BB's Rybka/Ippolit comparison

General discussion about computer chess...
User avatar
Angel
Posts: 103
Joined: Thu Jun 10, 2010 12:57 am
Contact:

Re: BB's Rybka/Ippolit comparison

Post by Angel » Tue Jun 15, 2010 3:58 pm

Anonimity is not a crime and a very sensible thing in todays cyber space

BBs finding can be checked by independat unbiased or simply honest members of this forum so that should be done and would end this nonsense once and for all so the ippolit engines can be officialy recognised
this is only fair even without BBs findings being confirmed given the evidence that Vas undoubtly plaugarised fruits sources to make his compiles has been clearly recognised as fact
Nothing is ever truly lost , just miss placed and awaiting us

Jeremy Bernstein
Site Admin
Posts: 1226
Joined: Wed Jun 09, 2010 7:49 am
Real Name: Jeremy Bernstein
Location: Berlin, Germany
Contact:

Re: BB's Rybka/Ippolit comparison

Post by Jeremy Bernstein » Tue Jun 15, 2010 4:04 pm

Olivier Deville wrote:
Jeremy Bernstein wrote:Anonymity is just a fact of life on the internet. I agree that, in an ideal world, the author(s) would have faces and (real) names, as it would be reassuring somehow. But I personally don't see how revealing your personal data can be a prerequisite to participation in any net-based activities.

Jeremy
Hi Jeremy

It seems to me all other engines around have known authors, with full names.

Before the troll attacks me, let me state I only want to know the truth about the whole story. When/if Ippolit is proven to be legit (the report is a good start), and the authors drop the mask, I will gladly add it to my tournaments.

Olivier
Of course. My point is that in nearly every other "creative" subculture of the internet, pseudonymous and anonymous production is normal. For whatever reason, this community is very rigid on this point. Musicians, visual artists and programmers within these fields are accustomed to dealing with nameless entities. Possibly CC's roots in academia have led to this, perhaps the competitive rankings.

Here's an interesting story from my personal online history -- this is from a few years back, but I/other interested parties still don't know who this person was, although "she" contributed immensely to the beginnings of real-time (consumer) computer video and was, indirectly, responsible for me getting my job: http://en.wikipedia.org/wiki/Netochka_Nezvanova . "She" was also an abusive nut, but that's one of the common unpleasant side effects of anonymity. The nickname "N.N." makes the reference at least associatively chess-related... :)

Jeremy

User avatar
thorstenczub
Posts: 593
Joined: Wed Jun 09, 2010 12:51 pm
Real Name: Thorsten Czub
Location: United States of Europe, germany, NRW, Lünen
Contact:

Re: BB's Rybka/Ippolit comparison

Post by thorstenczub » Tue Jun 15, 2010 4:25 pm

BTO7 wrote:Funny how the Rybka Lovers are STILL some how some way ...even after a very thorough report STILL are trying to somehow affix guilt to Ippolitt??? They are now trying to jam a square peg in a round hole ...trying to still some how some way BLAME Ippo for cloning R3. I mean when all you have is ...well the author is anonymous now??? You guys get 33 PAGES of PROOF and all you can say is ...but why no names? How about WHY are you guys so determine to put blame on these guys still......everyone of you should be APOLOGIZING! GIVE IT UP your wrong...these guys should have never been persecuted in the first place ....now we have proof of no CLONING yet some still wanna blame these guys....WHY? There should be a public apology by Vas and every one of you that was saying guilty until proven innocent. Now Rybka lovers are just trying to dream up way to blame IPPO's anyways. Seems Rybka guys are not wanting to hear the truth ...they are just hell bent on IPPO's are guilty in spite of the facts. Stop arguing just to argue ...try starting to apologize for knowing nothing with not a shred of evidence ...yet being judge, jury, the whole time on NOTHING.

BT
from my eperience over the many years of computerchess i did,
people are mainly interested in washing dirty laundry.

not in computerchess.

and believers.... do you really think believers in whatever need facts ?
they do what they always do.
they discuss about programmer because they have no instrument to discuss program.
these guys need no facts to stone somebody.

hyatt
Posts: 1242
Joined: Thu Jun 10, 2010 2:13 am
Real Name: Bob Hyatt (Robert M. Hyatt)
Location: University of Alabama at Birmingham
Contact:

Re: BB's Rybka/Ippolit comparison

Post by hyatt » Tue Jun 15, 2010 4:33 pm

benstoker wrote:
Chris Whittington wrote:
Jeremy Bernstein wrote:
govert wrote:
Jeremy Bernstein wrote:
kingliveson wrote:Where do we go from here?
Ideally, we get some independent confirmation from a reliable source that the data presented by BB is legit. I don't mean to be in any way disrespectful when I say that, BB. But given the events of the last months, it seems wise to verify claims, even (especially) those which resonate with one's own gut feeling. I wonder if Zach could be convinced to at least spot-check a few things.

Jeremy
Also, I think we should get to the bottom of this anonymity thingy;

Assuming IPPO & co. are legal, why does the author(s) stay anonymous, and could something be done to get them to reveal their identities?
Anonymity is just a fact of life on the internet. I agree that, in an ideal world, the author(s) would have faces and (real) names, as it would be reassuring somehow. But I personally don't see how revealing your personal data can be a prerequisite to participation in any net-based activities.

Jeremy
well, I don't see a problem with IPPOLIT team being anonymous, I suppose it makes it a bit more difficult for them if they are accused of whatever, but anonymity, in itself, is not a precursor or proof of the misbehaviours they have been accused of.

the anonymity problem does hold for BB (unfortunately, because he may well have good reason). BB is effectively the expert witness in this affair. Expert witnesses need a track record to be credible, without one their evidence can be dismissed in many ways, for example the suggestions that BB is part of IPPOLIT and the work is a fraud (not that I suggest it is, just that it's impossible to assert the evidence in the face of that attack on it).

Whilst BB has been of incalculable help in moving forward to a possible resolution, comp chess is not going to get there without some sort of usual expert witness demonstration of independence and track record.

To put it another way, BB has produced a category A report, us category B guys (who can read and understand, ie, other cc programmers) can interpret what he says and, pretty uniformly, we interpret it as meaning IPPOLIT is clean, and deliver our conclusions to the rest of cc, but what we can't do is assess the overall status of the category A expert witness because of the anonymity.
I would like to see BB followup with a detailed explanation of his methods, with at least some illustrations of what utilities he used to disassemble and how he came to certain conclusions by examination of the assembly. This would allow third parties to easily test things out themselves and confirm assertions or otherwise raise questions. Scientific theories must be based on experimental results that are repeatable using the same conditions and methods. I hope BB helps by showing us exactly how he conducted his examination.

If the examination can be easily repeated, then it is largely irrelevant who BB is, since other experts can easily test the hypotheses and see for themselves.

What he has done is _not_ remarkable. It represents a lot of work. Transplanting a heart and lungs is remarkable and I would have no chance in pulling this off. Building a barbed-wire fence around the state of Texas is a simple task that would take forever. I know how to do that task. This is true of disassembling code. Most CS students (certainly those in my assembly language / architecture course) understand a high-level language like C/C++, and a very low-level language in assembler code. And they know how to translate directly from C to assembler to do the same task. And in learning that, they learn how to go in "the other direction" at the same time.

benstoker
Posts: 110
Joined: Thu Jun 10, 2010 7:32 pm
Real Name: Ben Stoker

Re: BB's Rybka/Ippolit comparison

Post by benstoker » Tue Jun 15, 2010 4:57 pm

hyatt wrote:
benstoker wrote:
Chris Whittington wrote:
Jeremy Bernstein wrote:
govert wrote:
Jeremy Bernstein wrote:
kingliveson wrote:Where do we go from here?
Ideally, we get some independent confirmation from a reliable source that the data presented by BB is legit. I don't mean to be in any way disrespectful when I say that, BB. But given the events of the last months, it seems wise to verify claims, even (especially) those which resonate with one's own gut feeling. I wonder if Zach could be convinced to at least spot-check a few things.

Jeremy
Also, I think we should get to the bottom of this anonymity thingy;

Assuming IPPO & co. are legal, why does the author(s) stay anonymous, and could something be done to get them to reveal their identities?
Anonymity is just a fact of life on the internet. I agree that, in an ideal world, the author(s) would have faces and (real) names, as it would be reassuring somehow. But I personally don't see how revealing your personal data can be a prerequisite to participation in any net-based activities.

Jeremy
well, I don't see a problem with IPPOLIT team being anonymous, I suppose it makes it a bit more difficult for them if they are accused of whatever, but anonymity, in itself, is not a precursor or proof of the misbehaviours they have been accused of.

the anonymity problem does hold for BB (unfortunately, because he may well have good reason). BB is effectively the expert witness in this affair. Expert witnesses need a track record to be credible, without one their evidence can be dismissed in many ways, for example the suggestions that BB is part of IPPOLIT and the work is a fraud (not that I suggest it is, just that it's impossible to assert the evidence in the face of that attack on it).

Whilst BB has been of incalculable help in moving forward to a possible resolution, comp chess is not going to get there without some sort of usual expert witness demonstration of independence and track record.

To put it another way, BB has produced a category A report, us category B guys (who can read and understand, ie, other cc programmers) can interpret what he says and, pretty uniformly, we interpret it as meaning IPPOLIT is clean, and deliver our conclusions to the rest of cc, but what we can't do is assess the overall status of the category A expert witness because of the anonymity.
I would like to see BB followup with a detailed explanation of his methods, with at least some illustrations of what utilities he used to disassemble and how he came to certain conclusions by examination of the assembly. This would allow third parties to easily test things out themselves and confirm assertions or otherwise raise questions. Scientific theories must be based on experimental results that are repeatable using the same conditions and methods. I hope BB helps by showing us exactly how he conducted his examination.

If the examination can be easily repeated, then it is largely irrelevant who BB is, since other experts can easily test the hypotheses and see for themselves.

What he has done is _not_ remarkable. It represents a lot of work. Transplanting a heart and lungs is remarkable and I would have no chance in pulling this off. Building a barbed-wire fence around the state of Texas is a simple task that would take forever. I know how to do that task. This is true of disassembling code. Most CS students (certainly those in my assembly language / architecture course) understand a high-level language like C/C++, and a very low-level language in assembler code. And they know how to translate directly from C to assembler to do the same task. And in learning that, they learn how to go in "the other direction" at the same time.
Does this mean that you are satisfied that the BB report's assertions are adequately explained, shown, demonstrated? BTW, maintaining that barbed wire fence along the Mexico border would be harder than BB's project.

User avatar
kingliveson
Posts: 1388
Joined: Thu Jun 10, 2010 1:22 am
Real Name: Franklin Titus
Location: 28°32'1"N 81°22'33"W

Re: BB's Rybka/Ippolit comparison

Post by kingliveson » Tue Jun 15, 2010 5:30 pm

hyatt wrote: What he has done is _not_ remarkable. It represents a lot of work. Transplanting a heart and lungs is remarkable and I would have no chance in pulling this off. Building a barbed-wire fence around the state of Texas is a simple task that would take forever. I know how to do that task. This is true of disassembling code. Most CS students (certainly those in my assembly language / architecture course) understand a high-level language like C/C++, and a very low-level language in assembler code. And they know how to translate directly from C to assembler to do the same task. And in learning that, they learn how to go in "the other direction" at the same time.
Here at UCF, ASM introduction begins with Embedded Systems using Motorola 6811 micro-controller and Computer Architecture is VHDL. Combine these courses with Digital Systems and C programming, you have decent background. Where I disagree is saying what was done here is not remarkable. Sure with your extensive background in the field it may come easy. Even some very talented programmers have a difficult time doing do the reverse -- that is ASM to C. I think he's shown to have good knowledge in computer chess programming by the detailed report he produced.
PAWN : Knight >> Bishop >> Rook >>Queen

User avatar
Chris Whittington
Posts: 437
Joined: Wed Jun 09, 2010 6:25 pm

Re: BB's Rybka/Ippolit comparison

Post by Chris Whittington » Tue Jun 15, 2010 5:59 pm

kingliveson wrote:
hyatt wrote: What he has done is _not_ remarkable. It represents a lot of work. Transplanting a heart and lungs is remarkable and I would have no chance in pulling this off. Building a barbed-wire fence around the state of Texas is a simple task that would take forever. I know how to do that task. This is true of disassembling code. Most CS students (certainly those in my assembly language / architecture course) understand a high-level language like C/C++, and a very low-level language in assembler code. And they know how to translate directly from C to assembler to do the same task. And in learning that, they learn how to go in "the other direction" at the same time.
Here at UCF, ASM introduction begins with Embedded Systems using Motorola 6811 micro-controller and Computer Architecture is VHDL. Combine these courses with Digital Systems and C programming, you have decent background. Where I disagree is saying what was done here is not remarkable. Sure with your extensive background in the field it may come easy. Even some very talented programmers have a difficult time doing do the reverse -- that is ASM to C. I think he's shown to have good knowledge in computer chess programming by the detailed report he produced.

It's certainly true that the detailed report is a recommendation at to its own status. I'm assuming that the detailed comments next to the asm sections really are meaningful in context, it would not be too good if the comments had been placed next to random bits of assembler, in other words I hope the community is not relying on the comment-asm match by default ;-)

Nevertheless, what we have is:

a) An assertion by Vas to the tune that IPPOLIT is reverse engineered with modifications from Rybka3. The assertion is without evidential backing.

b) A denial by IPPOLIT team. The denial is without evidential backing, but that's hardly surprising, proving a negative is not too easy. As in XYZ denies being a murderer but is unsurprisingly unable to account for his movements 24 hours a day 7 days a week etc. The onus has to be on the accuser to provide the evidence.

c) There is one and only one piece of detailed evidence. That points to a completely clean IPPOLIT. The detailed evidence suffers from anonymity, but it is the ONLY evidence.

d) IPPOLIT remains blackballed on various forums.

Given the current situation as is, and judging, fairly and without bias, it is difficult not to suggest that since the ONLY evidence says IPPOLIT is clean, IPPOLIT should be given immediate status as any other chess engine on forums, with links, in tournaments etc.

Meanwhile, everything, including many reputations, hangs on the status of BB and whether or not that status is good, or, on the repeatability of his report. Do others, prepared to do the work, confirm. Or, do others, with detailed checking of the asm-comments confirm they match.

User avatar
Rebel
Posts: 515
Joined: Wed Jun 09, 2010 7:45 pm
Real Name: Ed Schroder

Re: BB's Rybka/Ippolit comparison

Post by Rebel » Tue Jun 15, 2010 7:52 pm

BB, thanks a lot, I will try gdb and see if things are understandable for my rusty brain.

As far as I see it the acceptance of your document stands or falls with your credibility and skills. In this respect your anonymity (although understandable) works against you. I mean, if the document was signed by a known-name (say for instance Frans Morsch to name a somebody) I would give him a call and if he confirms there is no reason why I shouldn't accept the document as real.

I have a proposal that might help. If you are able to tell me how ProDeo detects a passed pawn that would be proof enough for me you are able to do a RE job, thus your skills. After all your document is all about RE.

BTW, how many months of work did it take? Just curious........

Ed
BB+ wrote:Take the executable. Run it. Run "gdb --pid X" where X is the process number. Then "disassemble 0x400000 0x500000" (more than enough, usually). Then grep for "call" on the dump of this, to get an idea of where functions start. Do "go infinite", stop a random point, and note where the instruction pointer is. More forwards/backwards from the current subroutine.

After this, it is can be guesswork, or trial/error, or comparison with IPPOLIT and friends. For instance, if a function calls no other functions, there are a limited number of possibilities for it. If it is also doing something with a hash, perhaps it is the pawn-eval function. If a function starts by anding with 0x3f and shifting by 6, it is probably doing something with a move (make/undo or move-is-ok, for instance). If it is passed a stack that gets moves (again noted by their structure with 0x3f and shifts by 6) put into it, it is probably move generation. For search functions, you can note that there is access to the hash table at the start, and they call lots of other functions. I think the castling code given is one of the more easy things to understand. In my daily job, I often have to track/fix bugs of other people (and sometimes myself), so this helps me be more facile about asm code and gdb. Maybe I should also stress that it is a lot of work, with guessing (sometimes wrong, to be corrected by later info) and trial/error prevalent.

Here is the relevant parts of a Rybka CUT/ALL node function (and also the "exclusion" search); I elide various instructions that are irrelevant to the current discussion (and I don't always have this many comments when disassembling)..

hyatt
Posts: 1242
Joined: Thu Jun 10, 2010 2:13 am
Real Name: Bob Hyatt (Robert M. Hyatt)
Location: University of Alabama at Birmingham
Contact:

Re: BB's Rybka/Ippolit comparison

Post by hyatt » Tue Jun 15, 2010 9:57 pm

benstoker wrote:
hyatt wrote:
benstoker wrote:
Chris Whittington wrote:
Jeremy Bernstein wrote:
govert wrote:
Jeremy Bernstein wrote:
kingliveson wrote:Where do we go from here?
Ideally, we get some independent confirmation from a reliable source that the data presented by BB is legit. I don't mean to be in any way disrespectful when I say that, BB. But given the events of the last months, it seems wise to verify claims, even (especially) those which resonate with one's own gut feeling. I wonder if Zach could be convinced to at least spot-check a few things.

Jeremy
Also, I think we should get to the bottom of this anonymity thingy;

Assuming IPPO & co. are legal, why does the author(s) stay anonymous, and could something be done to get them to reveal their identities?
Anonymity is just a fact of life on the internet. I agree that, in an ideal world, the author(s) would have faces and (real) names, as it would be reassuring somehow. But I personally don't see how revealing your personal data can be a prerequisite to participation in any net-based activities.

Jeremy
well, I don't see a problem with IPPOLIT team being anonymous, I suppose it makes it a bit more difficult for them if they are accused of whatever, but anonymity, in itself, is not a precursor or proof of the misbehaviours they have been accused of.

the anonymity problem does hold for BB (unfortunately, because he may well have good reason). BB is effectively the expert witness in this affair. Expert witnesses need a track record to be credible, without one their evidence can be dismissed in many ways, for example the suggestions that BB is part of IPPOLIT and the work is a fraud (not that I suggest it is, just that it's impossible to assert the evidence in the face of that attack on it).

Whilst BB has been of incalculable help in moving forward to a possible resolution, comp chess is not going to get there without some sort of usual expert witness demonstration of independence and track record.

To put it another way, BB has produced a category A report, us category B guys (who can read and understand, ie, other cc programmers) can interpret what he says and, pretty uniformly, we interpret it as meaning IPPOLIT is clean, and deliver our conclusions to the rest of cc, but what we can't do is assess the overall status of the category A expert witness because of the anonymity.
I would like to see BB followup with a detailed explanation of his methods, with at least some illustrations of what utilities he used to disassemble and how he came to certain conclusions by examination of the assembly. This would allow third parties to easily test things out themselves and confirm assertions or otherwise raise questions. Scientific theories must be based on experimental results that are repeatable using the same conditions and methods. I hope BB helps by showing us exactly how he conducted his examination.

If the examination can be easily repeated, then it is largely irrelevant who BB is, since other experts can easily test the hypotheses and see for themselves.

What he has done is _not_ remarkable. It represents a lot of work. Transplanting a heart and lungs is remarkable and I would have no chance in pulling this off. Building a barbed-wire fence around the state of Texas is a simple task that would take forever. I know how to do that task. This is true of disassembling code. Most CS students (certainly those in my assembly language / architecture course) understand a high-level language like C/C++, and a very low-level language in assembler code. And they know how to translate directly from C to assembler to do the same task. And in learning that, they learn how to go in "the other direction" at the same time.
Does this mean that you are satisfied that the BB report's assertions are adequately explained, shown, demonstrated? BTW, maintaining that barbed wire fence along the Mexico border would be harder than BB's project.
It means that the "technique used" doesn't need dozens of posts explaining how it was done. Whether it is factual or not I could not say with 100% certainty. It seems sound, well thought out, and detailed. I tend to take things at face value unless there is something that stands out like a red flag. I do not see one, but that doesn't mean there is not one. In light of absolutely no data to contradict any of this, it certainly looks like "the real deal" imho.

hyatt
Posts: 1242
Joined: Thu Jun 10, 2010 2:13 am
Real Name: Bob Hyatt (Robert M. Hyatt)
Location: University of Alabama at Birmingham
Contact:

Re: BB's Rybka/Ippolit comparison

Post by hyatt » Tue Jun 15, 2010 10:01 pm

kingliveson wrote:
hyatt wrote: What he has done is _not_ remarkable. It represents a lot of work. Transplanting a heart and lungs is remarkable and I would have no chance in pulling this off. Building a barbed-wire fence around the state of Texas is a simple task that would take forever. I know how to do that task. This is true of disassembling code. Most CS students (certainly those in my assembly language / architecture course) understand a high-level language like C/C++, and a very low-level language in assembler code. And they know how to translate directly from C to assembler to do the same task. And in learning that, they learn how to go in "the other direction" at the same time.
Here at UCF, ASM introduction begins with Embedded Systems using Motorola 6811 micro-controller and Computer Architecture is VHDL. Combine these courses with Digital Systems and C programming, you have decent background. Where I disagree is saying what was done here is not remarkable. Sure with your extensive background in the field it may come easy. Even some very talented programmers have a difficult time doing do the reverse -- that is ASM to C. I think he's shown to have good knowledge in computer chess programming by the detailed report he produced.

I'm using "not remarkable" in the context that the _methodology used_ is not esoteric or something that might be done in a government lab. It is a straightforward technique that most CS graduates could do, given the interest. My "not remarkable" had nothing to do with his computer chess knowledge. It was simply a response to someone asking him to explain how he did the RE analysis. That is what is "not remarkable" but is "time-consuming". BTW our asm course is based on X86, which is perfectly suited for this particular example, obviously.

Post Reply