The worst pawn weakness

[Lyudmil Tsvetkov]

It would be interesting to know your opinion about which pawn weakness you consider as the worst. What does your practice tell you?

In my humble opinion, based on the games I have played, I have never encountered a more detrimental pawn weakness as the double horizontally isolated backward-fated pawn.

[d]4k3/1p1b2p1/5p1p/1p1p4/3N4/P1P5/1P3PPP/4K3 w - - 0 1

b5 and b7 represent such a pawn. I think its penalty should be huge, do not know how much exactly, but really huge (maybe a full pawn, or at least 0.90-0.95cps). There is nothing creative to do with such a pawn, probably just sacrifice it, so that it does not stand in the way of the own pieces.

Maybe the only pawn that beats it in terms of detrimental qualities is the 'blocked double horizontally isolated backward-fated pawn'

[d]8/1p1b2p1/1k3p1p/1p1p4/1K1N4/P1P5/1P3PPP/8 w - - 0 1

Same diagram, but the unfortunate black b pawns are already even blocked by the white king (but better by a minor piece), so that even sacrificing b5 to attempt doubling enemy pawns becomes impossible. Maybe the penalty is already exactly a full pawn, or, in some cases where the pawn thwarts the friendly pieces activity, even larger than a pawn (might seem like a paradoxical statement, but for some game spans it is not).

And another diagram to highlight the importance of backward pawns.

[d]4k3/1p1b2p1/5p1p/1p1p4/1P1N4/P1P5/5PPP/4K3 w - - 0 1

The position is exactly the same as in the first diagram, the only distinction being that the b5 pawn is now already blocked by b4. With this arrangement, doubleness and isolation for b5 remain, but backward-fatedness suddenly disappears, which, of course, favours black. And indeed, I think the position here is a draw (or a likely draw), while the position on the first diagram should be easily won for white.

Interesting to know what penalty you would assign to b5 in both cases.

What is the biggest pawn weakness in your experience?

Best, Lyudmil

[Lyudmil Tsvetkov]

Other obvious examples that come to mind:

[d]8/6k1/7p/7p/7P/8/6K1/8 w - - 0 1
The double pawns part of symmetrical structures (but less than a full pawn, maybe 80cps penalty)

[d]8/5pk1/3p3p/5p2/5PpP/3P2P1/6K1/8 w - - 0 1
Same, but with more pawns

[d]6k1/3p4/3P4/8/8/8/8/6K1 w - - 0 1
Only in the opening and early middlegame, isolated pawn fixed/blocked by an enemy pawn on the 2nd rank/initial position, meaning that own piece development is thwarted considerably; penalty might go as high as 1.5-2 pawns, but I think this is not a proper pawn weakness and has to do more with development/piece activity.

[d]6k1/3p4/3p4/3P4/8/8/8/6K1 w - - 0 1
Very much similar, have not thought which penalty should be bigger, double isolated pawns on the 2nd and 3rd ranks (one of them initial), fixed/blocked by an enemy pawn. Again, huge penalty (1.5-2 pawns), but has to do more with development, not a real pawn weakness

[d]8/3p4/3p4/3p4/3p1k2/3p4/3P1K2/8 b - - 0 1
One might also think of unrealistic examples as this one: a quantupled pawn (with penalty, of course, going as high as 3-4 pawns), but this is more in the realm of science fiction.

Interesting btw., could black win here?

[Lyudmil Tsvetkov]

Well, if no one is going to post, my impression will be that this is the worst pawn weakness one can get.

[neelbasant]

Well, if no one is going to post, my impression will be that this is the worst pawn weakness one can get.

Friend All position except last one are known pawn weaknesses.

I don't know how many times the last one position has been reached.

[Uri Blass]

stockfish does not see a win in the first position
score at depth 55 is 1.21 but it is not better than the score at depth 47-54 that was 1.23 so I get the impression that maybe it is a draw

Here is the line of stockfish at depth 55 when I used a single cpu for the analysis

[d]4k3/1p1b2p1/5p1p/1p1p4/3N4/P1P5/1P3PPP/4K3 w - - 0 1

Stockfish_13091318_x64_modern_sse42:


55/83 4:00:46 30,230,402,820 2,092,647 +1.21 Ke1-d2 Ke8-e7 Kd2-e3 Ke7-d6 Nd4-b3 Bd7-g4 f2-f3 Bg4-f5 Ke3-d4 Bf5-d7 Nb3-c5 Bd7-c6 Nc5-d3 Bc6-e8 Nd3-f4 Be8-f7 Nf4-e2 Bf7-e8 Ne2-f4

[Lyudmil Tsvetkov]

stockfish does not see a win in the first position
score at depth 55 is 1.21 but it is not better than the score at depth 47-54 that was 1.23 so I get the impression that maybe it is a draw

Here is the line of stockfish at depth 55 when I used a single cpu for the analysis

[d]4k3/1p1b2p1/5p1p/1p1p4/3N4/P1P5/1P3PPP/4K3 w - - 0 1

Stockfish_13091318_x64_modern_sse42:


55/83 4:00:46 30,230,402,820 2,092,647 +1.21 Ke1-d2 Ke8-e7 Kd2-e3 Ke7-d6 Nd4-b3 Bd7-g4 f2-f3 Bg4-f5 Ke3-d4 Bf5-d7 Nb3-c5 Bd7-c6 Nc5-d3 Bc6-e8 Nd3-f4 Be8-f7 Nf4-e2 Bf7-e8 Ne2-f4

I don't know, Uri.
My experience is that Stockfish in general plays extremely well relatively simple endgames, so I am a bit afraid to take a categorical statement. However, what about trying to get this position instead:

[d]8/1p4p1/3kbp1p/1p1p4/3K1P2/P1P1N1P1/1P5P/8 b - - 0 1

Should be feasible, all white pawns on black squares, knight on e3 and king on d4. Do you think Stockfish with black could defend that?

[jdart]

I only give a modest penalty for doubled pawns, but more if isolated, and even more if on an open file. Even so the penalty is not going to approach -1 pawn. Empirically, more penalty does not help.

I think in your example though Black also benefits by having Bishop vs. Knight. It is well known that with pawns on both sides of the board, the Bishop tends to be better, at least if it is not blocked and has mobility.

But all these factors are generalizations. In the current state of the art programs have to choose evaluations that work "on average" across millions of similar positions.

--Jon

[PK]

I suppose that doubled pawn is not necessarily a weakness per se. Penalizing it works because more often than not a doubled pawn indicates that more subtle structural weakness exists somewhere else - be it an isolated pawn, a blocked piece or pawn majority on the other wing. Probably sufficiently robust and perfectly tuned evaluation function wouldn't need a penalty for doubled pawns.

[jdart]

Doubled pawn by itself is not a serious weakness IMO. In the Nimzo-Indian, White gets doubled pawns all the time and is not necessarily worse.

[Lyudmil Tsvetkov]

I suppose that doubled pawn is not necessarily a weakness per se. Penalizing it works because more often than not a doubled pawn indicates that more subtle structural weakness exists somewhere else - be it an isolated pawn, a blocked piece or pawn majority on the other wing. Probably sufficiently robust and perfectly tuned evaluation function wouldn't need a penalty for doubled pawns.

Hi Jon and Pawel.

I respect your opinions, but would agree with neither of you.

If you do not evaluate double pawns, what else more important could you evaluate?
The less material on the board, the more evident the importance of even least detrimental double pawns (part of a larger group, no symmetrical structures, etc.) becomes. In simple pawn endgames it is actually very difficult or almost impossible to encounter a situation where a single double pawn would not seal the fate of the side having it. One should start with pawn endgames, and then expand the knowledge. Thus, claiming that double pawns are unimportant seems at least an excentrical idea to me.