Author: Uri Blass
Date: 09:06:39 09/16/05
Go up one level in this thread
On September 16, 2005 at 11:49:06, Roger D Davis wrote: >On September 16, 2005 at 05:42:19, Uri Blass wrote: > >>On September 16, 2005 at 05:28:30, Gian-Carlo Pascutto wrote: >> >>>On September 16, 2005 at 05:13:08, Heinz van Kempen wrote: >>> >>>>Now we finally know that it was a myth claiming that engines at highest level >>>>can only improve by a few rating points each year with a lot of work and testing >>>>and by adding additional knowledge. >>> >>>Why is this a myth? >>> >>>You think Fabien hasn't been working hard on his engine and testing a lot? >>> >>>-- >>>GCP >> >>The question if Fabien works hard is not the question. >> >>The point is that people believed that programs at the high level need hard work >>to improve 50 elo points in a year when Fruit improves by 100 elo in few months >>even from the high level of Fruit2.1 and note that Fruit WCCC still has search >>bugs. >> >>Fabien proved that it is possible to get something near 100 elo improvement in 2 >>or 3 months(the time between fruit2.1 and Fruit WCCC). >> >>If it is possible to do it in 2 or 3 months by hard work then it is clearly >>possible to do it in one year when you work about the program not more than >>average time of 2 hours per day. >> >>Note that I never believed in the theory that programs at the high level needs >>hard work to improve by 50 elo and my thoery was simply that the programmers >>that work about chess programs are not good programmers. >> >>Uri > >And the code for Fruit is supposed to be very simple, right? How does he work so >hard and keep the code so simple? Seems like a contradiction: Working for the >simple. > >Roger I do not have that assumption that the code of fruit is very simple. I have more important things to do now then to understand fruit but I may try to understand nmore of fruit when I have time for it. Uri
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.