Crude, cruder, crudest

[arjuntemurnikar]

Where is white and where is black? Where is a1 and where h8? If it is as I understand it, I must turn upside down both my thought and my vision.

Stockfish uses the same convention.

a1....h1
..
..
a8....h8

OR

h1....a1
..
..
h8...a8

// It doesn't matter. They are both laterally symmetrical.

I think you missed this and your values are all UPSIDE DOWN! This might be the reason why they were doing so badly.

Can you send an updated PSQT? Also, it would greatly help if it was a .txt. Use Notepad instead of Word.

[Lyudmil Tsvetkov]

- strangely too, e6 and d6 pawns get good bonus, but f6 and c6 are negative, fully wrong

How can you say "fully wrong" if those values apparently have been shown to lead to better play than other values? Don't you see a problem in that?

Now, some people tell me my values are stupid and I do not have the right to suggest corrections based on sound chess knowledge. But how could I agree with this? Please look at the following diagrams.

[d]6k1/5p2/4pP2/3pP3/3P4/8/8/6K1 w - - 0 1
Some tables, for example Ippolit, would consider that the d4 white pawn deserves bigger bonus than the f6 pawn. Now, why is this wrong? In the first place, if f7 falls, and that is a possibility, f6 becomes a passer just 2 squares away from promotion, very dangerous one. In the second place, f6 restricts the enemy pieces' activity much more than d4 would do. It is ture that f6 is placed on a less central file, but the above 2 qualities more than compensate for that. Conclusion: f6 could not possibly be valued lower than d4.

Do not I have the right to voice my concern if I see something wrong? Do not you think it would be good to have in a single table terms like centralisation, space, potential distance to promotion square if a pawn becomes a passer?

[d]6k1/8/8/N7/8/3N4/8/6K1 w - - 0 1
Above you see 2 knights, on d3 and a5. Most tables would consider that the knight on d3 is much better placed, however, that is not true. Look at the squares both knights control.

Nd3 controls 4 squares which improve its placement, e5,c5,b4 and f4, and 4 squares which are regressions, e1,c1,b2,f2. So that chances are in 50% of cases it will land on a better square in the future.
Na5, although on the edge, controls 3 squares that are very good or improve its current placement, c4,c6 and b7 (where it also controls a potential outlet for an enemy rook on d8), and only one square which is a regression, b3. So that chances are in 75% of cases with a future move this knight will land on an optimal square, way above the 50% for the d3 knight. Besides, the knight on a5 restricts the enemy pieces' activity much more than the knight on d3. In spite of being on the edge, those 2 factor alone would suggest Na5 really deserves a very good value, not anything astonishing, but certainly not a dismal penalty which some tables assign. Mobility of the knights is another thing not directly related to psqt.

Now, please tell me why should I remain silent when I see an obvious incongruity? Is this throwing of mud?

[d]6k1/5p1p/6p1/8/3Q4/8/8/6K1 w - - 0 1

[d]6k1/5p1p/6p1/3Q4/8/8/8/6K1 w - - 0 1

[d]6k1/5p1p/3Q2p1/8/8/8/8/6K1 w - - 0 1

On the above diagrams you see 3 queens, placed on d4, d5 and d6 respectively. Which of the 3 queens is placed best? Most queen tables would say the queens on d4 and d5 are better placed than the queen on d6, however, this is not so. Reasons:

Qd4 controls only 3 squares on the 7th and 8th ranks, which are good for penetration, a7,d7 and d8, and is not able to deliver a check to the enemy king from a distance just one square away (something very important, we will see later).

Qd5 controls only 4 squares on the important for penetration 7th and 8th ranks, d7,b7,d8,b8, and similarly as Qd4 does not have the possibility to deliver a check to the enemy king from a distance just one square away.

Qd6 controls 6 squares on the 7th and 8th ranks, c7,d7,e7,b8,d8,f8, and is able to deliver check to the enemy king from a distance just one square away, from g6.

Make the calculations yourselves: Qd4 controls 3 important squares, Qd5 controls 4 important squares, while Qd6 controls 7! important squares. Which queen is placed best? Do not I have the right to tell that something might be wrong with engine psqt, if it really is?

[d]6k1/8/6Q1/8/8/8/8/6K1 b - - 0 1
Queen delivering check from just one square away (queen on e6 would also qualify for such a check, although diagonal checks with the queen are a bit less forceful). This is the most forceful and dangerous queen check of all available. Other queen checks are not so menacing.

Tell me names and whatever you wish, but I have the right to tell what I think when there are substantial reasons to do so.

[syzygy]

- strangely too, e6 and d6 pawns get good bonus, but f6 and c6 are negative, fully wrong

How can you say "fully wrong" if those values apparently have been shown to lead to better play than other values? Don't you see a problem in that?

Now, some people tell me my values are stupid and I do not have the right to suggest corrections based on sound chess knowledge. But how could I agree with this? Please look at the following diagrams.

I am not saying your values are stupid.
I am saying that it is silly to call something in a well-tuned engine "fully wrong" without having any solid data on it.
It is also silly to unconditionally submit that your proposed values are "corrections".

You claim to be right in everything. But there is nothing to back it up. On the contrary.

[arjuntemurnikar]

Nevermind, I did it myself.

Here is the edited code: (In order and without King PSQT)

Code: Select all

S(  0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S(  0, 0),
S(-14,-7), S(-6,-3), S(10, 5), S(14, 7), S(14, 7), S(10, 5), S(-6,-3), S(-14,-7),
S(-10,-5), S(-4,-2), S(14, 7), S(18, 9), S(18, 9), S(14, 7), S(-4,-2), S(-10,-5),
S( -6,-3), S(-2,-1), S(22,11), S(34,17), S(34,17), S(22,11), S(-2,-1), S( -6,-3),
S(  6, 3), S(10, 5), S(34,17), S(54,27), S(54,27), S(34,17), S(10, 5), S(  6, 3),
S( 26,13), S(54,27), S(64,32), S(74,37), S(74,37), S(64,32), S(54,27), S( 26,13),
S(  2, 1), S( 4, 2), S(14, 7), S(24,12), S(24,12), S(14, 7), S( 4, 2), S(  2, 1),
S(  0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S(  0, 0)

S(-80,-100), S(-60,-70), S(-50,-60), S(-40,-50), S(-40,-50), S(-50,-60), S(-60,-70), S(-80,-100),
S(-60, -70), S(-40,-50), S(-30,-30), S(-20,-20), S(-20,-20), S(-30,-30), S(-40,-50), S(-60, -70),
S(-20, -20), S(-10,-10), S(  0,  0), S( 20, 20), S( 20, 20), S(  0,  0), S(-10,-10), S(-20, -20),
S(-10, -10), S( 20, 20), S( 30, 30), S( 40, 40), S( 40, 40), S( 30, 30), S( 20, 20), S(-10, -10),
S( 20,  20), S( 40, 40), S( 50, 50), S( 60, 60), S( 60, 60), S( 50, 50), S( 40, 40), S( 20,  20),
S( 60,  30), S( 70, 60), S( 80, 70), S( 90, 80), S( 90, 80), S( 80, 70), S( 70, 60), S( 60,  30),
S( 10,  -5), S( 20,  5), S( 30, 10), S( 40, 20), S( 40, 20), S( 30, 10), S( 20,  5), S( 10,  -5),
S( 10, -40), S( 10,-20), S( 10,-10), S( 10,-10), S( 10,-10), S( 10,-10), S( 10,-20), S( 10, -40)

S(-80,-100), S(-60,-70), S(-50,-60), S(-40,-50), S(-40,-50), S(-50,-60), S(-60,-70), S(-80,-100),
S(-60, -70), S(-40,-50), S(-30,-30), S(-20,-20), S(-20,-20), S(-30,-30), S(-40,-50), S(-60, -70),
S(-20, -20), S(-10,-10), S(  0,  0), S( 20, 20), S( 20, 20), S(  0,  0), S(-10,-10), S(-20, -20),
S(-10, -10), S( 20, 20), S( 30, 30), S( 40, 40), S( 40, 40), S( 30, 30), S( 20, 20), S(-10, -10),
S( 20,  20), S( 40, 40), S( 50, 50), S( 60, 60), S( 60, 60), S( 50, 50), S( 40, 40), S( 20,  20),
S( 60,  30), S( 70, 60), S( 80, 70), S( 90, 80), S( 90, 80), S( 80, 70), S( 70, 60), S( 60,  30),
S( 10,  -5), S( 20,  5), S( 30, 10), S( 40, 20), S( 40, 20), S( 30, 10), S( 20,  5), S( 10,  -5),
S( 10, -40), S( 10,-20), S( 10,-10), S( 10,-10), S( 10,-10), S( 10,-10), S( 10,-20), S( 10, -40)

S(-25,-25), S(-20,-20), S(-15,-15), S(-10,-10), S(-10,-10), S(-15,-15), S(-20,-20), S(-25,-25),
S(-20,-20), S(-15,-15), S(-10,-10), S( -5, -5), S( -5, -5), S(-10,-10), S(-15,-15), S(-20,-20),
S(-10,-10), S( -5, -5), S(  0,  0), S( 10, 10), S( 10, 10), S(  0,  0), S( -5, -5), S(-10,-10),
S( -5, -5), S(  5,  5), S( 10, 10), S( 20, 20), S( 20, 20), S( 10, 10), S(  5,  5), S( -5, -5),
S(  5,  5), S( 10, 10), S( 20, 20), S( 30, 30), S( 30, 30), S( 20, 20), S( 10, 10), S(  5,  5),
S( 10, 10), S( 20, 20), S( 30, 30), S( 40, 40), S( 40, 40), S( 30, 30), S( 20, 20), S( 10, 10),
S( 50, 50), S( 55, 55), S( 60, 60), S( 60, 60), S( 60, 60), S( 60, 60), S( 55, 55), S( 50, 50),
S( 45, 45), S( 45, 45), S( 50, 50), S( 50, 50), S( 50, 50), S( 50, 50), S( 45, 45), S( 45, 45)

S(-60,-60), S(-50,-50), S(-40,-40), S(-30,-30), S(-30,-30), S(-40,-40), S(-50,-50), S(-60,-60),
S(-45,-45), S(-35,-35), S(-30,-30), S(-20,-20), S(-20,-20), S(-30,-30), S(-35,-35), S(-45,-45),
S(-40,-40), S(-20,-20), S(  0,  0), S( 20, 20), S( 20, 20), S(  0,  0), S(-20,-20), S(-40,-40),
S(-20,-20), S(  0,  0), S( 30, 30), S( 40, 40), S( 40, 40), S( 30, 30), S(  0,  0), S(-20,-20),
S( 10, 10), S( 30, 30), S( 40, 40), S( 60, 60), S( 60, 60), S( 40, 40), S( 30, 30), S( 10, 10),
S( 40, 40), S( 60, 60), S( 70, 70), S( 80, 80), S( 80, 80), S( 70, 70), S( 60, 60), S( 40, 40),
S( 10, 10), S( 30, 30), S( 40, 40), S( 60, 60), S( 60, 60), S( 40, 40), S( 30, 30), S( 10, 10),
S(  5,  5), S( 20, 20), S( 30, 30), S( 30, 30), S( 30, 30), S( 30, 30), S( 20, 20), S(  5,  5)

[jpqy]

I'am getting already better results!

After my first test this morning i begin to play around with Stockfish settings..because i find it plays too much defensive with these values..so adding more aggressiveness ,mobility mid & endgame ,cowardice it goes much better now.
Contempt didn't help..
In next week i continue search for better settings..to see if i can also get a positive score against Komodo.


Core i7 2670QM 4cores , Blitz 5m 0

Stockfish 050114IPx 64 SSE4.2 - Houdini 4 Pro x64B C0 4c 5.5 - 4.5 +3/=5/-2 55.00%
Stockfish 050114IPx 64 SSE4.2 - Stockwood 040114 64 SSE4.2 4c 6.0 - 4.0 +2/=8/-0 60.00%
Stockfish 050114IPx 64 SSE4.2 - Komodo TCEC 64-bit 4c 4.0 - 6.0 +2/=4/-4 40.00%

It looks like i'am 12hours busy with trying different settings ..who is the maniac you say

JP.

[Lyudmil Tsvetkov]

- strangely too, e6 and d6 pawns get good bonus, but f6 and c6 are negative, fully wrong

How can you say "fully wrong" if those values apparently have been shown to lead to better play than other values? Don't you see a problem in that?

Now, some people tell me my values are stupid and I do not have the right to suggest corrections based on sound chess knowledge. But how could I agree with this? Please look at the following diagrams.

I am not saying your values are stupid.
I am saying that it is silly to call something in a well-tuned engine "fully wrong" without having any solid data on it.
It is also silly to unconditionally submit that your proposed values are "corrections".

You claim to be right in everything. But there is nothing to back it up. On the contrary.

just for the record concerning the use of vocabulary: the harshest words I have said is that something might be incorrect, wrong, not perfect (well, I once said disastrous, but that was not directed at a person); on the other hand, you call me naive, silly, suggest I am throwing mud, etc.

[Lyudmil Tsvetkov]

I'am getting already better results!

After my first test this morning i begin to play around with Stockfish settings..because i find it plays too much defensive with these values..so adding more aggressiveness ,mobility mid & endgame ,cowardice it goes much better now.
Contempt didn't help..
In next week i continue search for better settings..to see if i can also get a positive score against Komodo.


Core i7 2670QM 4cores , Blitz 5m 0

Stockfish 050114IPx 64 SSE4.2 - Houdini 4 Pro x64B C0 4c 5.5 - 4.5 +3/=5/-2 55.00%
Stockfish 050114IPx 64 SSE4.2 - Stockwood 040114 64 SSE4.2 4c 6.0 - 4.0 +2/=8/-0 60.00%
Stockfish 050114IPx 64 SSE4.2 - Komodo TCEC 64-bit 4c 4.0 - 6.0 +2/=4/-4 40.00%

It looks like i'am 12hours busy with trying different settings ..who is the maniac you say

JP.

Thanks Jean-Paul.

I would not expect you would do that, but obviously you are one of the biggest Stockfish fans around.

Thanks again for the effort, would be curious for any new update.

[Lyudmil Tsvetkov]

Nevermind, I did it myself.

Here is the edited code: (In order and without King PSQT)

Code: Select all

S(  0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S(  0, 0),
S(-14,-7), S(-6,-3), S(10, 5), S(14, 7), S(14, 7), S(10, 5), S(-6,-3), S(-14,-7),
S(-10,-5), S(-4,-2), S(14, 7), S(18, 9), S(18, 9), S(14, 7), S(-4,-2), S(-10,-5),
S( -6,-3), S(-2,-1), S(22,11), S(34,17), S(34,17), S(22,11), S(-2,-1), S( -6,-3),
S(  6, 3), S(10, 5), S(34,17), S(54,27), S(54,27), S(34,17), S(10, 5), S(  6, 3),
S( 26,13), S(54,27), S(64,32), S(74,37), S(74,37), S(64,32), S(54,27), S( 26,13),
S(  2, 1), S( 4, 2), S(14, 7), S(24,12), S(24,12), S(14, 7), S( 4, 2), S(  2, 1),
S(  0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S( 0, 0), S(  0, 0)

S(-80,-100), S(-60,-70), S(-50,-60), S(-40,-50), S(-40,-50), S(-50,-60), S(-60,-70), S(-80,-100),
S(-60, -70), S(-40,-50), S(-30,-30), S(-20,-20), S(-20,-20), S(-30,-30), S(-40,-50), S(-60, -70),
S(-20, -20), S(-10,-10), S(  0,  0), S( 20, 20), S( 20, 20), S(  0,  0), S(-10,-10), S(-20, -20),
S(-10, -10), S( 20, 20), S( 30, 30), S( 40, 40), S( 40, 40), S( 30, 30), S( 20, 20), S(-10, -10),
S( 20,  20), S( 40, 40), S( 50, 50), S( 60, 60), S( 60, 60), S( 50, 50), S( 40, 40), S( 20,  20),
S( 60,  30), S( 70, 60), S( 80, 70), S( 90, 80), S( 90, 80), S( 80, 70), S( 70, 60), S( 60,  30),
S( 10,  -5), S( 20,  5), S( 30, 10), S( 40, 20), S( 40, 20), S( 30, 10), S( 20,  5), S( 10,  -5),
S( 10, -40), S( 10,-20), S( 10,-10), S( 10,-10), S( 10,-10), S( 10,-10), S( 10,-20), S( 10, -40)

S(-80,-100), S(-60,-70), S(-50,-60), S(-40,-50), S(-40,-50), S(-50,-60), S(-60,-70), S(-80,-100),
S(-60, -70), S(-40,-50), S(-30,-30), S(-20,-20), S(-20,-20), S(-30,-30), S(-40,-50), S(-60, -70),
S(-20, -20), S(-10,-10), S(  0,  0), S( 20, 20), S( 20, 20), S(  0,  0), S(-10,-10), S(-20, -20),
S(-10, -10), S( 20, 20), S( 30, 30), S( 40, 40), S( 40, 40), S( 30, 30), S( 20, 20), S(-10, -10),
S( 20,  20), S( 40, 40), S( 50, 50), S( 60, 60), S( 60, 60), S( 50, 50), S( 40, 40), S( 20,  20),
S( 60,  30), S( 70, 60), S( 80, 70), S( 90, 80), S( 90, 80), S( 80, 70), S( 70, 60), S( 60,  30),
S( 10,  -5), S( 20,  5), S( 30, 10), S( 40, 20), S( 40, 20), S( 30, 10), S( 20,  5), S( 10,  -5),
S( 10, -40), S( 10,-20), S( 10,-10), S( 10,-10), S( 10,-10), S( 10,-10), S( 10,-20), S( 10, -40)

S(-25,-25), S(-20,-20), S(-15,-15), S(-10,-10), S(-10,-10), S(-15,-15), S(-20,-20), S(-25,-25),
S(-20,-20), S(-15,-15), S(-10,-10), S( -5, -5), S( -5, -5), S(-10,-10), S(-15,-15), S(-20,-20),
S(-10,-10), S( -5, -5), S(  0,  0), S( 10, 10), S( 10, 10), S(  0,  0), S( -5, -5), S(-10,-10),
S( -5, -5), S(  5,  5), S( 10, 10), S( 20, 20), S( 20, 20), S( 10, 10), S(  5,  5), S( -5, -5),
S(  5,  5), S( 10, 10), S( 20, 20), S( 30, 30), S( 30, 30), S( 20, 20), S( 10, 10), S(  5,  5),
S( 10, 10), S( 20, 20), S( 30, 30), S( 40, 40), S( 40, 40), S( 30, 30), S( 20, 20), S( 10, 10),
S( 50, 50), S( 55, 55), S( 60, 60), S( 60, 60), S( 60, 60), S( 60, 60), S( 55, 55), S( 50, 50),
S( 45, 45), S( 45, 45), S( 50, 50), S( 50, 50), S( 50, 50), S( 50, 50), S( 45, 45), S( 45, 45)

S(-60,-60), S(-50,-50), S(-40,-40), S(-30,-30), S(-30,-30), S(-40,-40), S(-50,-50), S(-60,-60),
S(-45,-45), S(-35,-35), S(-30,-30), S(-20,-20), S(-20,-20), S(-30,-30), S(-35,-35), S(-45,-45),
S(-40,-40), S(-20,-20), S(  0,  0), S( 20, 20), S( 20, 20), S(  0,  0), S(-20,-20), S(-40,-40),
S(-20,-20), S(  0,  0), S( 30, 30), S( 40, 40), S( 40, 40), S( 30, 30), S(  0,  0), S(-20,-20),
S( 10, 10), S( 30, 30), S( 40, 40), S( 60, 60), S( 60, 60), S( 40, 40), S( 30, 30), S( 10, 10),
S( 40, 40), S( 60, 60), S( 70, 70), S( 80, 80), S( 80, 80), S( 70, 70), S( 60, 60), S( 40, 40),
S( 10, 10), S( 30, 30), S( 40, 40), S( 60, 60), S( 60, 60), S( 40, 40), S( 30, 30), S( 10, 10),
S(  5,  5), S( 20, 20), S( 30, 30), S( 30, 30), S( 30, 30), S( 30, 30), S( 20, 20), S(  5,  5)

It is much easier for you, Arjun.

You will not tell me that with completely reversed values Stockfish performed only 100 elo worse, will you?
I think I did it right the first time, as I read carefully the instructions for filling the table. I also think I understood correctly the Ippolit psqt. I would be surprised if it is otherwise. However, you never know, I already do not know where my left and my right are...

I would be afraid to look at any new results you might post if you retest with the reversed values Anyway, please share to unravel the enigma...

[arjuntemurnikar]

Where is white and where is black? Where is a1 and where h8? If it is as I understand it, I must turn upside down both my thought and my vision.

Stockfish uses the same convention.

a1....h1
..
..
a8....h8

OR

h1....a1
..
..
h8...a8

// It doesn't matter. They are both laterally symmetrical.

I think you missed this and your values are all UPSIDE DOWN! This might be the reason why they were doing so badly.

Can you send an updated PSQT? Also, it would greatly help if it was a .txt. Use Notepad instead of Word.

It is much easier for you, Arjun.

You will not tell me that with completely reversed values Stockfish performed only 100 elo worse, will you?
I think I did it right the first time, as I read carefully the instructions for filling the table. I also think I understood correctly the Ippolit psqt. I would be surprised if it is otherwise. However, you never know, I already do not know where my left and my right are...

I would be afraid to look at any new results you might post if you retest with the reversed values Anyway, please share to unravel the enigma...

Ok, I did a quick test with the edited values.

The good news is that it is an improvement. The bad news is that its still pretty bad.

Code: Select all

TC = 2" + 0.02
340 - 565 - 295 [0.406] 1200
Elo difference: -66

[syzygy]

just for the record concerning the use of vocabulary: the harshest words I have said is that something might be incorrect, wrong, not perfect (well, I once said disastrous, but that was not directed at a person);

I never saw you include the "might be" before. Instead, you usually write "is full wrong" and that your values "are optimal".

on the other hand, you call me naive, silly, suggest I am throwing mud, etc.

The term mud was probably poorly chosen. What I meant was you're just throwing spaghetti against the wall and leave it to others to clean up.