Tapered Eval in Fruit

[emadsen]

Okay, but why? Aren't 16 pawns on the board very middle-game and zero very end-game?

Because once most pieces have been captured, you want the king to march up the board to escort its pawns. Otherwise the king will hide in the corner due to MG PST and king safety eval terms. You don't want the king to wait until all the pawns are gone too. Then there's nothing to escort.

[hgm]

Okay, but why? Aren't 16 pawns on the board very middle-game and zero very end-game?

No, if there is nothing but Kings + 16 Pawns the game is very end-gamish. Kings are not in danger of being checkmated, and can roam the board freely. Zugzwang is a major motif.

Actually one can argue that Pawns should have a negative weight: the game phase can be used to create a trading incentive if piece values (and thus material advantages) are larger in the end-game. Smaller phase (e.g. due to an equal trade) then increases the advantage. But trading Pawns usually reduce the winning prospects, so that should increase the phase.

[Karlo Bala]

Okay, but why? Aren't 16 pawns on the board very middle-game and zero very end-game?

No, if there is nothing but Kings + 16 Pawns the game is very end-gamish. Kings are not in danger of being checkmated, and can roam the board freely. Zugzwang is a major motif.

Actually one can argue that Pawns should have a negative weight: the game phase can be used to create a trading incentive if piece values (and thus material advantages) are larger in the end-game. Smaller phase (e.g. due to an equal trade) then increases the advantage. But trading Pawns usually reduce the winning prospects, so that should increase the phase.

Unfortunately, there is no universal solution. If you play against isolated or backward pawns, you want to trade all minor and keep all major pieces on the board. If you play against doubled pawns, then you should probably trade major pieces first.

[Pio]

Okay, but why? Aren't 16 pawns on the board very middle-game and zero very end-game?

No, if there is nothing but Kings + 16 Pawns the game is very end-gamish. Kings are not in danger of being checkmated, and can roam the board freely. Zugzwang is a major motif.

Actually one can argue that Pawns should have a negative weight: the game phase can be used to create a trading incentive if piece values (and thus material advantages) are larger in the end-game. Smaller phase (e.g. due to an equal trade) then increases the advantage. But trading Pawns usually reduce the winning prospects, so that should increase the phase.

I have pawns as negative value. The main reason for this is that with many pawns the king is usually a lot safer and should be more involved earlier on