More proof of Strelka cloning

[tiger]

Excellent point Stan, I could not have said it better.


// Christophe

[Uri Blass]

Difficult to believe that. Take a look at his explanation. Nobody engage in such arduous job just to make a joke. BTW, people are not stupids because they believed or still believe there is not cheat.
According the explanation, I tend to believe is not a clone and not a not-clone either. I see a mix of things, a complicated event.
Else: have you used your positional tests with other programs? I am curious to know how in such weird position engines behave.

My best
Fernando

I did not do testing with many engines in these positions.

I did it with strelka only because I read in the rybka forum a claim that it is a clone so I decided to check it.

Note that kateky I used these type of positions to compare between movei's evaluation and rybka's evaluation to see difference and get an idea about general direction for changes but even testing the idea of changing the evaluation to be closer to rybka's evaluation is not going to cause my program to have identical evaluation because I care about cases of big difference in score and if the difference in evaluation is 0.1 pawns or 0.2 pawns I do not care about it.

Uri

[Michael Sherwin]

I have mixed feelings about strong open source chess programs. Learning to program chess was next to impossible for me. Books and pseudo code was not enough. Strong open source programs were to complicated. TSCP was just right, for me. I modified TSCP to use a bitboard move generator and sent it in to Mr. Kerrigan, the author of TSCP. I thought that he might be interested in it. I had no idea that he would put it on his web site. Last time I looked it was still there. If not for TSCP and the experiance I got from modifying it, I doubt that there would be a RomiChess.

So far RomiChess is all mine. I have not borrowed even one line of code from another program and it is chock full of both good and crappy original ideas and is close to 2500 elo at CCRL. Once that I sort out all the new good and crappy ideas then I expect to top 2500 real soon. If it were not for strong open source programs like Crafty and (in its day) Phalanx, I believe that today I would have a top contender and that I would be having a lot more fun!

Today, if I want to have a top contender then I must now learn all I can from studying strong open source engines and then improve upon them. Now I have no choice.

But, in a way I am looking forward to it as then I know that I will have one heck of a good engine!

[Dann Corbit]

Are you actually glad that you did not learn from a better program?

Is every algorithm in your program your invention?

If I had my way, all of the strongest algorithms would be openly known. I honestly cannot imagine that it would be a bad thing. Of course, I do recognize the right someone has to keep their own invention secret.

And besides, I'm not king of the forest anyway.

[Michael Sherwin]

Are you actually glad that you did not learn from a better program?

Is every algorithm in your program your invention?

If I had my way, all of the strongest algorithms would be openly known. I honestly cannot imagine that it would be a bad thing. Of course, I do recognize the right someone has to keep their own invention secret.

And besides, I'm not king of the forest anyway.

I knew of alpha/beta from books and finally understood it enough to get started from TSCP. I also read about hash tables and null move, but implemented them in my own way. All else is my own invention. The learning in RomiChess with the bonus/malus is from Pavlov's dog experiments.

I am glad that I have gotten this far on my own, though, I now realize that I will never catch up in this life time if I do not start to study the code of stronger prgrams.

[Dann Corbit]

Are you actually glad that you did not learn from a better program?

Is every algorithm in your program your invention?

If I had my way, all of the strongest algorithms would be openly known. I honestly cannot imagine that it would be a bad thing. Of course, I do recognize the right someone has to keep their own invention secret.

And besides, I'm not king of the forest anyway.

I knew of alpha/beta from books and finally understood it enough to get started from TSCP. I also read about hash tables and null move, but implemented them in my own way. All else is my own invention. The learning in RomiChess with the bonus/malus is from Pavlov's dog experiments.

I am glad that I have gotten this far on my own, though, I now realize that I will never catch up in this life time if I do not start to study the code of stronger prgrams.

I have written a disk based sort which is *MUCH* faster than replacement selection. If I had never studied the state of the art, I would never have known how to beat it. In order to understand the next step in computation, it is absolutely essential to understand the current step. Otherwise, we reinvent the wheel, often with very rough corners. I have (many times) see people put forward 'new and improved' algorithms in news:comp.programming and most of the time they are either inferior versions of existing algorithms or otherwise really old news. If they had read a few papers or books about how things are done, they would not have spent 50 hours inventing something {poorly} that had already been invented, honed, polished, and delivered.

So you are right, you will have to study the existing algorithms if you are going to beat them. But if you never advance your understanding and inject your own ingenuity, you will never do anything better than to tie them.