Houdini Engine Origins

[BB+]

Yes, I got the same with:

Houdini w32 1_CPU
build 2010-05-15

Guess I got the build one day earlier.

[Peter C]

I dug up this old thread from Talkchess, which has some pretty conclusive evidence that Houdini comes from Ippo*.

I wouldn't care that Houdini comes from Ippo*, except that the author consistently lies that it doesn't. So, no Houdini for me.

Peter

Another guy that dont care. And until the author fess up no Houdini for Peter. Will the world ever be ok until Peter gets his Houdini

Dr. Ivannik

So what's you're telling me is that it is fine for the author to lie about the origins of his engine and steal code from others? Dude, you're morally messed up.

Peter

[thorstenczub]

So what's you're telling me is that it is fine for the author to lie about the origins of his engine and steal code from others? Dude, you're morally messed up.

Peter

are you talking about houdini or about rybka here ?

[Dr. Ivannik]

I dug up this old thread from Talkchess, which has some pretty conclusive evidence that Houdini comes from Ippo*.

I wouldn't care that Houdini comes from Ippo*, except that the author consistently lies that it doesn't. So, no Houdini for me.

Peter

Another guy that dont care. And until the author fess up no Houdini for Peter. Will the world ever be ok until Peter gets his Houdini

Dr. Ivannik

So what's you're telling me is that it is fine for the author to lie about the origins of his engine and steal code from others? Dude, you're morally messed up.

Peter

Peter

I personally don't know if any author lied about anything or stole anything. These allegations are made by other people like yourself. I don't know where people like you come from??? Accusing people of lying and stealing and then questioning my moral standing in life. In the end I will be judged for my moral character but not by you peter. I am embarrassed for you!! I hope you get your head right.

Thank you

Dr. Ivannik

[Dr. Ivannik]

Jeremy

This ridiculous thread has not convinced anybody of anything. Just the same ole shiit that can be read on talkchess forum. However there is still time to redeem yourself and lessen your own embarrassment. Lock this thread or apologize to Robert Houdart for your baseless and silly allegations. Many chess fans are watching. Do the right thing Jeremy. Do it Jeremy!!!

Thank You

Dr. Ivannik

[Uly]

Lock this thread or apologize to Robert Houdart for your baseless and silly allegations.

Please read the thread.

[Peter C]

I dug up this old thread from Talkchess, which has some pretty conclusive evidence that Houdini comes from Ippo*.

I wouldn't care that Houdini comes from Ippo*, except that the author consistently lies that it doesn't. So, no Houdini for me.

Peter

Another guy that dont care. And until the author fess up no Houdini for Peter. Will the world ever be ok until Peter gets his Houdini

Dr. Ivannik

So what's you're telling me is that it is fine for the author to lie about the origins of his engine and steal code from others? Dude, you're morally messed up.

Peter

Peter

I personally don't know if any author lied about anything or stole anything. These allegations are made by other people like yourself. I don't know where people like you come from??? Accusing people of lying and stealing and then questioning my moral standing in life. In the end I will be judged for my moral character but not by you peter. I am embarrassed for you!! I hope you get your head right.

Thank you

Dr. Ivannik

There is sufficient evidence posted in this thread (disassemble/decompilations, behavior in strange positions, evaluations, PVs, etc) that it's pretty obvious that the author did steal something. Explain how Lance Perkins's disassembly of Houdindi matches almost exactly the disassembly of IvanHoe for example.

So what's you're telling me is that it is fine for the author to lie about the origins of his engine and steal code from others? Dude, you're morally messed up.

Peter

are you talking about houdini or about rybka here ?

I guess it applies to both (Rybka maybe more so, since that was actually a GPL violation whereas IvanHoe is in the public domain).

Peter

[BB+]

Explain how Lance Perkins's disassembly of Houdindi matches almost exactly the disassembly of IvanHoe for example.

I think that actually might be in the weaker part of the evidence given here. As Gerd Isenberg put it on the Chess Programming wiki:

The idea to index the material table in the same manner by combined counters of queens, rooks, light and dark bishops, knights and pawns, and to calculate piece counters from that table-index by a sequence of mod/div operations by {2,2,3,3,2,2,2,2,3,3,9,9} might be considered obvious after studying the mentioned source code, and if applied that scheme, there is hardly anything to avoid a sequence of almost identical x86 machine code with same constants for reciprocal multiplication.

So these could point back to the ideas/code dichotomy in any event. The postings of Jeremy, with identical evals from a variety of positions, seems to me to be more robust as evidence.

[Jeremy Bernstein]

I suppose one can simply say, "who cares?", and call it a day. As I wrote on the Rybka forum - if your tournament permits Ippolit, it will permit Houdini, and if not, it won't. That was true 6 months ago and it's true today and will likely be true in 2 years.

Houdini is getting some press attention now, so it's worthwhile to collect the, hell, I'll say it, the proof that it's a clone of Ippolit in one place and hope that curious people find it useful. I personally find the nearly identical evaluation and PV output conclusive.

I don't know if the original Ippo authors care, Kranium and Sentinel don't seem to mind (or maybe they're happy letting other people bitch on their behalf), and life goes on. On a programming level, Robert has demonstrated that he knows what he's doing, and I'm sure there's less and less Ippolit in Houdini every day.

What was it that Robert said, "the chess world has moved on"? Quite possibly it has, but it never hurts to remind the chess world that it's being lied to.

Jeremy

[Peter C]

Explain how Lance Perkins's disassembly of Houdindi matches almost exactly the disassembly of IvanHoe for example.

I think that actually might be in the weaker part of the evidence given here. As Gerd Isenberg put it on the Chess Programming wiki:

The idea to index the material table in the same manner by combined counters of queens, rooks, light and dark bishops, knights and pawns, and to calculate piece counters from that table-index by a sequence of mod/div operations by {2,2,3,3,2,2,2,2,3,3,9,9} might be considered obvious after studying the mentioned source code, and if applied that scheme, there is hardly anything to avoid a sequence of almost identical x86 machine code with same constants for reciprocal multiplication.

So these could point back to the ideas/code dichotomy in any event. The postings of Jeremy, with identical evals from a variety of positions, seems to me to be more robust as evidence.

Oh. I know very little about chess programming, so 2 nearly identical sequences of assembly looked pretty suspicious to me. Thanks.


Has anyone tried how they behave in illegal positions?
4k3/8/8/3PP3/2PPPP2/1PPPPPP1/PPPPPPPP/3K4 w - - 0 1
Houdini 1.01 (the earliest version I have) and IvanHoe 999963 both crash (albeit at different times; Ivan crashes just after "position fen ..." is sent and Houdini crashes as soon as "go infinite" is sent) while Naum 4.2, Stockfish 2.0.1, and Komodo 1.3 analyze it just fine. Critter 0.90 cleanly rejects the FEN.

Peter