Thoughts on Fruit=Rybka EVAL

[BB+]

(the posts would give a timestamp of when they were edited if so).

I don't think is true. I edit my posts all the time, and get no such message. However, if something else appears in the thread in the interim (between Submit and Edit), only then will such an "edited 1 times by BB+ ..." message show up. At least, that's my understanding of how the forum works.

I'm confused about the rest of the discussion, as: I don't remember mentioning "notes" one way or the other; and as Zach says, my correction was a minor one in a completely different post, and the original 2 posts with the enumeration were left intact (and have been there all along, possibly with the word "notes" in the first 10-15 minutes [the editing period], which would appear to be ~4am in the Netherlands). I also didn't respond sufficiently promptly to Ed's first inquiry about this (I have no Internet at home, and was not at work on Sunday [the first post was Sat 7:37pm my time, and the second Mon 9:54am), and so he asked a second time [with slightly more innuendo, such as: A remarkable coincidence. Care to explain?].

Or are we now in the state of mistrusting each others motives?

At least with ChrisW and his confrontational mode, I think this state occurred awhile back (indeed, probably dating from when Hyatt was named to the Secretariat, or maybe when Levy was last elected -- which reminds me that nominations should occur "soon" [see here] as the Triennial Meeting is just today 3 months away).

[Rebel]

This is why I'm even bringing it up. BB did nothing wrong, but you were quick to suggest that he was deleting or modifying evidence. Why do you mistrust him? Sorry to belabor this point...

One more note on this, if I ever had plans to discredit BB or you for that matter for agenda reasons then:

In retrospect, I regret only publishing evidence relating to the evaluation, since it's fuzzy in many ways, and people can easily see it as "circumstantial". A lot of the evidence hasn't even been written up properly. I view this as my own failure in many ways, but there's only so much effort I'm willing to put into this.

this confession of yours would have been a great starting point.

Instead I gave you a compliment for your honesty.

Zach, your credibility is not at stake, not in my book. What people need to understand is that your and also BB's documents are written for chess programmers in the first place. Much of the trouble comes by Bob's dishonesty exaggerating the facts such as the case SR addressed. Flat out misleading the people here or incompetence to understand the document. You should not be blamed for that.

Also note that when I question someones integrity (Bob) I am used to do in the open and not in a sneaky way.

[Rebel]

Further, I think all 6 of your points are debatable, if not dubious. I will just choose one (a), and refer you to this, which is where I think the PST issue currently stands [I can make a new thread, if you prefer]. I might also note that I estimated that it was less than 1% of the issue, so ignoring it doesn't change much.

In the documents the PST's are promoted as a major issue.

[Rebel]

So, can we leave out the disparaging comments or not ?

No worries. As long as we don't return to flimsy theories that "all top engines in 2005 used essentially the same evaluation code", or that 5.5 months is an unthinkably short time to rewrite code, or that the "evaluation ordering" could be the alpha-and-omega of the R/F case, I will cease any mocking. Any irritation, however, also derived from how you went about some things (as opposed to the colloquy in the Panel), such as the volcanic reaction to Wylie's "possible", pithy dismissals (directed at me) like "I am so sorry for all your work", the continued mis-impression that the Panel "had no dissenting voices", ...

Which means a "no" answer. It's notified.

Lemme pick just one to comment, because in the end that's the one what this is all about: "I am so sorry for all your work"

Yes, I feel that way for you after you told me the time and energy you put in to this.

As it seems to me you had too little opposition in the panel and also overlooked a couple aspects of chess programming. The PST's comes to mind, but also EVAL. And we haven't even really started a thorough discussion about EVAL as we are still discussing the PST. All in all, the documents are inconclusive.

And you (and Zach as well) should REALLY explain to me that when you state that the EVAL order in Rybka is basically the same as in Fruit that is downplayed (by Zach) as irrelevant. I give you guys a smoking gun and you call it irrelevant? The logic fully escapes me.

But maybe that's another flimsy question of mine. Solly, could not resist

[Rebel]

I do not dislike Ed, just the opposite, but in this case he is simply wrong and for whatever reason feels compelled to defend Vas and rescind all he said. I am disappointed that Ed has abandoned all rational . There are others that have less noble reasons. One, that is anything but anonymous.

It's okay Terry, you are a good man.

[hyatt]

This is why I'm even bringing it up. BB did nothing wrong, but you were quick to suggest that he was deleting or modifying evidence. Why do you mistrust him? Sorry to belabor this point...

One more note on this, if I ever had plans to discredit BB or you for that matter for agenda reasons then:

In retrospect, I regret only publishing evidence relating to the evaluation, since it's fuzzy in many ways, and people can easily see it as "circumstantial". A lot of the evidence hasn't even been written up properly. I view this as my own failure in many ways, but there's only so much effort I'm willing to put into this.

this confession of yours would have been a great starting point.

Instead I gave you a compliment for your honesty.

Zach, your credibility is not at stake, not in my book. What people need to understand is that your and also BB's documents are written for chess programmers in the first place. Much of the trouble comes by Bob's dishonesty exaggerating the facts such as the case SR addressed. Flat out misleading the people here or incompetence to understand the document. You should not be blamed for that.

Also note that when I question someones integrity (Bob) I am used to do in the open and not in a sneaky way.

"Then when are you going to remove the Rybka ideas you have added to YOUR code?"

"very direct and open, you say?"



I challenged you to find them. You cited two specific versions and the improved branching factor in the later one. I diff'ed the two search sources and showed the very small changes. And how nothing NEW was added, but that existing code was just tuned a small bit. Never saw any retraction in your "open way of doing things..."

So please, give that a rest, and stop pretending to be "Mr. Altruism". The arguments have been heated at times between many of us. Don't try to act like you politely side-stepped those and didn't jump in. You've made just as many exaggerated statements as anybody else in this discussion/process...

[zwegner]

This is why I'm even bringing it up. BB did nothing wrong, but you were quick to suggest that he was deleting or modifying evidence. Why do you mistrust him? Sorry to belabor this point...

One more note on this, if I ever had plans to discredit BB or you for that matter for agenda reasons then:

In retrospect, I regret only publishing evidence relating to the evaluation, since it's fuzzy in many ways, and people can easily see it as "circumstantial". A lot of the evidence hasn't even been written up properly. I view this as my own failure in many ways, but there's only so much effort I'm willing to put into this.

this confession of yours would have been a great starting point.

Instead I gave you a compliment for your honesty.

Zach, your credibility is not at stake, not in my book. What people need to understand is that your and also BB's documents are written for chess programmers in the first place. Much of the trouble comes by Bob's dishonesty exaggerating the facts such as the case SR addressed. Flat out misleading the people here or incompetence to understand the document. You should not be blamed for that.

Also note that when I question someones integrity (Bob) I am used to do in the open and not in a sneaky way.

I appreciate it.

I'm confused, though, how you would use that statement of mine to discredit me or Mark (what do you mean by "agenda reasons"?). Perhaps I wasn't very clear. I think that there is other convincing evidence that either hasn't been written up at all, or has just been mentioned in forum postings. Even other pieces of evidence written up by Mark or the ICGA panel have been mostly ignored. I still think the evaluation evidence is quite damning, but as we see, there are many shades of grey here, and reasonable people can come up with different interpretations (even if I think they are wrong). Ultimately, what I mean is that, if there are people who are capable of understanding the evidence, but still disagree with me, it is partially my own fault for not making a convincing enough case. I really don't see how my saying that does anything to "discredit" me.

Looks like I will have to spend a lot more effort on this after all.

[hyatt]

Zach, as you get older, you will begin to realize what is going on here.

Most of us are using a Merriam-Webster dictionary. And when we write using words in that dictionary, most get the correct interpretation and all is well. But some are using the "rybkarian-controversy dictionary" where some words mean different things. Any tiny qualification to a statement (I regret only publishing...) gets tristed.

First, "only" doesn't exist in the Rybkarian dictionary, so your statement becomes "I regret publishing ..." instead. So now you are sorry you published the data, which through a small stretch of semantics becomes "I no longer believe what I wrote so it should be discounted completely."

Second, you really blew it with "A lot of the evidence hasn't even been written up properly". That turns into "A lot of the evidence is improper and worthless."

You have to be VERY careful how you phrase your comments, because only a select few have a copy of the rybkarian dictionary, which makes it quite difficult to predict what your statements will mean after going thru the Rybkarian translator (which is way worse than Babblefish, by the way)..

But, on a serious note, you could spend the rest of your life on this, and RE every byte of Rybka and show that it matches Fruit in many other places as well. It won't be enough... However much data you produce, they always want N more pieces. You can NEVER catch up to that threshold condition, since it is what Ken Thompson might call a "self-referential definition" (for those that know what a self-referential structure is). Or from the movie, "The Money Pit." Always 2 more weeks.

[Rebel]

I appreciate it.

I'm confused, though, how you would use that statement of mine to discredit me or Mark (what do you mean by "agenda reasons"?).

You don't wanna know my dark side, my head sometimes is a scary place to be

Perhaps I wasn't very clear. I think that there is other convincing evidence that either hasn't been written up at all, or has just been mentioned in forum postings. Even other pieces of evidence written up by Mark or the ICGA panel have been mostly ignored. I still think the evaluation evidence is quite damning, but as we see, there are many shades of grey here, and reasonable people can come up with different interpretations (even if I think they are wrong). Ultimately, what I mean is that, if there are people who are capable of understanding the evidence, but still disagree with me, it is partially my own fault for not making a convincing enough case. I really don't see how my saying that does anything to "discredit" me.

Looks like I will have to spend a lot more effort on this after all.

I am not sure about the latter. Basically what bothers me on EVAL (besides that order thing) that Fruit has nothing in EVAL that makes it unique, just the usual stuff. A couple of years ago I went forwards (and not even backwards) through EVAL and was disappointed to find only 1 minor idea. Really, that Rybka inspects the squares around the enemy king, has penalties for bad pawns, code for trapped knights/bishops etc. etc. I ask, what is new? It's just the basic stuff. Instead it's Rybka that has something unique, that big table. It would have been a lot easier Fruit had such a specific.

In Rebel I have several unique things in eval which when found in other programs prior to 2005 would raise an eyebrow. Fruit has not.

Well, there is one after all, both have in common maintaining a midgame and endgame score at the same time. It's an elegant optimisation looking at the ASM output, no single branch (jump) to be taken. On the other hand one can argue if the gain of 2 branches really outperforms the extra IMUL on modern processors. That's a question for Gerd I think.

What's faster?

midgame=midgame+(x*11); // fruit way
endgame=endgame+(x*14);

versus,

if (phase) midgame=midgame+(x*11); else endgame=endgame+(x*14); // classic way

But then the real optimisation would be:

int score[2];
int factor[] = { 11,14 };

score[phase]=score[phase]+(x*factor[phase]);

[Rebel]

whereas score[0] represents midgame and score[1] the endgame variable.

No branches, 1 IMUL.

=================

Note to Jeremy, the time limit to edit posts is really small.