A Question for Our Sponsor..IPPO Links OK or Not??

[Michael Diosi]

Hi,


It is certainly a bigger task than writing a Tic-Tac-Toe engine. Did you manage to get it running in your favorite GUI, Chessbase ?



Michael
http://www.playwitharena.com

[Milos]

It is certainly a bigger task than writing a Tic-Tac-Toe engine. Did you manage to get it running in your favorite GUI, Chessbase ?

Since you do "so much" programming for Arena, it's obvious your programming skills are on Graham's level. So, you talking about programming is like a security guy in CERN parking talking about ATLAS detector.

Btw. since you obviously didn't figure that one yet, there is a button on the right side of each post saying quote. When you are referring to someone you usually press that one .

[bob]

ICD/ Your Move & Chess and company didn't specifically ask links to (IPPOLIT/Robbolito) to be deleted. They DIDN'T name _any_ engine. They just asked that the links to questionable software to be deleted.

Posts from both yourself and Jeremy in the mods forum straight after this show clearly that you two also knew that they were talking specifically about these engines as you clearly stated so.
I will not quote from that forum because it's private, but at least when a new moderation team gets access to it, they'll be able to see everything in there.

Cheers,
Graham.

They're not questionable anymore (more strong evidence from BB reports)

I have asked Jeremy to read this thread and ask him to comment since he's on a holiday.

The BB report is not universally accepted as far as I'm aware. Who is BB?

Someone who is quite capable technically, with respect to computer chess, and someone I have now had several interactions with and who offers far more convincing data than others I can name.

[bob]

The BB report is not universally accepted as far as I'm aware. Who is BB?

They're universally accepted... but then again there are always select few that disagree with the report. Finally we have this report as evidence and the report is enough to prove to many people and atleast 2 moderators that the engine is not in _any_ way questionable.

As for BB, I don't know who he is. I thought he was a good friend of Zach's and had communicated with Larry Kauffman. I have no doubt that he's genuine intelligent programmer.

Why won't any of these people put their names to what they claim?

perhaps because they know that anyone on "that side" of the discussion is subject to massive attacks??? --by a select few???

[bob]

gerold,

If _you_ think the Ippolit is questionable. Why do you still download and test the engine?

Here's your test: http://www.talkchess.com/forum/viewtopi ... ht=#353190

As for me, I don't think it's questionable.

If you haven't read the BB's report and want to read it. Please see this:
http://www.talkchess.com/forum/viewtopic.php?t=34914

The whole thing is much more simple. We can accept the BB report, even though the author decides to stay anonymous. We may also assume that it is quite accurate. You can be sure that Vas has the same opinion. But a lot of people, including many non-programmers and also a few programmers, are simply either not understanding it or misinterpreting its contents. Those who do not understand it might have repeated what others say. It is easy to see a "majority" this way. But I still hope that there is a tiny set of people left who care more about what they see and what they know, instead of what a "majority" says.

The key point is that those people who understand the BB report know that IP* is definitely based to a huge extent on the results of disassembling Rybka 3, with a lot of *changes* that were made. The major part of the R3 reuse obviously occurs in evaluation and search while the origin of other parts like board representation, move generator or UCI interface (to name only a few) is not clearly identified.

As an example I suggest to look at appendix B of the BB report (page 28ff), and do a thorough comparison of the listed search algorithms. There are four algorithms given in detail:

- PV node
- Low-depth (not in check)
- qsearch when not in check
- qsearch when in check

For each of these, copy the IP* and the Rybka section into separate text files, do some careful indenting, formatting and "normalizing" to improve comparability of the two sides, and then use a good text comparison tool (I use BeyondCompare, there are some others).

Look how much is different, and how much is IDENTICAL.

Then come back, please.

Hmm... Heard this argument before. Except replace ip* with fruit.

You may go a similar way for the evaluation part, although there it is slightly harder since the BB report does not contain such detailled listings of eval algorithms like for search. But you can still see a lot of DIFFERENCES and a lot of IDENTICAL things.

It is clear that drawing the line between "taking ideas" and "taking code" is not easy, especially in a case like this where we are considering disassembling. In the given situation I would say that the IP* authors have taken a very huge amount of ideas from R3 at least, so much that it is very obvious where most of the playing strength of IP* (and therefore its descendants) comes from.

Now a lot of different terms are being used here.

"Legal"? Difficult to say. Disassembling to understand what a program does is not illegal by itself. Taking ideas to implement them in an own work is surely legal, too. To a certain extent, at least. But when the amount of ideas taken approaches the whole set of ideas that are found in the original program then I'd raise some doubts, although this is clearly a "grey area" which will be difficult to resolve. Knowing the exact R3 source code version would definitely help a lot here but is not available.

Even if our result were "it's illegal" then the proof of illegality is missing, and Vas will probably not open a court case therefore. He will write up the case, that is for sure, and it will very probably take less than two years until then. This will perhaps convince some more people. Not everyone, of course, which is clear since we don't have a case being as simple as "2 + 2 = 4".

"Moral/legitimate"? One of my main doubts about legality arises from the fact that the IP* source code was published as "public domain", which for me is a copyright violation. Same as above: not easy to prove without R3 source, and "grey area" as far as the "ideas vs. code" line is concerned. But everyone should understand at least that the act of publicly exposing the results of uncovering huge parts of the intellectual property of someone else, being the closed source of the strongest chess engine so far, cannot be called "moral", even if it were "legal".

And that still holds *independent* from the well-known Fruit-R1 case, since it is clear that even a proof of a GPL violation in R1, if it would exist, would *not* turn Rybka (with its current version R4 as the only version where Vas is now saving the source) into an open source program, as long as the copyright owner of Fruit does not make a case. There is *no* double standard here. Publicly exposing something based by a huge amount on closed source of the strongest program so far is a "Robin Hood" style action which should not be regarded as "legitimate" in the CC community in my opinion.

"Clone"? The term is not clearly defined. Many different attempts of a definition are around. I simply avoid it, for me it is enough to know where IP* comes from and has got most of its strength from, I don't need a special word for it.

"Derivative"? Yes. Not in the sense of the GPL, which does not apply here, but in the sense that IP* is clearly (heavily) derived from another engine. But same as "clone": I don't need that word here.

"Original work"? IP* surely contains some of it, but the big picture is that a huge part of IP* is not an original work.

"Questionable"? Of course. Authors of IP* are anonymous, using phantasy names only. That alone, together with their dubious appearance in the ip*.w* forum and their use of language, makes IP* questionable for me. Now many might say, why not being anonymous? The answer is simple: accepting engines with anonymous authors will destroy CC sooner or later, that is for sure. Considering additionally the origins of IP* and the "moral" aspect makes it even more questionable. If we take the word of the CCC charter then we'll find the term "questionable legal status". That is difficult. If some say it's legal and other say it's not, and there is no court to decide, then what *is* the legal status? Don't know. But it surely remains "questionable" when considering mainly the anonymity aspect.

"Do not test it"? Subject of personal opinion as well as of some "general agreement". Personally I would vote for not testing engines with anonymous authors at least, and I would advertise for taking this as a general agreement.

The "closed source program derived from public domain program which is derived from another closed source program" topic (Fire, Houdini, ...) is clearly another - interesting and challenging - issue which I can't cover here.

"Do not link to it"? After the horse has left the barn it's too late to close the door. (I hope I got that right in English, in German the saying goes: "Das Kind ist in den Brunnen gefallen.") We cannot do much (actually: nothing) about hiding the existence of the Ipp* engine family from the world. Links are everywhere. But there is no need for additional advertising. I suggest that everyone stops pointing fingers, saying "they don't allow IP* links => they are censoring" or "they allow IP* links => they are advertising it". We have a variety of different places where CC is discussed, and each has its own rules. None of these places can claim for itself to have the "right" rules while others are wrong. If people do not like the rules of one place they go to somewhere else. This has happened recently, will always happen, and is quite normal. But the "pointing fingers" is not (from both sides).

As a conclusion, I vote for being consistent regarding the handling of links to IP* sites within CCC by *not* allowing them, for being consistent also everywhere else regarding their rules, and to everyone for accepting these attempts to be consistent, as well as any failure of these which are always part of a learning process.


Please agree, and let's have peace

Sven

[zamar]

The key point is that those people who understand the BB report know that IP* is definitely based to a huge extent on the results of disassembling Rybka 3, with a lot of *changes* that were made. The major part of the R3 reuse obviously occurs in evaluation and search while the origin of other parts like board representation, move generator or UCI interface (to name only a few) is not clearly identified.

I think this is excellent summary of the report. There are so many similarities that it can't be just a coincidence. On the other hand almost every snippet of code has been rewritten in a different way or otherwise modified, so it's far from pure copy+paste.

To censor links or not to censor? I have no strong opinions here. Perhaps we could vote about it?

[Steve B]

To censor links or not to censor? I have no strong opinions here. Perhaps we could vote about it?

Actually Joona
during the prior mod term there were several polls run about the Ippo issue
every poll that ran favored allowing discussions and linkage to Ippo and family
here are two that were run last November

http://www.talkchess.com/forum/viewtopic.php?t=30838

http://www.talkchess.com/forum/viewtopic.php?t=30772


this mind you was 6-7 months ago before any of the recent developments regarding Ippo

these and other polls changed hardly anyone's point of view

i think the best way to proceed regarding the Ippo and Family linkage issue is with elections which are 3 weeks away
i can assure you ..teams will be in contention with widely disparate views on this issue..

Best Regards
Steve

[mcostalba]

this mind you was 6-7 months ago before any of the recent developments regarding Ippo

Any poll or vote is a picture, a snapshot, in a precise point in the time.

People change ideas and opinions because new facts are added, old ones lose interest and so on.

We do political elections each 5 years (more or less) just for this reason, because you have the right to change your mind about what you considered your best vote last time you voted.

Here things are moving much faster: ippo sites filled with provocative statements by anonymous people we are quickly being repalced by people with a name and even a picture that takes that "forbid" sources and grows up an engine derived from that.

Is this still "forbidden" ? This is for people to debate, but what is sure is that things are moving and so the people's sentiment, I guess that by the end of this year looking back at threads posted in these days we will have the feeling of reading obsoleted stuff, as when we now read posts about ippo & co written 6 months ago.

[Dann Corbit]

The situation is made even worse by two facts: Ippolit has more differences with R3 than R1 does with Fruit, and it would be much easier to modify Fruit than the result of some decompilation.

But you've forgot the biggest Sven's argument. Fruit is not bitboard and Rybka is.
For Sven rewriting an engine from non-bitboard to bitboard is such an amazigly complicated task that after such a rewrite there is no doubt we got a complately new engine, while writing an engine from scratch following some decompiled code is just a piece of cake...

The man is just hopeless, it's not that he doesn't see that, it's that he doesn't want to see it. What is the reason (protecting a friend, worshiping, being paid) is another topic...

Without seeing what was actually done, I do not think we can say which process was more or would be more difficult.

[Graham Banks]

the fact that virtually every aspect of R1 is directly from Fruit gives a very clear picture.

Funny thing is that Ryan (Fruit) tells me that these claims are grossly exaggerated, but nobody seems prepared to listen to what he has had to say or believe it.