Zugzwang Test Suite ZPTS by Robert Allgeuer

[Uri Blass]

I took my GPS in car so he guides me from point A to point B. If it makes me go through xyz points, I throw it out the window.

No rule in chess says "you must play the shortest possible mate." Hence, not all engines try to do that. The first mate found is more than good enough.

There are no rules in effect, it's just a matter of common sense.
An engine that shows lazy in his search for checkmate, it leaves me very great doubts in his choice of moves in the rest of the game. If you can not understand it ...

I suppose ALL humans are inherently lazy then? Once you find and verify a forced mate, do you continue to burn time on your clock looking to see if you can find something shorter? Your comment about other moves in the rest of the game makes absolutely zero sense. A mate is forced. Period. Guaranteed by the search. Other moves are not forced, except in rare cases, so a program's choice there is "iffy" anyway.

Not all humans are the same and I know at least one player who continue to burn time on his clock to search for shorter mate after he found a mate(in case that he has time).

There are also some winning moves that engines can make that humans usually do not do like sacrificing a queen to get a winning tablebases KNN vs KP

Note that it is not that I criticize the strength of the engines for making non optimal winning moves but only respond to the claim about humans
and to explain the reason that engines can seem stupid to humans.

Humans are usually not going to sacrifice a queen for a pawn even if they know that sacrificing the queen lead to some KBN vs K position that they know that it is a win because they prefer a simpler win with KBNQ vs KP.

[phenri]

I took my GPS in car so he guides me from point A to point B. If it makes me go through xyz points, I throw it out the window.

Of course but the your target is not only to get from point A to point B but to do it fast.

When the target is to get higher chess rating it is unimportant if you win faster or slower.

You are mistaken ... The characteristic of a strong player or a strong engine is that most moves are more accurate, are rating increases, and vice versa. I am surprised to read this from you.

The sentence on the GPS was just a metaphor, as I will say this: "Over the tool is accurate, more efficient you are."

Now we do not go astray please, this thread is for discussions about the Zugzwang and the difficulty of engines in certain positions and not the checkmate in x or y number of moves started by someone.

(blah blah blah)
Are world champion candidates weak?
(blah blah blah)

Excuse me, but where did you read that?
With all the respect, now stop breaking my balls.

Got it from this:

The characteristic of a strong player or a strong engine is that most moves are more accurate, are rating increases,

ALL players make mistakes. I don't happen to call not finding the shortest mate a "mistake" however. The object of the game of chess is to beat your opponent. No extra points for doing it in fewer moves.

You misinterpret my words and you take it out of the context post by post.
I will not answer you anymore and bring me to your level because you seem to be a champion in this game

[bob]

I took my GPS in car so he guides me from point A to point B. If it makes me go through xyz points, I throw it out the window.

Of course but the your target is not only to get from point A to point B but to do it fast.

When the target is to get higher chess rating it is unimportant if you win faster or slower.

You are mistaken ... The characteristic of a strong player or a strong engine is that most moves are more accurate, are rating increases, and vice versa. I am surprised to read this from you.

The sentence on the GPS was just a metaphor, as I will say this: "Over the tool is accurate, more efficient you are."

Now we do not go astray please, this thread is for discussions about the Zugzwang and the difficulty of engines in certain positions and not the checkmate in x or y number of moves started by someone.

(blah blah blah)
Are world champion candidates weak?
(blah blah blah)

Excuse me, but where did you read that?
With all the respect, now stop breaking my balls.

Got it from this:

The characteristic of a strong player or a strong engine is that most moves are more accurate, are rating increases,

ALL players make mistakes. I don't happen to call not finding the shortest mate a "mistake" however. The object of the game of chess is to beat your opponent. No extra points for doing it in fewer moves.

You misinterpret my words and you take it out of the context post by post.
I will not answer you anymore and bring me to your level because you seem to be a champion in this game

Doesn't matter to me whether you respond or not. You raised TWO issues:

(1) I do not find normal to have random results in smp. And return the fault to the CPU is a false excuse.

(2) I return you the question, why do you prefer mate in 14 or even 12 moves when it is in 3 moves? Or why do you want a flying Paris-New York with a stop at the North Pole when you can not stop?


I tried to answer each politely AND accurately. SMP is non-deterministic due to a sequential algorithm executed in a parallel way. Like it or not, that is simply the way it is. It is the way it always has been and it is the way it always will be, when using alpha/beta. And with nothing coming even close to alpha/beta, we are stuck.

Your second point is nonsense. What about this, which is what the program is actually doing. You enter a maze in Paris, with the goal of getting to New York. When you get to New York, are you going to BACK UP and try to find a shorter path, even though you are already there? The program has no idea there is a direct path from Paris to New York. It's goal is to simply get to New York. It takes the first path it finds, because by the time it FINDS the path, it is already there. Stupid to back up and try to find a shorter path.

Your analogies are nonsense, like it or not.

[phenri]

I took my GPS in car so he guides me from point A to point B. If it makes me go through xyz points, I throw it out the window.

Of course but the your target is not only to get from point A to point B but to do it fast.

When the target is to get higher chess rating it is unimportant if you win faster or slower.

You are mistaken ... The characteristic of a strong player or a strong engine is that most moves are more accurate, are rating increases, and vice versa. I am surprised to read this from you.

The sentence on the GPS was just a metaphor, as I will say this: "Over the tool is accurate, more efficient you are."

Now we do not go astray please, this thread is for discussions about the Zugzwang and the difficulty of engines in certain positions and not the checkmate in x or y number of moves started by someone.

(blah blah blah)
Are world champion candidates weak?
(blah blah blah)

Excuse me, but where did you read that?
With all the respect, now stop breaking my balls.

Got it from this:

The characteristic of a strong player or a strong engine is that most moves are more accurate, are rating increases,

ALL players make mistakes. I don't happen to call not finding the shortest mate a "mistake" however. The object of the game of chess is to beat your opponent. No extra points for doing it in fewer moves.

You misinterpret my words and you take it out of the context post by post.
I will not answer you anymore and bring me to your level because you seem to be a champion in this game

Doesn't matter to me whether you respond or not. You raised TWO issues:

(1) I do not find normal to have random results in smp. And return the fault to the CPU is a false excuse.

(2) I return you the question, why do you prefer mate in 14 or even 12 moves when it is in 3 moves? Or why do you want a flying Paris-New York with a stop at the North Pole when you can not stop?


I tried to answer each politely AND accurately. SMP is non-deterministic due to a sequential algorithm executed in a parallel way. Like it or not, that is simply the way it is. It is the way it always has been and it is the way it always will be, when using alpha/beta. And with nothing coming even close to alpha/beta, we are stuck.

Your second point is nonsense. What about this, which is what the program is actually doing. You enter a maze in Paris, with the goal of getting to New York. When you get to New York, are you going to BACK UP and try to find a shorter path, even though you are already there? The program has no idea there is a direct path from Paris to New York. It's goal is to simply get to New York. It takes the first path it finds, because by the time it FINDS the path, it is already there. Stupid to back up and try to find a shorter path.

Your analogies are nonsense, like it or not.

Point 1 was not addressed to you specifically on the one hand, and secondly, please spare me your jargon because the present case I compared the Critter versus Stockfish engine. One always gave the same result fairly quickly while the other was random.

Point 2 It is same, I compared engines such as Critter or Houdini against Stockfish with results runs, constant, quick and accurate in this TestSuite.
Know that there are other much better programmed and optimized engines than Crafty... and that the world does not revolve around you and your dogmatic theories.

QED

So, when is a checkmate in 3 moves, it is 3 and not 12!! You can think of this?

[Uri Blass]

I took my GPS in car so he guides me from point A to point B. If it makes me go through xyz points, I throw it out the window.

Of course but the your target is not only to get from point A to point B but to do it fast.

When the target is to get higher chess rating it is unimportant if you win faster or slower.

You are mistaken ... The characteristic of a strong player or a strong engine is that most moves are more accurate, are rating increases, and vice versa. I am surprised to read this from you.

The sentence on the GPS was just a metaphor, as I will say this: "Over the tool is accurate, more efficient you are."

Now we do not go astray please, this thread is for discussions about the Zugzwang and the difficulty of engines in certain positions and not the checkmate in x or y number of moves started by someone.

(blah blah blah)
Are world champion candidates weak?
(blah blah blah)

Excuse me, but where did you read that?
With all the respect, now stop breaking my balls.

Got it from this:

The characteristic of a strong player or a strong engine is that most moves are more accurate, are rating increases,

ALL players make mistakes. I don't happen to call not finding the shortest mate a "mistake" however. The object of the game of chess is to beat your opponent. No extra points for doing it in fewer moves.

You misinterpret my words and you take it out of the context post by post.
I will not answer you anymore and bring me to your level because you seem to be a champion in this game

Doesn't matter to me whether you respond or not. You raised TWO issues:

(1) I do not find normal to have random results in smp. And return the fault to the CPU is a false excuse.

(2) I return you the question, why do you prefer mate in 14 or even 12 moves when it is in 3 moves? Or why do you want a flying Paris-New York with a stop at the North Pole when you can not stop?


I tried to answer each politely AND accurately. SMP is non-deterministic due to a sequential algorithm executed in a parallel way. Like it or not, that is simply the way it is. It is the way it always has been and it is the way it always will be, when using alpha/beta. And with nothing coming even close to alpha/beta, we are stuck.

Your second point is nonsense. What about this, which is what the program is actually doing. You enter a maze in Paris, with the goal of getting to New York. When you get to New York, are you going to BACK UP and try to find a shorter path, even though you are already there? The program has no idea there is a direct path from Paris to New York. It's goal is to simply get to New York. It takes the first path it finds, because by the time it FINDS the path, it is already there. Stupid to back up and try to find a shorter path.

Your analogies are nonsense, like it or not.

Point 1 was not addressed to you specifically on the one hand, and secondly, please spare me your jargon because the present case I compared the Critter versus Stockfish engine. One always gave the same result fairly quickly while the other was random.

Point 2 It is same, I compared engines such as Critter or Houdini against Stockfish with results runs, constant, quick and accurate in this TestSuite.
Know that there are other much better programmed and optimized engines than Crafty... and that the world does not revolve around you and your dogmatic theories.

QED

So, when is a checkmate in 3 moves, it is 3 and not 12!! You can think of this?

1)I did not know that Critter or Houdini gave the same result
and I will be surprised if Critter with more than 1 core or Houdini with more than 1 core are deterministic in their choice.

It is possible that they have less variety than stockfish and Critter can need sometimes 1,000,000 nodes and sometimes 1,200,000 nodes to solve the same position but not sometimes 1,000,000 nodes and sometimes 10,000,000 nodes.

If you meant to say that stockfish has a bigger variety then it was not good to claim
"I do not find normal to have random results in smp. And return the fault to the CPU is a false excuse." and it was better to say that you do not find normal variety that is as big as stockfish's variety.

2)I do not see a reason to assume that Critter or Houdini are better only because they have a smaller variety in time to solution and I see no reason to compare with Crafty.

The facts are that we have no evidence that Critter or Houdini are better than stockfish so I see no reason to assume that the behaviour of Houdini and Critter is the normal behaviour and not the behaviour of stockfish.

Note that it is easy to search for shorter mate after you already found a mate if you want to do it and some programmers do it not because it improve the engine but for different reasons.

I do not think that finding the shorter mate is relevant for playing strength.

[phenri]

1)I did not know that Critter or Houdini gave the same result
and I will be surprised if Critter with more than 1 core or Houdini with more than 1 core are deterministic in their choice.

It is possible that they have less variety than stockfish and Critter can need sometimes 1,000,000 nodes and sometimes 1,200,000 nodes to solve the same position but not sometimes 1,000,000 nodes and sometimes 10,000,000 nodes.

If you meant to say that stockfish has a bigger variety then it was not good to claim
"I do not find normal to have random results in smp. And return the fault to the CPU is a false excuse." and it was better to say that you do not find normal variety that is as big as stockfish's variety.

2)I do not see a reason to assume that Critter or Houdini are better only because they have a smaller variety in time to solution and I see no reason to compare with Crafty.

The facts are that we have no evidence that Critter or Houdini are better than stockfish so I see no reason to assume that the behaviour of Houdini and Critter is the normal behaviour and not the behaviour of stockfish.

Note that it is easy to search for shorter mate after you already found a mate if you want to do it and some programmers do it not because it improve the engine but for different reasons.

I do not think that finding the shorter mate is relevant for playing strength.

1) Wrong! For you to understand, you just reverse the name of the engines.
2) And since the point 2) is based on the same theory as you advance in point 1), then it is equally wrong!

[bob]

I took my GPS in car so he guides me from point A to point B. If it makes me go through xyz points, I throw it out the window.

Of course but the your target is not only to get from point A to point B but to do it fast.

When the target is to get higher chess rating it is unimportant if you win faster or slower.

You are mistaken ... The characteristic of a strong player or a strong engine is that most moves are more accurate, are rating increases, and vice versa. I am surprised to read this from you.

The sentence on the GPS was just a metaphor, as I will say this: "Over the tool is accurate, more efficient you are."

Now we do not go astray please, this thread is for discussions about the Zugzwang and the difficulty of engines in certain positions and not the checkmate in x or y number of moves started by someone.

(blah blah blah)
Are world champion candidates weak?
(blah blah blah)

Excuse me, but where did you read that?
With all the respect, now stop breaking my balls.

Got it from this:

The characteristic of a strong player or a strong engine is that most moves are more accurate, are rating increases,

ALL players make mistakes. I don't happen to call not finding the shortest mate a "mistake" however. The object of the game of chess is to beat your opponent. No extra points for doing it in fewer moves.

You misinterpret my words and you take it out of the context post by post.
I will not answer you anymore and bring me to your level because you seem to be a champion in this game

Doesn't matter to me whether you respond or not. You raised TWO issues:

(1) I do not find normal to have random results in smp. And return the fault to the CPU is a false excuse.

(2) I return you the question, why do you prefer mate in 14 or even 12 moves when it is in 3 moves? Or why do you want a flying Paris-New York with a stop at the North Pole when you can not stop?


I tried to answer each politely AND accurately. SMP is non-deterministic due to a sequential algorithm executed in a parallel way. Like it or not, that is simply the way it is. It is the way it always has been and it is the way it always will be, when using alpha/beta. And with nothing coming even close to alpha/beta, we are stuck.

Your second point is nonsense. What about this, which is what the program is actually doing. You enter a maze in Paris, with the goal of getting to New York. When you get to New York, are you going to BACK UP and try to find a shorter path, even though you are already there? The program has no idea there is a direct path from Paris to New York. It's goal is to simply get to New York. It takes the first path it finds, because by the time it FINDS the path, it is already there. Stupid to back up and try to find a shorter path.

Your analogies are nonsense, like it or not.

Point 1 was not addressed to you specifically on the one hand, and secondly, please spare me your jargon because the present case I compared the Critter versus Stockfish engine. One always gave the same result fairly quickly while the other was random.

Point 2 It is same, I compared engines such as Critter or Houdini against Stockfish with results runs, constant, quick and accurate in this TestSuite.
Know that there are other much better programmed and optimized engines than Crafty... and that the world does not revolve around you and your dogmatic theories.

QED

So, when is a checkmate in 3 moves, it is 3 and not 12!! You can think of this?

You apparently can't follow any of the discussion, so I suppose it really is futile to continue trying to explain that which you refuse to acknowledge. NO parallel search produces the same move, with the same score, in the same time, every time it is run. Just does not happen. For some positions they might produce the same move every time, but the time will vary wildly. For some positions, it might produce different moves each time.

But there is no point in discussing this since you are not listening and not testing...

The computer doesn't necessarily know about the checkmate in three, when it first finds the checkmate in twelve moves. You understand that or not? If you want to waste time after finding a forced mate in 12, to search for one that is shorter, that is fine. But you already have the mate in 12 "in hand". Your example is completely unrelated to how chess programs work, which makes it meaningless...

Carry on in the foggy world you live in, I can only explain, I can't make you understand.

[Uri Blass]

1)I did not know that Critter or Houdini gave the same result
and I will be surprised if Critter with more than 1 core or Houdini with more than 1 core are deterministic in their choice.

It is possible that they have less variety than stockfish and Critter can need sometimes 1,000,000 nodes and sometimes 1,200,000 nodes to solve the same position but not sometimes 1,000,000 nodes and sometimes 10,000,000 nodes.

If you meant to say that stockfish has a bigger variety then it was not good to claim
"I do not find normal to have random results in smp. And return the fault to the CPU is a false excuse." and it was better to say that you do not find normal variety that is as big as stockfish's variety.

2)I do not see a reason to assume that Critter or Houdini are better only because they have a smaller variety in time to solution and I see no reason to compare with Crafty.

The facts are that we have no evidence that Critter or Houdini are better than stockfish so I see no reason to assume that the behaviour of Houdini and Critter is the normal behaviour and not the behaviour of stockfish.

Note that it is easy to search for shorter mate after you already found a mate if you want to do it and some programmers do it not because it improve the engine but for different reasons.

I do not think that finding the shorter mate is relevant for playing strength.

1) Wrong! For you to understand, you just reverse the name of the engines.
2) And since the point 2) is based on the same theory as you advance in point 1), then it is equally wrong!

I do not understand what you mean by reverse the name of the engines.
I ran Critter with 4 cores twice on position number 14 and got different results(one time it needed depth 17 and one time it needed depth 18) so my theory that Critter is not deterministic like stockfish is proved.

For houdini I do not have Houdini4 so I cannot check it.

[lech]

Sting very good ¡¡¡

Sting SF 4 x64-509MB/ 28 positions Total time: 0:01:41

IvanHoe Zuzgwang x64-509MB/ 26 positions Total time: 0:03:10


IvanHoe Zuzgwand + source.
https://www.dropbox.com/s/f361aayti20mw ... uzgwand.7z

Thanks Jose, Good work for this interesting topic.
Theme of zugzwang is my liked.

[phenri]

Conditions: 4 cores - 20 seconds
Sting SF 4-512MB
Houdini 3 Pro x64-512MB
Houdini 4 Pro x64A-512MB
Critter 1.6a 64-bit-512MB
IvanHoe Zuzgwang x64-512MB
Komodo TCEC 64-bit-512MB
Gull 2.8 beta x64-512MB
Gull 2.9 alpha x64-512MB
DON 1.0b pop64-512MB
Stockfish 120214 64 SSE4.2 yzygy-512MB
Stockfish 2.3.1 JA 64bit SSE4.2-512MB
Stockfish 120214 64 SSE4.2 -512MB

Data: cbv & pgn, calc & csv
ZPTS_results.7z mirror ZPTS_results.7z