Don wrote:I hope you go by much more than just the info strings, that is not convincing at all as there is nothing wrong with patterning your output based on another program.
It's not just the order of the info strings, it's also the place where they appear. This is different in different positions and equal in the output of Fruit and Houdini.
After I saw Pablo's post http://talkchess.com/forum/viewtopic.ph ... 95&t=40294 It looks like Houdini is by far closer to RobboLito than to Fruit. I guess even a sceptical person can't think this is coincidence. Even the counted nodes are equal until depth 4.
Can we agree that Houdini is based on Robbo?
Then my OP would mean that Robbo is based on Fruit. I think a lot of discussions are done about this topic. No need to warm this up
Certainly, the node match is the strongest proof. And this happens in any position:
Houdini w32 1_CPU
build 2010-05-15
by Robert Houdart
position startpos moves e2e4
go depth 8
info depth 1 seldepth 2 score cp -13 time 0 nodes 21 nps 0 pv g8h6
info depth 1 seldepth 2 score cp 13 time 0 nodes 22 nps 0 pv g8f6
info depth 2 seldepth 15 score cp -1 time 1 nodes 284 nps 284000 pv g8f6 b1c3 b8c6 g1f3 d7d5
info depth 3 seldepth 15 score cp -1 time 1 nodes 418 nps 418000 pv g8f6 b1c3 b8c6 g1f3 d7d5
info depth 4 seldepth 15 score cp -5 time 2 nodes 915 nps 457000 pv g8f6 e4e5 f6e4 b1c3 d7d5 d1f3
info depth 4 seldepth 15 score cp -1 time 3 nodes 1126 nps 375000 pv b8c6 b1c3 g8f6 g1f3 d7d5
info depth 5 seldepth 15 score cp -1 time 3 nodes 1439 nps 479000 pv b8c6 b1c3 g8f6 g1f3 d7d5
info depth 6
info depth 6 seldepth 22 score cp 1 time 5 nodes 2521 nps 504000 pv b8c6 b1c3 g8f6 g1f3 d7d5 e4d5 f6d5 f1d3 e7e5 d1e2
info depth 7
info depth 7 seldepth 22 score cp 1 time 7 nodes 3781 nps 540000 pv b8c6 b1c3 g8f6 g1f3 d7d5 e4d5 f6d5 f1d3 e7e5 d1e2
info depth 8
info depth 8 seldepth 25 score cp 1 time 12 nodes 5618 nps 468000 pv b8c6 b1c3 g8f6 g1f3 d7d5 e4d5 f6d5 f1d3 e7e5 d1e2
info depth 9
info depth 8 seldepth 25 score cp 1 time 20 nodes 9569 nps 478000 pv b8c6 b1c3 g8f6 g1f3 d7d5 e4d5 f6d5 f1d3 e7e5 d1e2
bestmove b8c6 ponder b1c3
You can ignore the evaluations, since they are obfuscated in Houdini.
Oh my God
I rest my case regards,
Dr.D
_No one can hit as hard as life.But it ain’t about how hard you can hit.It’s about how hard you can get hit and keep moving forward.How much you can take and keep moving forward….
Don wrote:I hope you go by much more than just the info strings, that is not convincing at all as there is nothing wrong with patterning your output based on another program.
It's not just the order of the info strings, it's also the place where they appear. This is different in different positions and equal in the output of Fruit and Houdini.
After I saw Pablo's post http://talkchess.com/forum/viewtopic.ph ... 95&t=40294 It looks like Houdini is by far closer to RobboLito than to Fruit. I guess even a sceptical person can't think this is coincidence. Even the counted nodes are equal until depth 4.
Can we agree that Houdini is based on Robbo?
Then my OP would mean that Robbo is based on Fruit. I think a lot of discussions are done about this topic. No need to warm this up
Certainly, the node match is the strongest proof. And this happens in any position:
Houdini w32 1_CPU
build 2010-05-15
by Robert Houdart
position startpos moves e2e4
go depth 8
info depth 1 seldepth 2 score cp -13 time 0 nodes 21 nps 0 pv g8h6
info depth 1 seldepth 2 score cp 13 time 0 nodes 22 nps 0 pv g8f6
info depth 2 seldepth 15 score cp -1 time 1 nodes 284 nps 284000 pv g8f6 b1c3 b8c6 g1f3 d7d5
info depth 3 seldepth 15 score cp -1 time 1 nodes 418 nps 418000 pv g8f6 b1c3 b8c6 g1f3 d7d5
info depth 4 seldepth 15 score cp -5 time 2 nodes 915 nps 457000 pv g8f6 e4e5 f6e4 b1c3 d7d5 d1f3
info depth 4 seldepth 15 score cp -1 time 3 nodes 1126 nps 375000 pv b8c6 b1c3 g8f6 g1f3 d7d5
info depth 5 seldepth 15 score cp -1 time 3 nodes 1439 nps 479000 pv b8c6 b1c3 g8f6 g1f3 d7d5
info depth 6
info depth 6 seldepth 22 score cp 1 time 5 nodes 2521 nps 504000 pv b8c6 b1c3 g8f6 g1f3 d7d5 e4d5 f6d5 f1d3 e7e5 d1e2
info depth 7
info depth 7 seldepth 22 score cp 1 time 7 nodes 3781 nps 540000 pv b8c6 b1c3 g8f6 g1f3 d7d5 e4d5 f6d5 f1d3 e7e5 d1e2
info depth 8
info depth 8 seldepth 25 score cp 1 time 12 nodes 5618 nps 468000 pv b8c6 b1c3 g8f6 g1f3 d7d5 e4d5 f6d5 f1d3 e7e5 d1e2
info depth 9
info depth 8 seldepth 25 score cp 1 time 20 nodes 9569 nps 478000 pv b8c6 b1c3 g8f6 g1f3 d7d5 e4d5 f6d5 f1d3 e7e5 d1e2
bestmove b8c6 ponder b1c3
You can ignore the evaluations, since they are obfuscated in Houdini.
I have not looked at the node match information. However I can probably make nodes match by running several million positions on different programs and keeping the ones that match - sooner or later I can probably get lots of match examples.
I don't know anything about the node match stuff here - but you would have to pick a lot of positions independently and then show that "too many" are the same. You could do that with positions that have certain characteristics of course, such as mates or something - but you need to show that no other program matches and of course combine this with other evidence too. You must be careful that the match is not easily explained by simple things.
I have no idea whether that applies in this case without carefully looking at it myself.
I would hope that being the author of the best chess entity in the world is surely an achievement boosting your consulting carrier even if the consulting may apply to totally unrelated areas.
This is also a paragraph on your CV that may raise a few eyebrows
I just looked at this more carefully, I thought it was a hand picked example but it's right after playing "e4" in the opening. Not only do the nodes match exactly, but so does the PV. That is pretty damning evidence even by itself.
Do you have that version of Houdini somewhere that I can get a copy? I want to check this myself.
By the way, when I built the similarity tester I found that 2 of the closest matching programs was Robolitto and Houdini. The match percentage was higher than even 2 different releases of most programs.
Pablo Vazquez wrote:
Alexander Schmidt wrote:
Don wrote:I hope you go by much more than just the info strings, that is not convincing at all as there is nothing wrong with patterning your output based on another program.
It's not just the order of the info strings, it's also the place where they appear. This is different in different positions and equal in the output of Fruit and Houdini.
After I saw Pablo's post http://talkchess.com/forum/viewtopic.ph ... 95&t=40294 It looks like Houdini is by far closer to RobboLito than to Fruit. I guess even a sceptical person can't think this is coincidence. Even the counted nodes are equal until depth 4.
Can we agree that Houdini is based on Robbo?
Then my OP would mean that Robbo is based on Fruit. I think a lot of discussions are done about this topic. No need to warm this up
Certainly, the node match is the strongest proof. And this happens in any position:
Don wrote:I just looked at this more carefully, I thought it was a hand picked example but it's right after playing "e4" in the opening. Not only do the nodes match exactly, but so does the PV. That is pretty damning evidence even by itself.
Do you have that version of Houdini somewhere that I can get a copy? I want to check this myself.
By the way, when I built the similarity tester I found that 2 of the closest matching programs was Robolitto and Houdini. The match percentage was higher than even 2 different releases of most programs.
Pablo Vazquez wrote:
Alexander Schmidt wrote:
Don wrote:I hope you go by much more than just the info strings, that is not convincing at all as there is nothing wrong with patterning your output based on another program.
It's not just the order of the info strings, it's also the place where they appear. This is different in different positions and equal in the output of Fruit and Houdini.
After I saw Pablo's post http://talkchess.com/forum/viewtopic.ph ... 95&t=40294 It looks like Houdini is by far closer to RobboLito than to Fruit. I guess even a sceptical person can't think this is coincidence. Even the counted nodes are equal until depth 4.
Can we agree that Houdini is based on Robbo?
Then my OP would mean that Robbo is based on Fruit. I think a lot of discussions are done about this topic. No need to warm this up
Certainly, the node match is the strongest proof. And this happens in any position:
I will try to find the exact thread at www.open-chess.org where this attachment was originally posted, since there were interesting discussions about the topic.
melajara wrote:I would hope that being the author of the best chess entity in the world is surely an achievement boosting your consulting carrier even if the consulting may apply to totally unrelated areas.
This is also a paragraph on your CV that may raise a few eyebrows
Don wrote:I just looked at this more carefully, I thought it was a hand picked example but it's right after playing "e4" in the opening. Not only do the nodes match exactly, but so does the PV. That is pretty damning evidence even by itself.
Do you have that version of Houdini somewhere that I can get a copy? I want to check this myself.
By the way, when I built the similarity tester I found that 2 of the closest matching programs was Robolitto and Houdini. The match percentage was higher than even 2 different releases of most programs.
Don, are you completely oblivious to the results of your own Sim03 test? Houdini even 1.5 is closer in its output to Robbo (and Ivan, for example) than Rybka 4 is to Rybka 3. Komodo is not, and I wouldn't mind if you go commercial.
