I have the same first Qxd4 PV, and don't use SEE to order initial root moves.Gian-Carlo Pascutto wrote:By the way, an underlying issue that causes the funny output is that Ippolit is using MVV/LVA scoring to order the moves at the root (in fact, they are passed from a function higher up), whereas they use (proper) SEE in the search tree, and correctly order Qxd4 backwards there.
This is why you won't find much engines with such output. You must have almost the same bug.
a Telltale position
Moderators: hgm, Rebel, chrisw
-
- Posts: 2250
- Joined: Wed Mar 08, 2006 8:47 pm
- Location: Hattingen, Germany
Re: a Telltale position
-
- Posts: 6073
- Joined: Sat Apr 01, 2006 9:34 pm
- Location: Scotland
Re: a Telltale position
I never said that Robbolito was akin to Rybka 2.3.2. I said that the easiest Rybka to disassemble was Rybka 2.3.2b (the tuning/programmable/leaked one) because the symbols had not been stripped and that it is my belief that was what was used to create Ippolit..... (please note Ippolit).
With regard to the evals you see in my first post, what say you Miguel?
Chris
With regard to the evals you see in my first post, what say you Miguel?
Chris
-
- Posts: 2250
- Joined: Wed Mar 08, 2006 8:47 pm
- Location: Hattingen, Germany
Re: a Telltale position
Hi Richard,Richard Allbert wrote:Hi Gerd!
I think if you feed this to a lot of engines, there aren't many that produce a score -900 < val < -800 on the first pv string.
you mean the score is too low? I have statically -982.
I would liked to come as a visitor. I will not come. My mother is very ill and dying ...Richard Allbert wrote:See you Belgium
Richard
Gerd
-
- Posts: 644
- Joined: Fri Feb 02, 2007 3:11 am
- Location: New Zealand
Re: a Telltale position
Thanks.I think I know what is going on and can connect the dotslkaufman wrote:I ran off a quick 100 games using the Monte Carlo feature of Rybka 4 at five ply (which is really 8 ply). White won by 78 to 22 confirming the human GM assessment. I imagine that really ancient engines might score this around zero, but this should have no relevance to how current engines evaluate.Cubeman wrote:How do the games from that position end, it would be interesting for some test games between the so called Ippo clones and the traditional other strong engines.A wrong evaluation would show up in game results.Sometimes I think Human evaluations are not necessary the absolute truth.I also imagine that there could be many engines out there even before Rybka beta that would evaluate similar scores as Houdini and Critter.
-
- Posts: 6401
- Joined: Thu Mar 09, 2006 8:30 pm
- Location: Chicago, Illinois, USA
Re: a Telltale position
I know, but Larry extrapolated that to evals, and that is not correct.Christopher Conkie wrote:I never said that Robbolito was akin to Rybka 2.3.2. I said that the easiest Rybka to disassemble was Rybka 2.3.2b (the tuning/programmable/leaked one) because the symbols had not been stripped and that it is my belief that was what was used to create Ippolit..... (please note Ippolit).
Like I implied in my post, you have a very good point here (the coincidences in PV + eval are really striking.). That should be stressed, not Qxd4 and the near neutral eval. Otherwise, it dilutes the argument.With regard to the evals you see in my first post, what say you Miguel?
Miguel
Chris
-
- Posts: 6073
- Joined: Sat Apr 01, 2006 9:34 pm
- Location: Scotland
Re: a Telltale position
Ok. Just wanting to make myself clear.michiguel wrote:I know, but Larry extrapolated that to evals, and that is not correct.Christopher Conkie wrote:I never said that Robbolito was akin to Rybka 2.3.2. I said that the easiest Rybka to disassemble was Rybka 2.3.2b (the tuning/programmable/leaked one) because the symbols had not been stripped and that it is my belief that was what was used to create Ippolit..... (please note Ippolit).
Like I implied in my post, you have a very good point here (the coincidences in PV + eval are really striking.). That should be stressed, not Qxd4 and the near neutral eval. Otherwise, it dilutes the argument.With regard to the evals you see in my first post, what say you Miguel?
Miguel
Chris
I think it is helpful at this point to remember this thread Miguel.
http://www.talkchess.com/forum/viewtopi ... 01&t=34333
In this thread he said the following......
http://www.talkchess.com/forum/viewtopi ... 44&t=34333
I am here to tell you all that there is a 0% chance that Houdini is not based on Robbolito.
Chris
-
- Posts: 792
- Joined: Wed Jul 19, 2006 9:58 am
Re: a Telltale position
I'm sorry to read that, Gerd. I had a bad 2010 for the same reason.
Thoughts with you
Richard
Thoughts with you
Richard
-
- Posts: 644
- Joined: Fri Feb 02, 2007 3:11 am
- Location: New Zealand
Re: a Telltale position
So these words "based on" and "taking ideas from" have different meanings and can be interpreted how ever you want.I wonder why he choose to also take the bad ideas too, like the evaluation of this position.I just wish that some authors will come clean with their real contributions.We don't want a repeat of the Rybka-Fruit fiasco.
Is it possible to take code from other engine and then change it ever so slightly and then turn around with a straight face and say you only took the idea?You may be able to kid others but not yourself.
Is it possible to take code from other engine and then change it ever so slightly and then turn around with a straight face and say you only took the idea?You may be able to kid others but not yourself.
-
- Posts: 5960
- Joined: Sun Jan 10, 2010 6:15 am
- Location: Maryland USA
Re: a Telltale position
You are quite correct that the coincident PV and eval is far more telling than the near-zero evals. It is just that the evals are what called my attention to this position initially. I felt that any top program that doesn't see White's advantage either has a poorly written eval or has duplicated Rybka's error in this respect. If your program doesn't see White's advantage here, I think you can improve it by adjusting parameters so as to show White's advantage in this position. I guess I'm now helping Houdini, Ivanhoe, and Critter improve their programs!michiguel wrote:I know, but Larry extrapolated that to evals, and that is not correct.Christopher Conkie wrote:I never said that Robbolito was akin to Rybka 2.3.2. I said that the easiest Rybka to disassemble was Rybka 2.3.2b (the tuning/programmable/leaked one) because the symbols had not been stripped and that it is my belief that was what was used to create Ippolit..... (please note Ippolit).
Like I implied in my post, you have a very good point here (the coincidences in PV + eval are really striking.). That should be stressed, not Qxd4 and the near neutral eval. Otherwise, it dilutes the argument.With regard to the evals you see in my first post, what say you Miguel?
Miguel
Chris
-
- Posts: 6401
- Joined: Thu Mar 09, 2006 8:30 pm
- Location: Chicago, Illinois, USA
Re: a Telltale position
Or course I agree, there is room for improvement for my engine here (in many other departments too ).lkaufman wrote:You are quite correct that the coincident PV and eval is far more telling than the near-zero evals. It is just that the evals are what called my attention to this position initially. I felt that any top program that doesn't see White's advantage either has a poorly written eval or has duplicated Rybka's error in this respect. If your program doesn't see White's advantage here, I think you can improve it by adjusting parameters so as to show White's advantage in this position.michiguel wrote:I know, but Larry extrapolated that to evals, and that is not correct.Christopher Conkie wrote:I never said that Robbolito was akin to Rybka 2.3.2. I said that the easiest Rybka to disassemble was Rybka 2.3.2b (the tuning/programmable/leaked one) because the symbols had not been stripped and that it is my belief that was what was used to create Ippolit..... (please note Ippolit).
Like I implied in my post, you have a very good point here (the coincidences in PV + eval are really striking.). That should be stressed, not Qxd4 and the near neutral eval. Otherwise, it dilutes the argument.With regard to the evals you see in my first post, what say you Miguel?
Miguel
Chris
In my case, I think I cannot improve it by tuning the parameters. I bet that if I try, I will make it weaker (it is not that simple without messing up everything else). Most likely, I am missing a parameter, which may not be the same. That is true in science too (and it may have been R2 case). When the fitting is not good, a parameter may be missing.
Miguel
I guess I'm now helping Houdini, Ivanhoe, and Critter improve their programs!