Don wrote:If the ICGA were to open this without any kind of restriction, there would be no reason not to bring Houdini to the tournament. Since this is "unrestricted" it should not even be required to modify it. So why not have several Houdini's at the tournament?
Very strange logic.
If Houdini is allowed in a tournament, all of the sudden "several Houdini's" would appear? From where exactly?
In reality there would be as much chance of having "several Houdini's" as there would be of having "several Komodo's".
Robert
I thing you misunderstood my statement as an accusation against you. Here is a re-wording of what I said and I think you will see the logic is sound:
I said that if the ICGA were to have NO restrictions on authorship (which means anybody can bring anything) to a tournament, then what would prevent every wannabe author from bringing the strongest program available (which happens to be Houdini) to tournaments?
Do you have an answer for that that defeats my logic?
The tone of these discussions is that the ICGA has no right to question the origins of our programs and that this is none of their business. There are many who feel that way.
Nobody would pick Komodo or any other program in such a situation as Houdini is stronger.
From where would these Houdini's appear you ask? From your web site of course. Any wannabe author could grab the latest version of Houdini and without modification represent it as their own program, that is if the ICGA would only stay out of our business and let us do anything we want to (sarcasm intended.)
Don wrote:If the ICGA were to open this without any kind of restriction, there would be no reason not to bring Houdini to the tournament. Since this is "unrestricted" it should not even be required to modify it. So why not have several Houdini's at the tournament?
Very strange logic.
If Houdini is allowed in a tournament, all of the sudden "several Houdini's" would appear? From where exactly?
In reality there would be as much chance of having "several Houdini's" as there would be of having "several Komodo's".
Robert
Well, I would make a copy, of course, and do a hex-edit job or adapter trick on it to change the name, and enter it. No one would check the binary or ask for sources, since anything flies... So what is stopping me?
No, you don't even have to do that. I specified that there is no restriction. In fact I can just announce that I'm running Houdini in the spirit of cooperative progress in computer chess and perhaps I could change one of the parameters to make it a "truly unique" entry in order to make myself feel better about what I'm doing.
If the ICGA standards were slightly higher, I would be forced to obfuscate what I was doing. I could of course do as you say, edit my name into the binary and/or wrap it up in a bogus front end. The extra code would make the binary size different and I could filter and transform the output. You only have to create just a little doubt in order to get the support of those who are gullible and have lost the ability to think critically, objectively.
Uri Blass wrote:I can only say that I dislike the rules.
My opinion is that it is better not to have rules that not allow copying.
If we want better programs then we should allow it for everyone.
I suspect that for example an average of evaluations of stockfish and komodo may give a better program and if the author is not allowed to try it then it is bad for computer chess.
I understand that people do not like that other use their work to get a better starting point but the rules should be the same for everyone and they should also be allowed to use the work of other to get a better starting point.
Are you going to play in a tournament with 30 copies of robolito? I am not. That's the reason for this rule. Anyone can organize a tournament. HGM has been doing one for quite a while. Others can follow suit. And establish any rules they think are appropriate. You will likely end up with only clones, however, as the original authors will not be interested.
Uri Blass wrote:I can only say that I dislike the rules.
My opinion is that it is better not to have rules that not allow copying.
If we want better programs then we should allow it for everyone.
If on the other hand, we want a fair competition among the world's best chess engine programmers, then the current rules are necessary.
Otherwise, the next WCCC would be a battle between five Ippolit clones, none of whose "authors" had put more than one month of effort into them. I don't think such a "championship" would be very interesting, do you?
I do not think that Ippolit clones when the authors did not put more than one month of effort into them can be competitive with the latest houdini or the latest cluster Rybka.
Do a little analysis. If you have 30 programs, all within 50-60 points of each other, and you play a swiss, you have a coin-toss winner. With so many strong opponents, everybody is going to lose games...
How, exactly, is such an event going to be fun for an _original_ program author?
Uri Blass wrote:I can only say that I dislike the rules.
My opinion is that it is better not to have rules that not allow copying.
If we want better programs then we should allow it for everyone.
If on the other hand, we want a fair competition among the world's best chess engine programmers, then the current rules are necessary.
Otherwise, the next WCCC would be a battle between five Ippolit clones, none of whose "authors" had put more than one month of effort into them. I don't think such a "championship" would be very interesting, do you?
I do not think that Ippolit clones when the authors did not put more than one month of effort into them can be competitive with the latest houdini or the latest cluster Rybka.
Do a little analysis. If you have 30 programs, all within 50-60 points of each other, and you play a swiss, you have a coin-toss winner. With so many strong opponents, everybody is going to lose games...
How, exactly, is such an event going to be fun for an _original_ program author?
Most of the ones in favor of this are not program authors, so they do not see our point of view.
However I don't see how this would be much fun for the observers either.
Don wrote:If the ICGA were to open this without any kind of restriction, there would be no reason not to bring Houdini to the tournament. Since this is "unrestricted" it should not even be required to modify it. So why not have several Houdini's at the tournament?
Very strange logic.
If Houdini is allowed in a tournament, all of the sudden "several Houdini's" would appear? From where exactly?
In reality there would be as much chance of having "several Houdini's" as there would be of having "several Komodo's".
Robert
I thing you misunderstood my statement as an accusation against you. Here is a re-wording of what I said and I think you will see the logic is sound:
I said that if the ICGA were to have NO restrictions on authorship (which means anybody can bring anything) to a tournament, then what would prevent every wannabe author from bringing the strongest program available (which happens to be Houdini) to tournaments?
Do you have an answer for that that defeats my logic?
The tone of these discussions is that the ICGA has no right to question the origins of our programs and that this is none of their business. There are many who feel that way.
Nobody would pick Komodo or any other program in such a situation as Houdini is stronger.
From where would these Houdini's appear you ask? From your web site of course. Any wannabe author could grab the latest version of Houdini and without modification represent it as their own program, that is if the ICGA would only stay out of our business and let us do anything we want to (sarcasm intended.)
I dislike the idea of many programs that are the same but I think that the solution to the problem should be to ask the programmers to run some test suites with their program before the event and if more than one program get almost the same outpt in the test suite(no more difference than running the same program twice) then only one of them should be allowed.
almost the same output include also time of getting the same move choice and my opinion is that if 2 programs use the same evaluation when one of them has a significantly different search then both should be allowed.
Although this could help to keep the tournament interesting, it would be impossible to implement without some method to decide which of the identical or nearly identical programs will be allowed to enter, and which will be sent home.
Don wrote:If the ICGA were to open this without any kind of restriction, there would be no reason not to bring Houdini to the tournament. Since this is "unrestricted" it should not even be required to modify it. So why not have several Houdini's at the tournament?
Very strange logic.
If Houdini is allowed in a tournament, all of the sudden "several Houdini's" would appear? From where exactly?
In reality there would be as much chance of having "several Houdini's" as there would be of having "several Komodo's".
Robert
I thing you misunderstood my statement as an accusation against you. Here is a re-wording of what I said and I think you will see the logic is sound:
I said that if the ICGA were to have NO restrictions on authorship (which means anybody can bring anything) to a tournament, then what would prevent every wannabe author from bringing the strongest program available (which happens to be Houdini) to tournaments?
Do you have an answer for that that defeats my logic?
The tone of these discussions is that the ICGA has no right to question the origins of our programs and that this is none of their business. There are many who feel that way.
Nobody would pick Komodo or any other program in such a situation as Houdini is stronger.
From where would these Houdini's appear you ask? From your web site of course. Any wannabe author could grab the latest version of Houdini and without modification represent it as their own program, that is if the ICGA would only stay out of our business and let us do anything we want to (sarcasm intended.)
I dislike the idea of many programs that are the same but I think that the solution to the problem should be to ask the programmers to run some test suites with their program before the event and if more than one program get almost the same outpt in the test suite(no more difference than running the same program twice) then only one of them should be allowed.
almost the same output include also time of getting the same move choice and my opinion is that if 2 programs use the same evaluation when one of them has a significantly different search then both should be allowed.
It's my opinion that the evaluation is the most difficult part of a chess program. I don't think most authors even have the ability produce a really top notch evaluation function. If I am some unknown programmer and suddenly show up with a program called Hoodwinki which is pretty strong and plays exactly like Komodo for instance, it's a sure bet the evaluation was based on Komodo, even if the search and data structure is not. If the search is completely different I don't care, the most difficult part was a cheat because the author was too lazy to take the year or two required to build a highly refined evaluation function or else didn't know how. I can tell you from the example of Komodo that every version improved on the evaluation of Komodo, and we are STILL working on the evaluation function of Komodo. It has evolved and improved over YEARS of time.
So it's very generous of you to give away the evaluation function I worked so hard on for free.
I also am not impressed if the author improved the evaluation function. It's far easier to improve an existing product than to engineer a product of the same quality from scratch - this is not even a close call.
Your argument is still based on the concept of being highly sympathetic to the concept of copying the work of others. I know that you have good intentions but your view does not reflect the views of the program authors. You may be thinking, "who cares what they think, they are in the minority", but the inescapable truth is that the program authors are the ones providing the actual programs. If you cut us off at the knees you won't have any software to argue about.
Don wrote:........ If I am some unknown programmer and suddenly show up with a program called Hoodwinki which is pretty strong and plays exactly like Komodo for instance, it's a sure bet the evaluation was based on Komodo, even if the search and data structure is not.........
Don wrote:
It's my opinion that the evaluation is the most difficult part of a chess program. I don't think most authors even have the ability produce a really top notch evaluation function. I can tell you from the example of Komodo that every version improved on the evaluation of Komodo, and we are STILL working on the evaluation function of Komodo. It has evolved and improved over YEARS of time.
I can tell you, it's really a hard job!
That is exactly or almost what Vas did with Rybka, he took Crafty and Fruit and started from these discoveries, craftmanship and European flavour. Then it took him years to develop something like Rybka 4 and he is still working on very hard. All that has little to do with what the ICGA is alleging.
BTW it's kom-o-do which is a Hawaian dialect and means "comme all do". Hawaian is partially French. I dont know if you knew this, Don.
So I dont get it why Vas is singled out by his peer group.
My advice is, let's not waste our short life on this Earth with bickering and bullying.
Both Vas and Robert H. should be free to be as creative as possible. But I wish to address this to all here around.
-Popper and Lakatos are good but I'm stuck on Leibowitz
