Pawn value estimation

[Vinvin]

Is there any scientific measurement that yields reliable pawn value estimation?

It seems that the standard methods are very good at showing the relative values of pieces (compared to one another), but not pawns.

By definition, the value of a pawn is 100cp.

I don't understand the point of your question. Simply pawn=100 (or rather pawn value endgame = 100) is a fixed point, and all other piece values (and any eval terms) are relative to that value.

The point is the values of pieces are much stable than the values of a pawn.
For example, a sample with pawn=100 and the same players with knight=300 :

Code: Select all

	      Anand 	Carlsen 	Andersen 
Pawn	  100		100		100
Knight	216		213		286
Bishop	230		243		289
Rook	  355		352		531
Queen	 762		786		1013
King	∞		∞		∞

Code: Select all

          Anand   Carlsen  Andersen 
Pawn        139		141		105
Knight      300		300		300
Bishop      319		342		303
Rook	     493		496		557
Queen      1058 	 1107	  1063
King	       ∞		 ∞		  ∞	

One can see that more values are much stable.

[abulmo]

By definition, the value of a pawn is 100cp.

This is a tautology, not a definition. Everybody knows the value of a pawn is a pawn or 100 centipawns.

I don't understand the point of your question. Simply pawn=100 (or rather pawn value endgame = 100) is a fixed point, and all other piece values (and any eval terms) are relative to that value.

Choosing the endgame (why not the opening?) pawn is a little bit more accurate definition. What I am missing, is in which kind of structure (passer, doubled, isolated, chained, none of that) is this pawn and on which square? Or, if you prefer, which psq value is worth zero for a pawn, and which structure value is worth zero, which mobility value for a pawn is worth zero, etc.
As the original poster Vladimir Medvedev ignores anything except the material value, it actually gives to the value a pawn an average psq value of zero, an average structure value of zero, etc. This may explain the great pawn value volatility observed by Dann Corbit, as the pawn value then depends on how a player uses its pawns.

[lucasart]

Is there any scientific measurement that yields reliable pawn value estimation?

It seems that the standard methods are very good at showing the relative values of pieces (compared to one another), but not pawns.

By definition, the value of a pawn is 100cp.

I don't understand the point of your question. Simply pawn=100 (or rather pawn value endgame = 100) is a fixed point, and all other piece values (and any eval terms) are relative to that value.

Engines tend to use other values (see, for instance Stockfish)
PawnValueMg = 198, PawnValueEg = 258,

Though pawns are generally reported to the user as 100 centipawns.

Be that as it may, I am talking about the ratio of the value of a pawn to the value of other chessmen.

The fitting methods arrive at very similar ratios for the value of any two pieces, using many different input data sets, if there are enough games.
However, the ratio of a pawn to the other chessmen varies wildly.

Let me make this even more clear, because you don't seem to understand it:
* centipawn: "centi" means 1/100, hence centipawn = pawn/100. It's like centimeter = 1/100 * meter, ok ?
* pawn = 100 * centipawn

It's completely tautologic. Your question is ill defined, really.

[Evert]

Pawns are like the photons of Chess: they have no rest mass!

I guess the interaction of the pawn with the other pieces is through Higgs bosons.

It's a bit unfortunate in a way that the nominal value of a pawn is used as a scale factor for the evaluation, given that the actual value of a pawn can differ substantially from this nominal value. When doing piece-value scalings or estimates, I usually fix R=500 instead.

[hgm]

I usually calibrate on Q=950. Although most engines rate a Queen a bit higher.

Perhaps we should switch to using milliQueens instead of centiPawns?

[Vinvin]

Perhaps we should switch to using milliQueens instead of centiPawns?

Very good idea too !

[michiguel]

I usually calibrate on Q=950. Although most engines rate a Queen a bit higher.

Perhaps we should switch to using milliQueens instead of centiPawns?

I used to use as a standard the Knight in Gaviota.
The Knight is more stable, and it is in the middle of the scale. Those are very good reasons to use it.

The atomic unit, which is supposed to represent the mass of a proton, it is not defined as the mass of the proton, but 1/12th of the mass of the carbon atom.

Miguel

[michiguel]

Is there any scientific measurement that yields reliable pawn value estimation?

It seems that the standard methods are very good at showing the relative values of pieces (compared to one another), but not pawns.

By definition, the value of a pawn is 100cp.

I don't understand the point of your question. Simply pawn=100 (or rather pawn value endgame = 100) is a fixed point, and all other piece values (and any eval terms) are relative to that value.

Engines tend to use other values (see, for instance Stockfish)
PawnValueMg = 198, PawnValueEg = 258,

Though pawns are generally reported to the user as 100 centipawns.

Be that as it may, I am talking about the ratio of the value of a pawn to the value of other chessmen.

The fitting methods arrive at very similar ratios for the value of any two pieces, using many different input data sets, if there are enough games.
However, the ratio of a pawn to the other chessmen varies wildly.

Let me make this even more clear, because you don't seem to understand it:
* centipawn: "centi" means 1/100, hence centipawn = pawn/100. It's like centimeter = 1/100 * meter, ok ?
* pawn = 100 * centipawn

It's completely tautologic. Your question is ill defined, really.

It does not work that way. A pawn unit is not necessarily "just" the value of a pawn. It is the value of a pawn in certain conditions. That is the question rephrased: "what conditions are those so the scale is robust and stable?". Even in science, the value of "something" seems to represent "something" but is defined in terms of something "else" that is more accurate to measure. The atomic unit, for instance, as I responded in another post.

Miguel

[hgm]

The atomic unit, which is supposed to represent the mass of a proton, it is not defined as the mass of the proton, but 1/12th of the mass of the carbon atom.

That reminds me of the new SI prefix jocto = 1e-24. That would make a joctoMole equal to 1.6 molecule. Now 1.6 is a bit of an inconvenient number. So a chemist proposed another prefix, 'guaca', defined such that a guacaMole would be exactly 1 molecule. For this, 'guaca' would have to be "the inverse of Avocado's number".

[Dann Corbit]

Does this matter ?

Yes. I want to know the answer to this question and I also want to know the reason for the answer.

Computers now beat the best human player,

Without a doubt.

so the current programs must have it pretty right.

Does not follow.