Push a patch

[lantonov]

I believe you're misinterpreting the PSQT. PSQT values are not penalties, but bonuses. So what you're seeing is that in the endgame, SF prefers to have its king centralized.

If this is directed to me, I do not see where is my misinterpretation. I know that the PSQT table is Bonus, at least its name says so. But negative bonus is a penalty. As concerns the king, the fact that such large bonuses are given to it in the PSQT means that these are not given to the king elsewhere where they would be more specific.

[Haikouichthys]

I believe you're misinterpreting the PSQT. PSQT values are not penalties, but bonuses. So what you're seeing is that in the endgame, SF prefers to have its king centralized.

I am misinterpreting nothing.

Of course, they are bonuses, and not penalties.

Here is my corrected table, but I did not correct any ranks above 4.

S(303, 27), S(327, 81), S(270,108), S(220,116),
S(290, 74), S(321,128), S(260,155), S(210,165),
S(220,113), S(256,167), S(196,195), S(148,202),
S(192,135), S(226,193), S(167,219), S(119,228),
S(173,135), S(207,189), S(148,216), S(100,224),
S(146,111), S(180,165), S(121,192), S( 73,200),
S(119, 74), S(153,128), S( 94,155), S( 46,163),
S( 98, 27), S(132, 81), S( 73,108), S( 25,116),


As you see, SF prefers 4th rank in the mg to 8th rank, but king is mated easier not in the center, but on the 8th rank.

SF prefers 4th rank to 5th and 6th ranks in the eg, but 5th and 6th ranks are better for the king in the eg.

etc., etc.

So SF has it all wrong, but, when you tune 5 bad terms well, they produce a working unity, that is far from optimal, but difficult to correct, as you do not know what related terms interact and how you should change their values.

That is why I am already afraid to touch anything no matter how suspicious it looks: it seems that currently only small gradual steps are possible, both in tuning and in general knowledge implementation.

My mistake, I was confused by your saying that fourth and fifth ranks had a higher mg 'penalty' than 7th and 8th. I think I'm still confused. You say the king is mated more easily on the 8th rank than on the 4th, but isn't that a reason to prefer the 4th to the 8th in the middlegame, exactly as SF does?

EDIT2: Edited out my first edit. Memo to self: No posting while hungry.

[Lyudmil Tsvetkov]

I believe you're misinterpreting the PSQT. PSQT values are not penalties, but bonuses. So what you're seeing is that in the endgame, SF prefers to have its king centralized.

If this is directed to me, I do not see where is my misinterpretation. I know that the PSQT table is Bonus, at least its name says so. But negative bonus is a penalty. As concerns the king, the fact that such large bonuses are given to it in the PSQT means that these are not given to the king elsewhere where they would be more specific.

This was directed to me, I suppose, but, as we are namesakes...

Large bonuses are quite in place, as, you slightly go wrong with the square where you are, and you get mated, I do not see a problem here; the issue for me is that squares that are due larger bonus get smaller one and vice-versa.

But, as the problem happens in almost all kinds of tables and terms, this seems to autocorrect itself.

[Lyudmil Tsvetkov]

I believe you're misinterpreting the PSQT. PSQT values are not penalties, but bonuses. So what you're seeing is that in the endgame, SF prefers to have its king centralized.

I am misinterpreting nothing.

Of course, they are bonuses, and not penalties.

Here is my corrected table, but I did not correct any ranks above 4.

S(303, 27), S(327, 81), S(270,108), S(220,116),
S(290, 74), S(321,128), S(260,155), S(210,165),
S(220,113), S(256,167), S(196,195), S(148,202),
S(192,135), S(226,193), S(167,219), S(119,228),
S(173,135), S(207,189), S(148,216), S(100,224),
S(146,111), S(180,165), S(121,192), S( 73,200),
S(119, 74), S(153,128), S( 94,155), S( 46,163),
S( 98, 27), S(132, 81), S( 73,108), S( 25,116),


As you see, SF prefers 4th rank in the mg to 8th rank, but king is mated easier not in the center, but on the 8th rank.

SF prefers 4th rank to 5th and 6th ranks in the eg, but 5th and 6th ranks are better for the king in the eg.

etc., etc.

So SF has it all wrong, but, when you tune 5 bad terms well, they produce a working unity, that is far from optimal, but difficult to correct, as you do not know what related terms interact and how you should change their values.

That is why I am already afraid to touch anything no matter how suspicious it looks: it seems that currently only small gradual steps are possible, both in tuning and in general knowledge implementation.

My mistake, I was confused by your saying that fourth and fifth ranks had a higher mg 'penalty' than 7th and 8th. I think I'm still confused. You say the king is mated more easily on the 8th rank than on the 4th, but isn't that a reason to prefer the 4th to the 8th in the middlegame, exactly as SF does?

EDIT2: Edited out my first edit. Memo to self: No posting while hungry.

What do you mean top left is a1/a8 square?

I thought top left is a1/h1 square, as the board is cut vertically, and not horizontally; the a1 and h1 squares get equal, symmetrical values, and not the a1 and a8 squares.

Considering the specific values, I was right for the eg, but for the mg I simply meant that SF could decrease even more the bonus for more advanced ranks. One way or another, I am not able to look at SF values at all.

For example, look at the mg penalties for the h8/a8, g8/b8 and c8/f8 squares. 98;132;73.

Why on earth the king is much safer on g8 than on h8?
And on h8 is safer than on f8?

Is not that a bit of a contradiction?

[Haikouichthys]

I believe you're misinterpreting the PSQT. PSQT values are not penalties, but bonuses. So what you're seeing is that in the endgame, SF prefers to have its king centralized.

I am misinterpreting nothing.

Of course, they are bonuses, and not penalties.

Here is my corrected table, but I did not correct any ranks above 4.

S(303, 27), S(327, 81), S(270,108), S(220,116),
S(290, 74), S(321,128), S(260,155), S(210,165),
S(220,113), S(256,167), S(196,195), S(148,202),
S(192,135), S(226,193), S(167,219), S(119,228),
S(173,135), S(207,189), S(148,216), S(100,224),
S(146,111), S(180,165), S(121,192), S( 73,200),
S(119, 74), S(153,128), S( 94,155), S( 46,163),
S( 98, 27), S(132, 81), S( 73,108), S( 25,116),


As you see, SF prefers 4th rank in the mg to 8th rank, but king is mated easier not in the center, but on the 8th rank.

SF prefers 4th rank to 5th and 6th ranks in the eg, but 5th and 6th ranks are better for the king in the eg.

etc., etc.

So SF has it all wrong, but, when you tune 5 bad terms well, they produce a working unity, that is far from optimal, but difficult to correct, as you do not know what related terms interact and how you should change their values.

That is why I am already afraid to touch anything no matter how suspicious it looks: it seems that currently only small gradual steps are possible, both in tuning and in general knowledge implementation.

My mistake, I was confused by your saying that fourth and fifth ranks had a higher mg 'penalty' than 7th and 8th. I think I'm still confused. You say the king is mated more easily on the 8th rank than on the 4th, but isn't that a reason to prefer the 4th to the 8th in the middlegame, exactly as SF does?

EDIT2: Edited out my first edit. Memo to self: No posting while hungry.

What do you mean top left is a1/a8 square?

I thought top left is a1/h1 square, as the board is cut vertically, and not horizontally; the a1 and h1 squares get equal, symmetrical values, and not the a1 and a8 squares.

Considering the specific values, I was right for the eg, but for the mg I simply meant that SF could decrease even more the bonus for more advanced ranks. One way or another, I am not able to look at SF values at all.

For example, look at the mg penalties for the h8/a8, g8/b8 and c8/f8 squares. 98;132;73.

Why on earth the king is much safer on g8 than on h8?
And on h8 is safer than on f8?

Is not that a bit of a contradiction?

Sorry, yeah, my edit was wrong. I posted before thinking it through.

[Haikouichthys]

You might be right about the rank 8 middlegame values being poorly tuned, but it seems like it should matter very little in terms of ELO, as it's extremely rare for the king to reach so advanced a rank in the midgame, and most realistic search lines should run out of depth due to reductions and pruning long before the king gets so far advanced.

[Lyudmil Tsvetkov]

You might be right about the rank 8 middlegame values being poorly tuned, but it seems like it should matter very little in terms of ELO, as it's extremely rare for the king to reach so advanced a rank in the midgame, and most realistic search lines should run out of depth due to reductions and pruning long before the king gets so far advanced.

Good point, but that is not always true, sometimes those squares are also relevant.

But that is of course not the most salient SF weakness.

As you know, SF thinks knight values increases more relatively to bishop value from mg to eg: fully wrong, but when you try to correct it, the more decisive in your changes you are, the easier you fail.

So SF wrong piece values are tuned as a whole to other many wrongly-postulated parameters, and separately seen they are very inadequate, as a whole they work somehow, you try to change them, and it is very difficult, as many other terms also depend on those changes.

So simply no way forward - you should do only very small, infinitesimal changes, if you want to succeed, otherwise SF refuses to comply.

[Lyudmil Tsvetkov]

Now, how I would like to see a patch on this in the framework.

Blocked pawns are of course more serious weakness than non-blocked pawns, but not all blocked pawns are equal - if there is a single blocked pawn you should penalise, this is of course d4/e4 central blocked pawn.

[Haikouichthys]

Hm. Have you looked at the material imbalance tables? I notice that knights gain a significant bonus with more pawns on the board, whilst bishops gain a significantly smaller bonus. As the game progresses, not always but generally you'll see the number of pawns on the board decrease, and so the knight will lose value relative to bishops.