Author: Scott Gasch
Date: 13:29:03 05/28/02
Go up one level in this thread
Beating human GMs and other strong engines is a motivator for me. Related to this, figuring out how the hell (insert strong engine or "the human brain" here) understands position X so quickly... The real reason I love chess programming, though, is hard to understand unless you do it. Here's an analogy: imagine you are an lead engineer of a rally car team designing a car to compete. You can lay the engine out however you want it, change cylinder size / configuration, control the fuel type / mixture ratio, cooling system, chassis weight, suspension, etc. You have complete control of everything. Building a new testmodel is free and fast; all it takes is the time to engineer it and about 30 seconds to "build" it. Then you can test it out... both on "standard" test courses and in real races against other engines whose mettle is already proven. If you are radical and smart enough maybe you'll come up with something no one has ever tried before... Will you new idea be the wankle engine of the chess world? Or if you are less ambitious today then why don't you try to replace the air filter and tweak your fuel injector algorithm looking for 5 extra horsepower... or tweak the suspension for some terrain that most cars are not so good at handling... Scott On May 28, 2002 at 14:47:17, K. Burcham wrote: > > >I have been curious about what motivates chess programmers. What do they really >want from their creation? > >1. Money >2. Number one on SSDF list >3. To Win most games against human GM >4. To gain more Fans >5. To be looked at as number one programmer >6. Attention & fame >7. Only to beat their last version >8. could care less about any of the above---just a casual hobby > > > >I have often wondered who is buying these programs. I have a large family. >If I ask any of these people who Fritz7 is, they would not have a clue. If I ask >someone I work with who Century 4 is, they would not have a clue. 95% of the >people on the street have not heard of Shredder 6 Paderborn. No one I work with >has ever heard of Chessbase. No one in my family has ever heard of chessbase. I >wonder what the income is for the top programmers. Not to know their salary, but >I am curious if the income from their programs is enough to motivate them to >pursue a better program than their last. Of course we are talking about the top >selling commercial programs. > >Also I have noticed in playing GM, most programs could care less if the GM is >closing all files. If there was an expensive program that had been tuned and >tested to play humans, I would buy this. But i would assume there is no market >for the work that it would take to produce this. maybe also this human program >would be weak against other programs on SSDF list. also if there was a "tuned >for programs only book", that was expensive, I would buy this. > >I wonder how some of the programmers test their changes as they decide to work >on another version. This would be very critical to decide to make changes to a >strong program that is already top five on SSDF list. Maybe some have test >positions that they trust to use as a standard. maybe they play other programs >in different time controls and see what results are compared to version before >changes. I cant believe that beta testers give accurate feedback for program >strength adjustments. (except maybe Sarah, Jonas, etc.) it would seem from some >posts and program releases, beta testers are not giving the type of feedback >necessary to prevent bugs being released to public. One of the top programs was >released last year, and it was very unprofessionally managed before release. as >soon as it was released, a huge list developed for patches and complaints. I >laughed at that one, but I bought it anyway. > >If there was an expensive program that was optimized for smp, that was stronger >than other smp commercial programs, I would buy it. But only if the price was >high enough to keep the casual program buyer from purchasing it. > >If a single processor program was released, that proved it was much better than >any other top program, and if it was expensive, I would buy it. > >You have seen what some chess program buyers will spend on hardware. It would be >the same with an expensive program. If a new program was released, and sold for >$350 to $500, became number one on SSDF list by the largest margin. This same >program everyone says is beating all programs, most of the time. At game >servers, there is small group of members that win most games using this program. >Would this sell? I think there is a market for this level of program. >But it would seem that one or more GM would have to be on payroll for helping >programmer with new version. if three 2600 GM were on Chess tiger 16 payroll for >one year full time working on book, knowledge and playing 1000's of games, with >feed back to programmer----seems this would be very strong. also it would seem >if a programmer could employ 20 Sarah types, to test matches with new changes to >program, playing 1000's of games against other top programs to give feedback to >programmer so that he can decide on final changes to make to program. > >Just some thoughts I have. >kburcham
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.