Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Bob - Taking notes CAN capture exactly what the code does.
- - By Rebel (****) Date 2013-11-03 09:47
Your statements again.

Bob - According to ME, taking a source program, making detailed notes about everything that is done, and then using that to "write" your own code is NOT ORIGINAL.

Bob - Taking notes CAN capture exactly what the code does. That IS "copying code".

Then we read.

Bob - I've tested everything in Glaurung 2, fruit 2, toga 2 [regarding LMR], and a lot of other ideas as well

You copied Pradu Magic bit-board code.

You were influenced (and I am mild here) by Strelka's (Vas code) Super aggressive razoring.

You took from Stockfish according to Marco.

Adding from Vas 400 elo legacy to Crafty.

Bob - As I have previously mentioned, the _only_ "Rybka idea" I have tested was the tt-singular extension.

Do I need to go on?

One more... Crafty 22.2
1. You copied the Fruit Bishop PST.
2. You took Fruit's interpolation code.

Your (cited) statements are contradicting your actions.

You are doing exactly the same as Vas and everybody else.

-----------

Furthermore you support reverse engineering.

Bob - Taking notes CAN capture exactly what the code does. That IS "copying code".

What else is reverse engineering (RE) other than figuring out what the code does?

Reverse engineering is illegal in many countries and yet you support it. It's why you are not allowed to subscribe the the programmer code of honor, shame on you. 31 chess programmers have. And you, Mr. Ethical and scalp hunter of the cloners Bob Hyatt is not on that list. That's odd, you should top that list as your statements about making notes are far more strict and radical than the programmer code.

I leave it up to you to to create something meaningful and consistent about the mess you created.

Suggest clear rules what is allowed, and what is not.
Parent - - By kunah (**) [cz] Date 2013-11-03 11:22
You took from Stockfish according to Marco.

Using futility margins in a table instead of a couple of constants is so straightforward that I wouldn't even call it an idea.
And I would even bet it was used by engines before Stockfish.
This is worth 0.00... elo (it just a microoptimization that simplifies code a saves a couple of instructions/branches per node).
Parent - By bob (Gold) [us] Date 2013-11-03 16:10
This was exactly what it was.  Originally there were separate tests for depth == 1, depth == 2.  I simplified the code, and by pure accident made it prune for depth==3 also.  I "fixed" the bug and Elo dropped slightly, which is where I discovered that pruning at depth==3 also worked.  When I noticed that, I tried 4, 5 and 6.  4 was a bit better, 5 and 6 never worked for me.

Ed knows all of this, it was discussed on open-chess a few months back.  He is just doing his usual dishonest style of argument here.  Of course HE did copy robe's (or strelka's) pruning because he stated this.  But you won't find that mentioned in his arguments here or in his "code of honor" nonsense.  Perhaps code of dishonor would fit better.
Parent - - By bob (Gold) [us] Date 2013-11-03 16:06
One at a time. 

1.  I didn't write the initial futility code for Crafty.  I "believe" it was written by Jeremiah Pennery but I am not certain.  It was in Crafty LONG before there was a stockfish.  Heinz's book explains the idea, gives pseudo-code, discusses bounds (which did not work well for me but cluster tuning fixed that.)  Ergo this is a bogus statement.  In fact, stockfish uses MY history counter algorithm, which I do not consider a big deal.  At one point you claimed my pruning came from robolito, remember that?  I showed you code that predated robolito, fruit, etc.  So you might as well get off of that bandwagon.  Nothing copied / borrowed from stockfish whatsoever.

2.  Read the quote from me in your post.  "making detailed notes about everything that is done, and then using that to "write" your own code is NOT ORIGINAL".  I didn't make detailed notes about everything that was done in Fruit, Glaurung or Toga.  I ONLY looked at LMR history counters.  NONE of those ideas ended up in Crafty.  So what, exactly, is your point, other than being dishonest here?

3.  I've not looked at Strelka's "super-aggressive razoring."   The razoring code in Crafty was present before there was a strelka.  Sorry.  Yet another outright lie, which you are becoming famous for.  You imagine something and then state it as fact.  I DID look at Heinz's book, which covered razoring.  There's no code to copy, however, just an explanation of the idea.

Keep trying, your FQ is steadily rising (FQ = foolishness quotient).

Feel free to show me one line of "fruit's interpolation code".  I took the interpolation "idea" and implemented it in Crafty.  Using someone's idea is not a violation of rule 2, just copying code or a complete engine design meets that standard, sorry again.

For your "code of honor", what's the point?  signing a list does absolutely nothing.  Your "code of honor" shows it to be meaningless, since you like to fabricate, distort, take out of context, and other such bogus forms of argument.  Where's the "honor" eh?
Parent - - By Venator (Silver) [nl] Date 2013-11-08 15:21
I took the interpolation "idea" and implemented it in Crafty.

You took notes, captured the way the code worked and therefore you have been copying code. Beaten by your own silly words and ridiculous argumentation.

By the way, you still didn't tell us the truth about the Fruit bishop PST you directly copied.
Parent - - By bob (Gold) [us] Date 2013-11-08 16:34
1.  I used the "idea" of interpolation.  Implemented in a different way from Fruit, BTW, but the idea of opening/eg interpolation is there.  But no copied code.

2.  no copied bishop PST, regardless of your incessant idiotic ranting...

grow up.

If that is possible.
Parent - - By Venator (Silver) [nl] Date 2013-11-08 17:04
You have difficulty reading, haven't you? Here is the sentence again:

You took notes, captured the way the code worked and therefore you have been copying code.

I know, it is hard to defend against your own silly arguments.

OK, correction: you copied the bishop PST and added 8 to each value to cover it up....
Parent - - By bob (Gold) [us] Date 2013-11-08 19:35
Didn't take any notes.  Didn't look at fruit code to do interpolation.  It was discussed AT LENGTH on CCC years ago and several of us tried the idea.  Which is NOT "copying code".  You are letting your ignorance get in the way of your education.  Using ideas discovered by others is NOT the same as copying code. 

For example, fruit's "interpolation code" looks like this:

eval = ((opening * (256 - phase)) + (endgame * phase)) / 256;

In the above, phase is computed as

phase = 16*pawn_phase + 4*knight_phase + 4*bishop_phase + 4*rook_phase + 2*queen_phase
           = 16*1 + 4*1 +4*1 + 4*2 +2*4
           = 40

Then

phase = phase - total_pawns*1 - total_knights*1 - total_bishops*1 - total_rooks*2 - total_queens*4

And finally

phase = (phase * 256 + 20) / 40

Crafty is a tad simpler:

  phase = Min(62, TotalPieces(white, occupied) + TotalPieces(black, occupied));

Which produces a number between 0 and 62, using normal material values of P=1, N=B=3, R=5 and Q=9 (8 + 6 + 6 + 10 + 9 for one side, 39, but pawns don't count, so it is maxed at 31 for one side or 62 for both.

  score = ((tree->score_mg * phase) + (tree->score_eg * (62 - phase))) / 62;

To compare, assume opening score = 0, endgame = 100

Here are some interpolated scores for fruit and crafty.  I used an op score of 0, and an eg score of 100.

First pair has all pieces still on board, fruit's interpolated score followed by crafty:

fruit = 0, crafty = 0

next, remove just queens.

fruit = 19, crafty = 29

Next, remove all rooks and queens.

fruit = 39, crafty = 61

next, all minors gone, rooks and queens still on:

fruit = 0, crafty = 38

Next, nothing but pawns left:

fruit = 100, crafty = 100

So, YES, crafty uses the "idea" of linearly interpolated scores, with an opening score that ignores endgame concepts, and an endgame score that ignores opening concepts.  They interpolate completely differently.  They only produce the same scores in exactly two cases, all pawns and pieces present, or just all pawns present, no pieces.  ALL other cases are interpolated differently.

Do you realize just how idiotic you look?  And Ed too, of course.  There is a huge difference between seeing the idea of computing two scores and then interpolating between them, versus copying Fruit's interpolation code exactly.  My code looks nothing like fruit's.  The results look nothing like fruit's.  You look exactly like an idiot.  Which is NOT a new thing.  You should stick to things you understand and stay away from programming.  Your lack of understanding shines about as brightly as the light on top of the Luxor in Las Vegas.  Everyone sees it.

Now compare that to the hashing code Vas copied DIRECTLY from fruit, line for line.  There is a huge difference between the idea of storing two bounds, and copying the exact code Fruit uses to do so.
Parent - - By Venator (Silver) [nl] Date 2013-11-08 20:01 Edited 2013-11-08 20:04
It is all very clear to me: you took notes and discussed them - and the code - at CCC. Which, in your own words, is tantamount to 'copying code'.

That you now fail to refute your own sillyness is no surprise, as you have been scoring own goals on many occasions.

The only piece of code I have seen copied directly from Fruit, is your famous bishop PST. Even amoeba's on Mars don't back you up on this one anymore.
Parent - - By bob (Gold) [us] Date 2013-11-08 20:38
Nope, using a single idea, implemented in a completely different way is NOT copying code.  Taking an evaluation, making detailed notes about EVERY term used, and then using that to write an equivalent evaluation si copying code.  One day you will get it, maybe.

The only one scoring "own goals" is you.  You look like a complete idiot.  Unfortunately, you appear to be used to it and don't realize a change is in order...
Parent - By Venator (Silver) [nl] Date 2013-11-08 21:43
Because you are stuck on stupid and unable to refute your own silly statements, I'll say it to you one more time:

The only piece of code I have seen copied directly from Fruit, is your famous bishop PST. Even amoeba's on Mars don't back you up on this one anymore.
Up Topic Rybka Support & Discussion / Rybka Discussion / Bob - Taking notes CAN capture exactly what the code does.

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill