Author: Robert Hyatt
Date: 19:54:07 02/27/04
Go up one level in this thread
On February 27, 2004 at 14:19:39, Vincent Diepeveen wrote: >On February 25, 2004 at 22:36:46, Robert Hyatt wrote: > >>On February 25, 2004 at 21:51:02, Vincent Diepeveen wrote: >> >>>On February 25, 2004 at 20:15:26, Bob Durrett wrote: >>> >>>>On February 25, 2004 at 19:33:38, Vincent Diepeveen wrote: >>>> >>>>>On February 25, 2004 at 18:24:05, Bob Durrett wrote: >>>>> >>>>>> >>>>>>Ladies and Gentlemen: >>>>>> >>>>>>The ideal CCC moderator should be mature, very smart, tough as nails, and above >>>>>>all NOT senile. >>>>> >>>>>And ideally it should not be folks that makes up stories that they own a purple >>>>>heart, when they don't. >>>>> >>>>>>Technical expertise is somewhat important. >>>>>>Currently and in the past we have had some excellent moderators and I trust the >>>>>>same will be true in the future. >>>>> >>>>>In general the candidate level is deteriorating here. >>>>> >>>>>Additionally the computerchess scene gets dominated in reality by >>>>>european/middle east progress, but moderation at CCC by North-Americans. >>>>> >>>>>>Bob Durrett >>>>>> >>>>>>P.S. A few more sophisticated and elegant chess algorithms would be nice too. : >>>>>>) >>>>> >>>>>Good elegant algorithms (or enhancements) never get posted in CCC. >>>>> >>>>>In fact i have invented many algorithms / search methods, which i never posted >>>>>and do not plan to post either. >>>>> >>>>>All but one appeared to be big BS in the end anyway, but one looks real >>>>>promising. >>>>> >>>>>I lack time to implement it, because making money is important in life and in >>>>>general that means not working onto search algorithms, no matter how cool it is >>>>>to do. >>>>> >>>>>Perhaps i will give it a shot 1 week before ict4 :) >>>>> >>>>>Most miserably failed the algorithm where i had put a lot of months work in, >>>>>which started off as a CNS implementation (conspiracy number search). >>>>> >>>>>Also failed was a selective searching search method where i had put in 2 years >>>>>of work (1998+1999). >>>>> >>>>>In general in computerchess experimenting with new search methods is what takes >>>>>a lot of time. >>>>> >>>>>Nowadays also time consuming is of course parallellization. >>>>> >>>>>When talking about search algorithms (also parallel search) i am sure there is >>>>>still a lot to invent. Majority of simple stuff has already been discovered. it >>>>>is very difficult to find new algorithms that use very simple general working >>>>>principles. >>>>> >>>>>However i'm sure there is still a lot to discover when complexity gets added. >>>>> >>>>>The reason why in general at universities never complex stuff gets invented in >>>>>game tree search is simply because the vaste majority, so everyone with one or 2 >>>>>exceptions (Jonathan Schaeffer is one such an exception of a good guy), they are >>>>>busy at a level which is so simple. They still must learn basic stuff and are >>>>>simply busy reinventing what already has been invented then they put 1 condition >>>>>different and they call it a new algorithm (which IMHO is not a new algorithm >>>>>then but at most a new tuning of an existing algorithm). >>>>> >>>>>So they simply are not *busy* creating complex working algorithms. And as i >>>>>already said, all effort spent so far by the same majority of researchers has >>>>>already been put in finding simple algorithms. >>>>> >>>>>Of course it would be cool if someone out of that group comes up with a new >>>>>simple working algorithm that works great. >>>>> >>>>>But the odds are small that they will find it. If someone will find it, it will >>>>>be a computer chess programmer who's not going to post it. >>>>> >>>>>This where when you add complexity to algorithms, there is an entire field open >>>>>to discover new algorithms in. The number of complex search methods published >>>>>(not counting parallellization algorithms of course which are all non trivial to >>>>>implement) which you cannot implement within 5 minutes of your time and from >>>>>which you know in advance that they *must* be tried just in case they work, you >>>>>really can count them on 1 hand. >>>>> >>>>>Yet i'm sure that no coming researcher will focus upon complex algorithms >>>>>either. The problem is simply it takes yourself to program a quite good playing >>>>>chessprogram in order to test simple algorithms and figure out whether they >>>>>work. >>>>> >>>>>Only when a researcher has understanding there he can move on to create some >>>>>more complex algorithms when he has the time. >>>> >>>>Let me put my fortune teller hat on: >>>> >>>>I see considerable change on the near horizon. In the next 20 or 30 years, we >>>>should see great technological improvements in digital, computer, and software >>>>areas. There should be tons of opportunity for chess programming enthusiasts to >>>>delve into new hardware and software concepts and no one should become bored. >>>>Indeed, the very definition of "programming" may change radically in our >>>>lifetimes. >>>> >>>>Now I will take my fortune teller hat off again. >>>> >>>>Hmmmm. What were we talking about? I forgot. Oh well, Spring will be here >>>>soon. That should be good enough. I need another snack. >>>> >>>>Bob D. >>> >>>In general scientists are known for being lazy programmers. In fact they program >>>just too little. So more powerful computers and another few new object oriented >>>programming languages will at most extend their holiday with 1 extra afternoon. >>>Instead of 2 afternoons coding they then code for 1 afternoon a year :) >> >> >>I would bet that this "scientist" has written 100X the number of lines of code >>you have written. >> >>You should stop such stupid generalizations. They make you look like a complete >>idiot. >> >>Actually whether you stop the generalizations or not really won't change that... > >In this posting mine i didn't mean you Bob, if so i would for sure have >mentionned it :) > >Exceptions proof the general rule :) > >I doubt though that you have written more code than i have, despite you being >longer around in programming life. > >Crafty would have its own interface otherwise, to just mention one point :) Poor logic. I've written multiple complete chess programs. At least five in the blitz/cray blitz family, plus crafty. But how many assemblers and compilers have you written? I've written assemblers for the IBM /360, the xerox sigma 9, the HP 2100a (two here, assembler and microcode assembler as well), two basic interpreters, a major part of a cobol compiler for an old TI supercomputer (ASC). A complete hardware emulator for the xerox sigma 9 so I could boot the operating system as a user, to test changes without taking the real machine down. Those represent a million lines of code at a minimum. Should I go on? I wrote about 160K lines of assembly for the xerox operating system to add support for a separate communcations processor offloading all the terminal interrupts to a second processor. I wrote about 150K lines of assembly for the IBM/360 system we used to make it far faster to run student programs. Perhaps you should re-think your comparison? Just because I chose to not write a GUI doesn't mean I haven't written a lot of other things. I just choose to avoid re-inventing the wheel when I can avoid it, and use xboard which works just fine for my requirements. If I wanted to write a GUI, I would, I have done lots of x-windows program development. The key is "if I wanted to" which I don't...
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.