The Strelka affair to date

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
fern
Posts: 8755
Joined: Sun Feb 26, 2006 3:07 pm

The Strelka affair to date

Post by fern » Sat May 12, 2007 11:38 pm

After the resolution of experts that took a look inside the bowels of Trelka, it appears that it is not a clone of Rybka.
Still, some people is dubious.
It seems that a law has been proclamed that any time two players play the same in the same position, that does not means both reached the positional truth of it, but someone is cloning the other.
It is curious to think like that, as much chess, differently of sheer artistic endeavour, has a layer of objetive truth. In many positions, like in mate-in-two problems solved equally by different players, human or artificial, it is only natural to play the same move.
In fact, it is NECCESARY that players of any kind should play the same if they are good enough to get the esentials of a position. Even different players of different levels can play the same.
So, if we see two engines playing the samemove in some positions, is it an enough good evidence that something wrong is going on?
Besides that, there is another consideration: commons sense indicates that you must take what is said to you in its value UNLESS there is HARD evidence not to believe it.
We can be deceived, but that matter less than to become a suspicious kind of guys. Any community is grounded in mutual trust and this one is not an exception.
Last: as much sooner or later truth comes to the fore, why to poison the ambience prematurely?
Not to believe what it is said to us was already made, at first, with Vas; now it was commited with this new russian guy.
How many extra times it will be done?

My best
Fernando

Christopher Conkie
Posts: 6068
Joined: Sat Apr 01, 2006 7:34 pm
Location: Scotland
Contact:

Re: The Strelka affair to date

Post by Christopher Conkie » Sun May 13, 2007 2:18 am

fern wrote:It seems that a law has been proclamed that any time two players play the same in the same position, that does not means both reached the positional truth of it, but someone is cloning the other.
No, you have misunderstood the positions that were shown to you and their usage.
fern wrote:It is curious to think like that, as much chess, differently of sheer artistic endeavour, has a layer of objetive truth. In many positions, like in mate-in-two problems solved equally by different players, human or artificial, it is only natural to play the same move.
No, certain positions show certain techniques implemented in engines. Some of those techniques are known and some are not as in the case of both Rybka and Strelka. If you had looked (i hope you did) at some of the positions provided by Thomas you might have seen that. It does however depend on whether or not you know what to look for. My last question will explain this (or not as the case may be, it depends on whether you know a little about a certain technique).
fern wrote:In fact, it is NECCESARY that players of any kind should play the same if they are good enough to get the esentials of a position. Even different players of different levels can play the same.
Yes, but that is more by luck than design. Still you fail to understand that engines behave certain ways when they have certain things implemented in them.
fern wrote:So, if we see two engines playing the samemove in some positions, is it an enough good evidence that something wrong is going on?
It depends upon the position. Some positions prove exactly something with no doubts.
fern wrote:Besides that, there is another consideration: commons sense indicates that you must take what is said to you in its value UNLESS there is HARD evidence not to believe it.
We can be deceived, but that matter less than to become a suspicious kind of guys. Any community is grounded in mutual trust and this one is not an exception.


Of the 15 or so new engines this year 7 of them were clones.

Trust? We check them all. No one is singled out in particular unless there is concern. Would you rather we did not?
fern wrote:Last: as much sooner or later truth comes to the fore, why to poison the ambience prematurely?
Not to believe what it is said to us was already made, at first, with Vas; now it was commited with this new russian guy.
How many extra times it will be done?

My best
Fernando
As many times as any new engine is suspect, now and forever.

Now may I ask you something that is quite on topic as I said I would above.

Do you know what nullmove is?

Any programmer will know that there is something strangely similar about Rybka and Strelka based on their behaviour concerning that technique and its (non)implementation.

I bet this is as clear as mud to you. If so don't worry about it. Leave it to the experts is what I say.

Regards

Christopher

User avatar
GenoM
Posts: 910
Joined: Wed Mar 08, 2006 8:46 pm
Location: Plovdiv, Bulgaria
Contact:

Re: The Strelka affair to date

Post by GenoM » Sun May 13, 2007 10:45 am

hi Christopher

its probably very exciting to be part of 'clonebusters'
no doubt it's a useful for CompChess Community job

but new authors of engine has not be treated as they were cloners a priori
firstly is good to prove that their programme is a clone

regards,
Geno
take it easy :)

Christopher Conkie
Posts: 6068
Joined: Sat Apr 01, 2006 7:34 pm
Location: Scotland
Contact:

Re: The Strelka affair to date

Post by Christopher Conkie » Sun May 13, 2007 11:44 am

GenoM wrote:hi Christopher

its probably very exciting to be part of 'clonebusters'
no doubt it's a useful for CompChess Community job
I would not describe it as exciting. It's quite time consuming and although it may be useful I am sure you know that if it were not needed, my time could be much better utilised.
GenoM wrote:but new authors of engine has not be treated as they were cloners a priori
firstly is good to prove that their programme is a clone

regards,
Geno
They are all tested with an open mind. If suspicious they are examined in more detail.

I will say it again even now after what has happened. Listen.

There is something quite unique that is only found in Streka and Rybka. It is exactly the same behaviour and is not what one would expect at all.

Regards

Christopher

User avatar
fern
Posts: 8755
Joined: Sun Feb 26, 2006 3:07 pm

Re: The Strelka affair to date

Post by fern » Sun May 13, 2007 5:25 pm

Do not be so fast in pressuming I do not know nothing about chess programming. Certainly I am not programmer, but I do know what null move is. It is a technique implemented by many programs since lot of time ago. Fritz was the first to make use of it in strenght.
Now, tell me how null move probe -or probed- something AFTER the examination of the source code by an expert probed It was NOT a clone of Rybka. I mean, anything you believed on the ground of this or that feature it is no longer valid and so it becomes moot by now.
Are we going to discus how null move could have probed something If it had probed something?

My best
Fernando

Christopher Conkie
Posts: 6068
Joined: Sat Apr 01, 2006 7:34 pm
Location: Scotland
Contact:

Re: The Strelka affair to date

Post by Christopher Conkie » Mon May 14, 2007 3:06 am

fern wrote:Do not be so fast in pressuming I do not know nothing about chess programming. Certainly I am not programmer, but I do know what null move is. It is a technique implemented by many programs since lot of time ago. Fritz was the first to make use of it in strenght.
Now, tell me how null move probe -or probed- something AFTER the examination of the source code by an expert probed It was NOT a clone of Rybka. I mean, anything you believed on the ground of this or that feature it is no longer valid and so it becomes moot by now.
Are we going to discus how null move could have probed something If it had probed something?

My best
Fernando
I am not going to even attempt to decypher this Fernando.

Let's just sat Strelka and Rybka are the similar in some ways and that in turn may make them play the same way.

I can't put it any easier.

Regards

Christopher

Dann Corbit
Posts: 10203
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: The Strelka affair to date

Post by Dann Corbit » Tue May 15, 2007 3:41 am

fern wrote:After the resolution of experts that took a look inside the bowels of Trelka, it appears that it is not a clone of Rybka.
Still, some people is dubious.
It seems that a law has been proclamed that any time two players play the same in the same position, that does not means both reached the positional truth of it, but someone is cloning the other.
It is curious to think like that, as much chess, differently of sheer artistic endeavour, has a layer of objetive truth. In many positions, like in mate-in-two problems solved equally by different players, human or artificial, it is only natural to play the same move.
In fact, it is NECCESARY that players of any kind should play the same if they are good enough to get the esentials of a position. Even different players of different levels can play the same.
So, if we see two engines playing the samemove in some positions, is it an enough good evidence that something wrong is going on?
Besides that, there is another consideration: commons sense indicates that you must take what is said to you in its value UNLESS there is HARD evidence not to believe it.
We can be deceived, but that matter less than to become a suspicious kind of guys. Any community is grounded in mutual trust and this one is not an exception.
Last: as much sooner or later truth comes to the fore, why to poison the ambience prematurely?
Not to believe what it is said to us was already made, at first, with Vas; now it was commited with this new russian guy.
How many extra times it will be done?

My best
Fernando
You make an interesting point that bears some expansion:
Take any two strong chess engines and have them perform WAC at 5 seconds. For well over 90% of the time, they will make the same choices. So is Fritz a clone of Tiger? Did Rybka borrow from Glaurung? What of Fruit and Hiarcs?

The answer is that all good chess engines will make the same choices here. So we should not get too excited when chess engines choose the same moves.

As far as "How many extra times it will be done?"
The answer is:
Every new chess engine stronger than 2400 Elo will provoke this reaction.

That's just the way it is. If history tells us anything, it is that history repeats itself. That's a real tragedy too. I remember Mr. Schleer (my 8th grade history teacher) wrote on the board:
"Those who do not learn from their mistakes are condemned to repeat them."
I raised my hand and asked about it:
"Exactly what lesson have we learned? About war? About prejudice or violence?"

We as humans repeat our mistakes every time. I wish it were otherwise, but I see it happen in myself all too frequently.

Uri Blass
Posts: 8607
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: The Strelka affair to date

Post by Uri Blass » Tue May 15, 2007 4:36 am

Dann Corbit wrote:
fern wrote:After the resolution of experts that took a look inside the bowels of Trelka, it appears that it is not a clone of Rybka.
Still, some people is dubious.
It seems that a law has been proclamed that any time two players play the same in the same position, that does not means both reached the positional truth of it, but someone is cloning the other.
It is curious to think like that, as much chess, differently of sheer artistic endeavour, has a layer of objetive truth. In many positions, like in mate-in-two problems solved equally by different players, human or artificial, it is only natural to play the same move.
In fact, it is NECCESARY that players of any kind should play the same if they are good enough to get the esentials of a position. Even different players of different levels can play the same.
So, if we see two engines playing the samemove in some positions, is it an enough good evidence that something wrong is going on?
Besides that, there is another consideration: commons sense indicates that you must take what is said to you in its value UNLESS there is HARD evidence not to believe it.
We can be deceived, but that matter less than to become a suspicious kind of guys. Any community is grounded in mutual trust and this one is not an exception.
Last: as much sooner or later truth comes to the fore, why to poison the ambience prematurely?
Not to believe what it is said to us was already made, at first, with Vas; now it was commited with this new russian guy.
How many extra times it will be done?

My best
Fernando
You make an interesting point that bears some expansion:
Take any two strong chess engines and have them perform WAC at 5 seconds. For well over 90% of the time, they will make the same choices. So is Fritz a clone of Tiger? Did Rybka borrow from Glaurung? What of Fruit and Hiarcs?

The answer is that all good chess engines will make the same choices here. So we should not get too excited when chess engines choose the same moves.

As far as "How many extra times it will be done?"
The answer is:
Every new chess engine stronger than 2400 Elo will provoke this reaction.

That's just the way it is. If history tells us anything, it is that history repeats itself. That's a real tragedy too. I remember Mr. Schleer (my 8th grade history teacher) wrote on the board:
"Those who do not learn from their mistakes are condemned to repeat them."
I raised my hand and asked about it:
"Exactly what lesson have we learned? About war? About prejudice or violence?"

We as humans repeat our mistakes every time. I wish it were otherwise, but I see it happen in myself all too frequently.
Of course no reason to suspect based on good moves that one engine is a clone.

The point is that not every strong engine will do the same type of mistakes.

some mistakes like null move with no verification are common to many engines so they may do the same type of mistakes but when you add not considering underpromotions and not understanding basic endgames then there are not many top engines that have this problem(I believe that all top engines except rybka and strelka or use tablebases or know that KBP vs K is a draw for the blind bishop).

Dann Corbit
Posts: 10203
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: The Strelka affair to date

Post by Dann Corbit » Tue May 15, 2007 4:52 am

Uri Blass wrote:
Dann Corbit wrote:
fern wrote:After the resolution of experts that took a look inside the bowels of Trelka, it appears that it is not a clone of Rybka.
Still, some people is dubious.
It seems that a law has been proclamed that any time two players play the same in the same position, that does not means both reached the positional truth of it, but someone is cloning the other.
It is curious to think like that, as much chess, differently of sheer artistic endeavour, has a layer of objetive truth. In many positions, like in mate-in-two problems solved equally by different players, human or artificial, it is only natural to play the same move.
In fact, it is NECCESARY that players of any kind should play the same if they are good enough to get the esentials of a position. Even different players of different levels can play the same.
So, if we see two engines playing the samemove in some positions, is it an enough good evidence that something wrong is going on?
Besides that, there is another consideration: commons sense indicates that you must take what is said to you in its value UNLESS there is HARD evidence not to believe it.
We can be deceived, but that matter less than to become a suspicious kind of guys. Any community is grounded in mutual trust and this one is not an exception.
Last: as much sooner or later truth comes to the fore, why to poison the ambience prematurely?
Not to believe what it is said to us was already made, at first, with Vas; now it was commited with this new russian guy.
How many extra times it will be done?

My best
Fernando
You make an interesting point that bears some expansion:
Take any two strong chess engines and have them perform WAC at 5 seconds. For well over 90% of the time, they will make the same choices. So is Fritz a clone of Tiger? Did Rybka borrow from Glaurung? What of Fruit and Hiarcs?

The answer is that all good chess engines will make the same choices here. So we should not get too excited when chess engines choose the same moves.

As far as "How many extra times it will be done?"
The answer is:
Every new chess engine stronger than 2400 Elo will provoke this reaction.

That's just the way it is. If history tells us anything, it is that history repeats itself. That's a real tragedy too. I remember Mr. Schleer (my 8th grade history teacher) wrote on the board:
"Those who do not learn from their mistakes are condemned to repeat them."
I raised my hand and asked about it:
"Exactly what lesson have we learned? About war? About prejudice or violence?"

We as humans repeat our mistakes every time. I wish it were otherwise, but I see it happen in myself all too frequently.
Of course no reason to suspect based on good moves that one engine is a clone.

The point is that not every strong engine will do the same type of mistakes.

some mistakes like null move with no verification are common to many engines so they may do the same type of mistakes but when you add not considering underpromotions and not understanding basic endgames then there are not many top engines that have this problem(I believe that all top engines except rybka and strelka or use tablebases or know that KBP vs K is a draw for the blind bishop).
I also agree that there were plenty of things that appeared strange about Strelka. I only meant to say that sometimes we may jump to conclusions prematurely. I was actually fairly well convinced that Strelka was a clone. On the other hand, not that I have seen the full evidence it is clear that Strelka is an interesting and original work.

User avatar
Rolf
Posts: 6081
Joined: Fri Mar 10, 2006 10:14 pm
Location: Munster, Nuremberg, Princeton

Re: The Strelka affair to date

Post by Rolf » Tue May 15, 2007 8:12 am

Dann Corbit wrote:
fern wrote:After the resolution of experts that took a look inside the bowels of Trelka, it appears that it is not a clone of Rybka.
Still, some people is dubious.
It seems that a law has been proclamed that any time two players play the same in the same position, that does not means both reached the positional truth of it, but someone is cloning the other.
It is curious to think like that, as much chess, differently of sheer artistic endeavour, has a layer of objetive truth. In many positions, like in mate-in-two problems solved equally by different players, human or artificial, it is only natural to play the same move.
In fact, it is NECCESARY that players of any kind should play the same if they are good enough to get the esentials of a position. Even different players of different levels can play the same.
So, if we see two engines playing the samemove in some positions, is it an enough good evidence that something wrong is going on?
Besides that, there is another consideration: commons sense indicates that you must take what is said to you in its value UNLESS there is HARD evidence not to believe it.
We can be deceived, but that matter less than to become a suspicious kind of guys. Any community is grounded in mutual trust and this one is not an exception.
Last: as much sooner or later truth comes to the fore, why to poison the ambience prematurely?
Not to believe what it is said to us was already made, at first, with Vas; now it was commited with this new russian guy.
How many extra times it will be done?

My best
Fernando
You make an interesting point that bears some expansion:
Take any two strong chess engines and have them perform WAC at 5 seconds. For well over 90% of the time, they will make the same choices. So is Fritz a clone of Tiger? Did Rybka borrow from Glaurung? What of Fruit and Hiarcs?

The answer is that all good chess engines will make the same choices here. So we should not get too excited when chess engines choose the same moves.

As far as "How many extra times it will be done?"
The answer is:
Every new chess engine stronger than 2400 Elo will provoke this reaction.

That's just the way it is. If history tells us anything, it is that history repeats itself. That's a real tragedy too. I remember Mr. Schleer (my 8th grade history teacher) wrote on the board:
"Those who do not learn from their mistakes are condemned to repeat them."
I raised my hand and asked about it:
"Exactly what lesson have we learned? About war? About prejudice or violence?"

We as humans repeat our mistakes every time. I wish it were otherwise, but I see it happen in myself all too frequently.
Look, after your intense exchange with C.Theron there was still this here:
http://216.25.93.108/forum/viewtopic.ph ... 89&t=13716

You asked "Exactly what lessons have we learned?" My answer is, it's the handling. Experienced judges know that the main point of their existence is their somewhat abstaining from taking part which then allows you to come to somewhat independent judgements.

If you once begin to answer questions, with details, then you enter a slippery slope because how could you re-analyse the whole topic (in public!) with people who dont have your abilities of a judge? A judge must bear in mind the whole at each litttle detail. And he must come to a final verdict. Re-analysing is for seminaries but for a judge it's nonsense.

Your juvenile question to the teacher is insofar nonsense because it's the context of the details which is reveiled only to someone with higher experience.

Actually I observe that details are explained with the intention that more people could accept your verdict. But this is a contradiction in itself.

Judges shouldnt discuss too much in public and never about their own verdicts. In special in a field where you are trusted to look upon normally secret code! Either you betray those who trusted you or you lose your own status as a trustworthy judge with expertise. So that experienced judges would never commit the error to "explain" their verdicts because they know about that zwick. All in all, a judge should better err in a specific case. That would be not as harmful as if he would violate the above iron rule.
-Popper and Lakatos are good but I'm stuck on Leibowitz

Post Reply