how far is too far: houdini for sell?

[geots]

I thought there was considerable evidence that houdini was a close derivative of rybka. Now it is for sell? Whatever your view of rybka (legal or derivitive), there is 0 doubt about Houdini being unoriginal. sigh.

If only we could see a return of crafty to number 1, or the program of any honest programmer.

Plagiarism rewarded.

Best
J

I did not buy houdini but I see no problem with buying non original programs.

Ok to buy "stolen merchandise" as well? I mean, YOU didn't actually steal it...
Wait, you might get charged with "receiving stolen merchandise" anyway... "caveat emptor".

Non original is not the same as stolen and
I see no proof that houdini is illegal.

Uri

I will repeat. Houdini came from Robo*. IF, and not a very big IF, robo is found to have fruit code, and we KNOW it contains Vas code, copying any of that code is certainly a criminal act, because it is a clear copyright infringement.

IF, and that is a GIGANTIC IF, houdini contains no Fruit or Rybka code, it will be OK. But Robolito certainly contains both Fruit and Rybka code. And that is a problem in that one can not then take the robo source and use it as if nothing is wrong, because that is very similar to receiving stolen property and then reselling it. That IS also a criminal act...

Just wait a bit. Took a while to prove Rybka was based on Fruit/Crafty. Give it a little more time for more evidence to be presented. Won't be long until the cat is out of the bag, again. And this won't be the last such case, either...

Well be sure the authorities know I have an illegal copy, and if there is a God in Heaven, they would send you to my door to confiscate it or make a citizen's arrest. Clint Eastwood aint got shit on me, IOW- MMD.

[rbarreira]

Rybka 1.6.1 [..] was a full copy of Crafty

I read that Rybka 1.6.1 was at 2100 ELO level. Can you tell us where the 'full copy' job went wrong, since Crafty was much stronger at that time.

I have not tried to look. It would require a complete RE of at least the full eval and search, and that isn't worth the significant effort it would require.

So it was not a full copy then?

[Harvey Williamson]

Rybka 1.6.1 [..] was a full copy of Crafty

I read that Rybka 1.6.1 was at 2100 ELO level. Can you tell us where the 'full copy' job went wrong, since Crafty was much stronger at that time.

I have not tried to look. It would require a complete RE of at least the full eval and search, and that isn't worth the significant effort it would require.

So it was not a full copy then?

One thing to note in that CCT tournament was that Crafty was running on good hardware. The Crafty(Rybka) engine was running on very average hardware.

[Uri Blass]

Rybka 1.6.1 [..] was a full copy of Crafty

I read that Rybka 1.6.1 was at 2100 ELO level. Can you tell us where the 'full copy' job went wrong, since Crafty was much stronger at that time.

I have not tried to look. It would require a complete RE of at least the full eval and search, and that isn't worth the significant effort it would require.

So it was not a full copy then?

One thing to note in that CCT tournament was that Crafty was running on good hardware. The Crafty(Rybka) engine was running on very average hardware.

It is not the main explanation to Rybka's bad results because programs that are not stronger than Crafty could score 50% with average hardware in that CCT tournament when Rybka1.6.1 was clearly weaker than that.

[Harvey Williamson]

Rybka 1.6.1 [..] was a full copy of Crafty

I read that Rybka 1.6.1 was at 2100 ELO level. Can you tell us where the 'full copy' job went wrong, since Crafty was much stronger at that time.

I have not tried to look. It would require a complete RE of at least the full eval and search, and that isn't worth the significant effort it would require.

So it was not a full copy then?

One thing to note in that CCT tournament was that Crafty was running on good hardware. The Crafty(Rybka) engine was running on very average hardware.

It is not the main explanation to Rybka's bad results because programs that are not stronger than Crafty could score 50% with average hardware in that CCT tournament when Rybka1.6.1 was clearly weaker than that.

The main explenation is probably that vas broke something.

[geots]

Rybka 1.6.1 [..] was a full copy of Crafty

I read that Rybka 1.6.1 was at 2100 ELO level. Can you tell us where the 'full copy' job went wrong, since Crafty was much stronger at that time.

I have not tried to look. It would require a complete RE of at least the full eval and search, and that isn't worth the significant effort it would require.

So it was not a full copy then?

One thing to note in that CCT tournament was that Crafty was running on good hardware. The Crafty(Rybka) engine was running on very average hardware.

It is not the main explanation to Rybka's bad results because programs that are not stronger than Crafty could score 50% with average hardware in that CCT tournament when Rybka1.6.1 was clearly weaker than that.

The more you sniff, the worse it smells doesn't it. It's "semantics and misdirection" time now, starring our favorite master of confusion.......
That shovel can't have much of an edge left, as many holes as it has dug.

[rbarreira]

Rybka 1.6.1 [..] was a full copy of Crafty

I read that Rybka 1.6.1 was at 2100 ELO level. Can you tell us where the 'full copy' job went wrong, since Crafty was much stronger at that time.

I have not tried to look. It would require a complete RE of at least the full eval and search, and that isn't worth the significant effort it would require.

So it was not a full copy then?

One thing to note in that CCT tournament was that Crafty was running on good hardware. The Crafty(Rybka) engine was running on very average hardware.

Not to diminish the value of that observation, but what I was going for with my post was to try and understand the degree of certainty of Bob's claim that Rybka 1.6.1 = Crafty.

[bob]

Rybka 1.6.1 [..] was a full copy of Crafty

I read that Rybka 1.6.1 was at 2100 ELO level. Can you tell us where the 'full copy' job went wrong, since Crafty was much stronger at that time.

I have not tried to look. It would require a complete RE of at least the full eval and search, and that isn't worth the significant effort it would require.

So it was not a full copy then?

Restated AGAIN, for the N+1th time:

Mark/Zach spot-checked the search, parts of the eval, the "move ordering/selection" code in NextMove, NextEvasion, etc. In every case, it matched Crafty's internal design / code perfectly. They found several "bugs" that were left over and which could not happen via independent development. I then went through the rybka 1.6.1 binary again. I found that code like the benchmark.c code was copied exactly also. The rotated bitboard code is an exact match, including the bishop (diagonal) shifts arrays and such. In all the checking that was done, which looked at a LOT of the code, everything matched up perfectly. Crafty's unusual "phase" method of ordering the moves, some hashing glitches where one thing is initialized twice, it's "mate" (white/black) was not initialized at all, etc...

There is zero doubt that 1.6.1 was a complete copy. What was modified? Don't know, but it was not significant. With all the spot-checks done, one would expect to see something different here and there if there were significant changes...

[bob]

Rybka 1.6.1 [..] was a full copy of Crafty

I read that Rybka 1.6.1 was at 2100 ELO level. Can you tell us where the 'full copy' job went wrong, since Crafty was much stronger at that time.

I have not tried to look. It would require a complete RE of at least the full eval and search, and that isn't worth the significant effort it would require.

So it was not a full copy then?

One thing to note in that CCT tournament was that Crafty was running on good hardware. The Crafty(Rybka) engine was running on very average hardware.

It is not the main explanation to Rybka's bad results because programs that are not stronger than Crafty could score 50% with average hardware in that CCT tournament when Rybka1.6.1 was clearly weaker than that.

IMHO, it is not even clear that 1.6.1 was what played in the CCT event. The prior versions seemed to be yet another program, looking nothing like Crafty. There is no way to actually verify what played in the tournament, all we have to go on was what Vas supplied at the time. If someone has the time, and the energy, it would be interesting to see what the FIRST version was based on if it was not original. We've found enough copying already to make this less than interesting to me.

[bob]

Rybka 1.6.1 [..] was a full copy of Crafty

I read that Rybka 1.6.1 was at 2100 ELO level. Can you tell us where the 'full copy' job went wrong, since Crafty was much stronger at that time.

I have not tried to look. It would require a complete RE of at least the full eval and search, and that isn't worth the significant effort it would require.

So it was not a full copy then?

One thing to note in that CCT tournament was that Crafty was running on good hardware. The Crafty(Rybka) engine was running on very average hardware.

It is not the main explanation to Rybka's bad results because programs that are not stronger than Crafty could score 50% with average hardware in that CCT tournament when Rybka1.6.1 was clearly weaker than that.

The more you sniff, the worse it smells doesn't it. It's "semantics and misdirection" time now, starring our favorite master of confusion.......
That shovel can't have much of an edge left, as many holes as it has dug.

Such "truths" often do "stink" when you don't like what they are. But they are STILL "truth", unfortunately for you.