Ant_Gugdin wrote:Almost all of the users of these lists will know that they are meaningless because they don't show the strongest engine. It's like having a football league table with the top team omitted, which would be ridiculous.
You make the false assumption that anyone would care what the strongest engine is. I, for one, could not care less. I am only interested where my own engines are on the list, and which are my nearby competitors that would be useful as opponents in my test gauntlets. I also do not care if the list is complete ornot, aslong as I can find many engines on the list. in the rating range I am interested in. A single engine is never of interest; you always have to test against many different opponents, so if engine X is in my standard gauntlet of 20 opponents or not is quite insignificant.
By your argument any national /regional / local football league table would be meaningless, because it does not include the World Champion. Yet my news-paper is fullf them every week. So this is another totally ridiculous assertion.
I think it is a very good thing that CCRL does not waste their time on all those nearly-alike Ippolit derivatives. That leaves them more time to spend on useful testing.
Ant_Gugdin wrote:
Jury, do you have a theory on where Ippolit comes from, then?
Of course, Ippolit was made from some version of Rybka. But not from Rybka3 - there are too many differences. Such modifications could make only the author of Rybka.
I have no idea who released Ippolit and for what purpose. But without Vas is just not done. Perhaps the source had been stolen, or we are dealing with a good joke of Vas.
I went through the Rybka code forwards and backwards and took many things.
hgm wrote:You don't seem to understand how different the alleged cases of Ippolit vs Rybka 3 and Rybka vs Fruit are. You talk about them as if they are exactly the same. While they are in fact vastly different:
Fruit was an open-source project. The code and ideas were made public by the author, and everyone was free to take and use them. If Rybka was based on Fruit, this was totally legal. Its author was known and addressble.
Ippolit, on the other hand, seems to be a decompiled code that comes from nowhere, published by 'authors' that do not want to divulge their true identity. If it was derived from a decompiled Rybka, that would make it totally illegal. Any work based on Ippolit would be based on _stolen_ code.
The cases are not _that_ different. All, IMHO, are bad of course. But one can not take a GPL program, copy it, and then sell the result without including the source. That is equally as bad as reverse-engineering a commercial program then releasing that source publicly. They are inverse operations, but I see nothing that makes one preferable over the other since both are _clearly_ wrong...
As far as this tread goes, anyone is free to rate or exclude any programs they want. Tournaments are not going to allow any of the suspected derivatives (robo*/ip*/houdini/fire* and dozens of others) access, which is perfectly reasonable. These events are not funded via public tax dollars and are allowed to be as inclusive/exclusive as they want...
Robert Flesher wrote:From my understanding of the GPL license you have left out ONE key point. I believe that once Vas used the Fruit code, which has been established, he did. He cannot then close the code and make it commercial.
( " The GPL is the first copyleft license for general use, which means that derived works can only be distributed under the same license terms")
By doing so he violated the terms of that license. So why is he exempt from the scrutiny ?
But, I have no problem with CCRL/CEGT, they can test whatever they want.
It is obvious that Rybka could not have any Fruit code, as Rybka is a bitboard engine, and Fruit is mailbox. That pervades the code everywhere, and requires a thorough rewrite.
Why? My search works just fine with bitboards or mailbox representations. My evaluation is tuned to bitboards, but if I rewrote to mailbox, it would not get tossed out. Just have bits and pieces replaced to access the different representation. Converting to bitboards is certainly an invasive process. But not _that_ invasive...
I borrowed lots of pieces of Cray Blitz when I started Crafty. CB was mailbox/vectorized, crafty was bitboard from day 1. yet lots of code fit right in.
Robert Flesher wrote:From my understanding of the GPL license you have left out ONE key point. I believe that once Vas used the Fruit code, which has been established, he did. He cannot then close the code and make it commercial.
( " The GPL is the first copyleft license for general use, which means that derived works can only be distributed under the same license terms")
By doing so he violated the terms of that license. So why is he exempt from the scrutiny ?
But, I have no problem with CCRL/CEGT, they can test whatever they want.
It is obvious that Rybka could not have any Fruit code, as Rybka is a bitboard engine, and Fruit is mailbox. That pervades the code everywhere, and requires a thorough rewrite.
Any code, even one line that is the same, is copying, even if ported/translated to bitboards. Perhaps going to bitboards is a way around the GPL license. But, like Dr. Hyatt said, " you cannot be a little bit pregnant"
Maybe you should give this a looksee.
hgm wrote:You make the false assumption that anyone would care what the strongest engine is. I, for one, could not care less.
Come on HGM, you are really clutching at straws here. Of *course* people care what the strongest engine is.
By your argument any national /regional / local football league table would be meaningless, because it does not include the World Champion. Yet my news-paper is fullf them every week. So this is another totally ridiculous assertion.
Come on, this is more straw clutching. If engine rating lists were split up into "divisions" (which they are not) then of course the absence of Houdini in Division 2 would not render Division 2 irrelevant. This is not an accurate analogy at all.
Ant_Gugdin wrote:
Jury, do you have a theory on where Ippolit comes from, then?
Of course, Ippolit was made from some version of Rybka. But not from Rybka3 - there are too many differences. Such modifications could make only the author of Rybka.
I have no idea who released Ippolit and for what purpose. But without Vas is just not done. Perhaps the source had been stolen, or we are dealing with a good joke of Vas.
Vas' involvement seems very unlikely. Whatever his flaws he is clearly a commercially minded guy. He is hardly going to ruin his own product by allowing stronger free engines to proliferate on the internet.
Perhaps Ippolit is a decompiled version of another strong engine - Shredder or Fritz - with some changes? (As these engines are not as strong as Rybka 3).
hgm wrote:You make the false assumption that anyone would care what the strongest engine is. I, for one, could not care less.
Come on HGM, you are really clutching at straws here. Of *course* people care what the strongest engine is.
By your argument any national /regional / local football league table would be meaningless, because it does not include the World Champion. Yet my news-paper is fullf them every week. So this is another totally ridiculous assertion.
Come on, this is more straw clutching. If engine rating lists were split up into "divisions" (which they are not) then of course the absence of Houdini in Division 2 would not render Division 2 irrelevant. This is not an accurate analogy at all.
Clutching at straws? I just give you the facts. So in your analogy CCRL is 'only' a rating list of division 2. Great! Because that is the division I happen to be interested in. It is the divsion in which most of us play.
If it was upto the whiners here, only 5 or 6 engines would ever be tested. (Houdini, Rybka, Shredder, Fritz, Junior in various versions). It is really marvelous that the CCRL does not have such a narrow-minded interest.
