs of course have to use these improvements - they need >every ounce of strength they can squeeze out of their programs, because they >are, after all, getting paid for it. > >My personal approach so far is to incorporate the "major" ideas into my program, >keep it as neat as possible, and release the source for everyone to hopefully >find useful. After I've fixed the bugs, and implemented requested features, >etc, I post a 'final version' on my web page. After that, I tend to tinker >around with my own ideas on my home computer. Some of them actually do end up >having significant payoffs (although they could quite likely have been thought >of already) .. but these are my own ideas. With this in mind, I am really >having a hard time understanding why an author would take an idea that isn't >original and add it to their program only give a very very marginal gain at the >risk of code clarity. > >Is the only goal of the amaeteur chess programmer to squeeze out that last >rating point that they can get out of their program in a constant quest for the >highest rating? If it is, I think we have missed the boat in a big way.. I think it depends on where your program is and where you want it to be. If you have a 1500 ELO program and you want to make big improvements, 5 ELO tweaks are rather silly. A major revamp is in order, such as changes to the fundamental model or addition of major pieces of functionality. But suppose that your program is a top 5 program and the other programs are about 10 ELO stronger than yours. Maybe you are willing to invest a big pile of time in little tweaky things because the investment is worth it to you.
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.