Pawn checks and center levers

[Lyudmil Tsvetkov]

I see that no one is interested in my stupid posts, but I will take the freedom to post one last very brief message, as I find the topics worth it.

One simple rule for pawn checks:

- pawn checks are much less important than the threat of the check itself, in sharp distinction to all kinds of piece checks, where the check itself is more important than the threat of a check
- unsafe pawn checks are far more important than safe pawn checks, so a check not supported by any own pawn or piece is usually more valuable than a check supported by own pawns and pieces

So that, with pawn checks, everything works the other way.
The main reason for this paradoxical behaviour is that pawn checks are mainly a tactical weapon.

[d]6k1/7P/2k5/8/1PK5/5p2/8/1B4K1 b - - 0 1
The white check with the h7 pawn, although defended by an own bishop, is much less important than the threat of a check/potential check by the black pawn on f3 upon the white king, with f3-f2, although that check would be unsafe and unsupported.

As said, pawn checks are mainly a tactical weapon.

In the endgame, potential pawn checks like b4-b5, with the checking pawn supported only by the own king, are also important.

I would give potential pawn checks some 5-10cps bonus, both for the mg and eg, though the mg bonus might be a bit higher.


Passing over to the topic of center levers.

Some engines implement bonus for advanced levers, i.e. bonus for levers on the 5th and 6th ranks. This makes sense.

However, it is also possible to break the symmetry and use the concept also for levers on the 4th rank, that would be symmetrical in terms of ranks, but not symmetrical in terms of files.
Of course, you will try to give the bonus for an outer, less central pawn, attacking an inner, more central pawn. This is one of the suggestions of Mr. Kmoch.

Unfortunately, it will not work in most cases.
It works only in the center, and apart from that, there are only 2 squares on the entire board that deserve the bonus: a c4 white pawn attacking a d5 black pawn, and an f4 white pawn, attacking an e5 black pawn. Other tries might not work, but those 2 squares are very important.

Reasons for giving bonus for those particular squares?

- a c4 white pawn attacking a d5 black pawn might be traded for a more valuable, more central e6 black pawn, but also for less valuable c6 black pawn, that is on the same file, but less advanced, so that is not the valid reason
- a good reason is that it is always well to attack the center, the earlier the better
- another good reason is that the concept helps flank attacks like c2-c4 and f2-f4, which are very efficient, especially f2-f4, and particularly when made earlier

[d]6k1/8/8/3pp3/2P2P2/8/8/6K1 w - - 0 1
c4 and f4 pawns/center levers get the bonus here, in spite of the fact that the attacked enemy pawns are also on the 4th rank

I would give some 5-10-15cps, who knows, to center c4-d5 and f4-e5 levers, just for the middlegame.

OK, sorry for posting, I know no one is interested, but at least I am sure Carl will read it.

[Lyudmil Tsvetkov]

As I said, the key to good, better and excellent evaluation is to apply specific concepts to specific squares.
In that case, the particular case is worth applying to only 2 squares.

Also, a similar concept could help players choose good opening lines like the Tarrasch Defence in the QGD, playing c4 on the second move after d4 d5 instead of the Trompovsky attack, and possibly start good king-side KID-type attacks with f2-f4. (the last of which, of course, I do not believe)

Of course, the above is under the supposition that no books or very short books are used for testing.

Please do not tell me that the King's Gambit, e4 e5 f4, is already busted, and there is the f4-e5 lever here, as, according to Fischer, black busts it by swiftly terminating the center lever, playing ef4 on the second move with the King's Gambit accepted, followed by d6.

Actually, if you think a bit more about it, the line with Bc5, followed by d6, declining the Gambit, and leaving the f4-e5 lever stay for a very long time, only favours white.

Any opinions on those suggestions?

[jpqy]

Lyudmil..you don't post stupid things..else you would not have always so many views with your topic's.
It's not that people react on your messages that they don't read it..and in many cases your idea's get tried..
For sure i follow them

Kind regards,
JP.

[Lyudmil Tsvetkov]

Lyudmil..you don't post stupid things..else you would not have always so many views with your topic's.
It's not that people react on your messages that they don't read it..and in many cases your idea's get tried..
For sure i follow them

Kind regards,
JP.

Merci beaucoup, Jean-Paul, merci beaucoup!
Je suis enchante.
Cést un grand plaisir dávoir quelqún qui lit ce que tu ecris.

So, do I have many views? Where are the stats?

Concerning center levers, there is one more single detail you could do: giving a small bonus, only for the opening and middlegame, to a e4-d5 lever. The bonus could be half that for the more important c4-d5 and f4-e5 levers.

You might think there is complete symmetry here, a central pawn on the 4th rank attacking another central pawn on the 4th rank, but this is not so.

The symmetry is broken in terms of right-left asymmetry. The e4 pawn is more far-right than the d5 pawn, as the e file is more to the right than the d file. So, the e4 pawn gets the bonus, and d5 is penalised.

Please do not ask me why this is so. I can explain, but it is a difficult thing to understand. It has a lot in common to do with the fact that the kings are usually to be found castled on the king side. So you can not completely avoid asymmetry in square eval, on the contrary, it is very much advisable to introduce some kind of asymmetry, at least in pawn evaluation.

The case that e4-d5 lever favours white is solidly supported. e6-e5 breaks are typical in finding counterplay for the Queen's Gambit Declined. Upon advancing to e5, this pawn will usually attack an enemy white d4 pawn, and as the e5 pawn is closer to the king side, the e5-d4 lever, a mirror image of the e4-d5 lever, favours black.

You know particularly well that the Scandinavian, e4 d5, is a very bad opening for black, I would say altogether lost. And here you have black playing d5 already on the first move, creating an e4-d5 unfavourable lever.

Many lines in the Caro-Kann with such e4-d5 levers are known to favour white.

And look at the Scotch game, e4 e5 Nf3 Nc6 d4. The Scotch game is known not to give white any advantage, at least not the advantage white could have in other lines, and here white creates the unfavourable d4-e5 lever already on the third move.

So there is nothing accidental here.

[d]6k1/8/8/3p4/4P3/8/8/6K1 w - - 0 1
This one favours white, the e4 pawn

[d]6k1/8/8/4p3/3P4/8/8/6K1 w - - 0 1
and this one black, the e5 pawn

Of course, those are mainly opening/middlegame terms, and using a very long book to test changes might certainly backfire.

[Haikouichthys]

So, do I have many views? Where are the stats?

How about 'two successful ideas having made it into Stockfish in the past 10 days'?

[Lyudmil Tsvetkov]

So, do I have many views? Where are the stats?

How about 'two successful ideas having made it into Stockfish in the past 10 days'?

Those were just random, marginal ideas.
They do not want to implement the good ones.

[hMx]

So, do I have many views? Where are the stats?

Look here: http://www.talkchess.com/forum/viewforum.php?f=2 fourth column (titled "Views").
Maybe you knew that already.

Cheers, Heiner

[carldaman]

I see that no one is interested in my stupid posts, but I will take the freedom to post one last very brief message, as I find the topics worth it.

One simple rule for pawn checks:

- pawn checks are much less important than the threat of the check itself, in sharp distinction to all kinds of piece checks, where the check itself is more important than the threat of a check
- unsafe pawn checks are far more important than safe pawn checks, so a check not supported by any own pawn or piece is usually more valuable than a check supported by own pawns and pieces

So that, with pawn checks, everything works the other way.
The main reason for this paradoxical behaviour is that pawn checks are mainly a tactical weapon.

[d]6k1/7P/2k5/8/1PK5/5p2/8/1B4K1 b - - 0 1
The white check with the h7 pawn, although defended by an own bishop, is much less important than the threat of a check/potential check by the black pawn on f3 upon the white king, with f3-f2, although that check would be unsafe and unsupported.

As said, pawn checks are mainly a tactical weapon.

In the endgame, potential pawn checks like b4-b5, with the checking pawn supported only by the own king, are also important.

I would give potential pawn checks some 5-10cps bonus, both for the mg and eg, though the mg bonus might be a bit higher.


Passing over to the topic of center levers.

Some engines implement bonus for advanced levers, i.e. bonus for levers on the 5th and 6th ranks. This makes sense.

However, it is also possible to break the symmetry and use the concept also for levers on the 4th rank, that would be symmetrical in terms of ranks, but not symmetrical in terms of files.
Of course, you will try to give the bonus for an outer, less central pawn, attacking an inner, more central pawn. This is one of the suggestions of Mr. Kmoch.

Unfortunately, it will not work in most cases.
It works only in the center, and apart from that, there are only 2 squares on the entire board that deserve the bonus: a c4 white pawn attacking a d5 black pawn, and an f4 white pawn, attacking an e5 black pawn. Other tries might not work, but those 2 squares are very important.

Reasons for giving bonus for those particular squares?

- a c4 white pawn attacking a d5 black pawn might be traded for a more valuable, more central e6 black pawn, but also for less valuable c6 black pawn, that is on the same file, but less advanced, so that is not the valid reason
- a good reason is that it is always well to attack the center, the earlier the better
- another good reason is that the concept helps flank attacks like c2-c4 and f2-f4, which are very efficient, especially f2-f4, and particularly when made earlier

[d]6k1/8/8/3pp3/2P2P2/8/8/6K1 w - - 0 1
c4 and f4 pawns/center levers get the bonus here, in spite of the fact that the attacked enemy pawns are also on the 4th rank

I would give some 5-10-15cps, who knows, to center c4-d5 and f4-e5 levers, just for the middlegame.

OK, sorry for posting, I know no one is interested, but at least I am sure Carl will read it.

I usually read your posts, Lyudmil, as long as I have the time, and reply if something needs to be said, but not always, since time may be too limited. A lot of times, a non-response may actually signify tacit approval

I've always been a fan of central levers, even flank levers. In particular, I like the f4-e5 lever even more if
a) White has a pawn on e4 as well,
b) the f5 square is empty and thus f4-f5 is a viable option to extend the pawn chain, and
c) Black has already castled Kingside.

I think a hefty bonus of 30cps (or more) is in order if all 3 conditions apply to the f4-e5 lever. Gaining space with f4-f5 should be encouraged by further bonuses if the King is on that side, while a c4-c5 advance on the Queenside should be penalized if extending the chain in the wrong direction, away from the King.

Brief Regards,
CL

[PaulieD]

So, do I have many views? Where are the stats?

How about 'two successful ideas having made it into Stockfish in the past 10 days'?

Those were just random, marginal ideas.
They do not want to implement the good ones.

When I visit this forum I actively seek your posts and threads. I believe your insight is very valuable to current Stockfish development. Your input also comes from your very fine games and prowess versus the chess engines.
Your ideas are opening doors to new much needed positional play for Stockfish.
Merci for all you do...

[Lyudmil Tsvetkov]

I see that no one is interested in my stupid posts, but I will take the freedom to post one last very brief message, as I find the topics worth it.

One simple rule for pawn checks:

- pawn checks are much less important than the threat of the check itself, in sharp distinction to all kinds of piece checks, where the check itself is more important than the threat of a check
- unsafe pawn checks are far more important than safe pawn checks, so a check not supported by any own pawn or piece is usually more valuable than a check supported by own pawns and pieces

So that, with pawn checks, everything works the other way.
The main reason for this paradoxical behaviour is that pawn checks are mainly a tactical weapon.

[d]6k1/7P/2k5/8/1PK5/5p2/8/1B4K1 b - - 0 1
The white check with the h7 pawn, although defended by an own bishop, is much less important than the threat of a check/potential check by the black pawn on f3 upon the white king, with f3-f2, although that check would be unsafe and unsupported.

As said, pawn checks are mainly a tactical weapon.

In the endgame, potential pawn checks like b4-b5, with the checking pawn supported only by the own king, are also important.

I would give potential pawn checks some 5-10cps bonus, both for the mg and eg, though the mg bonus might be a bit higher.


Passing over to the topic of center levers.

Some engines implement bonus for advanced levers, i.e. bonus for levers on the 5th and 6th ranks. This makes sense.

However, it is also possible to break the symmetry and use the concept also for levers on the 4th rank, that would be symmetrical in terms of ranks, but not symmetrical in terms of files.
Of course, you will try to give the bonus for an outer, less central pawn, attacking an inner, more central pawn. This is one of the suggestions of Mr. Kmoch.

Unfortunately, it will not work in most cases.
It works only in the center, and apart from that, there are only 2 squares on the entire board that deserve the bonus: a c4 white pawn attacking a d5 black pawn, and an f4 white pawn, attacking an e5 black pawn. Other tries might not work, but those 2 squares are very important.

Reasons for giving bonus for those particular squares?

- a c4 white pawn attacking a d5 black pawn might be traded for a more valuable, more central e6 black pawn, but also for less valuable c6 black pawn, that is on the same file, but less advanced, so that is not the valid reason
- a good reason is that it is always well to attack the center, the earlier the better
- another good reason is that the concept helps flank attacks like c2-c4 and f2-f4, which are very efficient, especially f2-f4, and particularly when made earlier

[d]6k1/8/8/3pp3/2P2P2/8/8/6K1 w - - 0 1
c4 and f4 pawns/center levers get the bonus here, in spite of the fact that the attacked enemy pawns are also on the 4th rank

I would give some 5-10-15cps, who knows, to center c4-d5 and f4-e5 levers, just for the middlegame.

OK, sorry for posting, I know no one is interested, but at least I am sure Carl will read it.

I usually read your posts, Lyudmil, as long as I have the time, and reply if something needs to be said, but not always, since time may be too limited. A lot of times, a non-response may actually signify tacit approval

I've always been a fan of central levers, even flank levers. In particular, I like the f4-e5 lever even more if
a) White has a pawn on e4 as well,
b) the f5 square is empty and thus f4-f5 is a viable option to extend the pawn chain, and
c) Black has already castled Kingside.

I think a hefty bonus of 30cps (or more) is in order if all 3 conditions apply to the f4-e5 lever. Gaining space with f4-f5 should be encouraged by further bonuses if the King is on that side, while a c4-c5 advance on the Queenside should be penalized if extending the chain in the wrong direction, away from the King.

Brief Regards,
CL

Precisely, one of the assets of such c and f file levers is to possibly extend, under certain conditions, a longer chain, either on the king, or queen side.

Another positive side is that f4 or c4 pawns are supported by the respective bishops on c1/f1, so white might wait with the tension created by the lever, until the opponent captures and those bishops are swiftly developed. So the center lever of this kind is more burdensome to the opponent. For example, with an f4-e5 lever, e5 is not supported by a bishop, usually by an already developed knight, besides being more central pawn and less advantageous to trade.

I am also a fan of applying flank lever concept to other possible levers, for example b4-c5, supposedly bonussing the less central b4 pawn in this case. However, try to do do this and give such a bonus. I think it will not work in the general case, as it is not clear who benefits from the tension created by the lever. Sometimes the b4 pawn, and sometimes the c5 pawn.

Although the b4 pawn is less central than the c5 pawn, and thus advantageous to trade, it is not supported by either a bishop or well placed knight in the usual case, and also can not possibly extend any useful chain with b5, so this flank lever is entirely conditional upon specific circumstances on the board, that might arise or not arise. Thus, you do not know which side usually benefits.

Practical play also shows that flank b4-c5, b5-c4, g4-f5 or g5-f4 lever thrusts appear mainly in the form of b7-b5 thrusts in specific and limited in number lines of the Benko/Volga and Bluemenfeld Gambits, the Modern Benoni and very rarely in a few lines of the black KID.

With c4-d5/f4-e5 center levers you might claim with a fair amount of certainty that they are advantageous. But who knows what works in an engine and what not?