Compensation

Compensation is what a side gets for material that has been given up. A sacrifice that wins immediate mate needs no compensation in the abstract sense — the material returns with interest by force. A sacrifice that does not immediately win must produce positional gains: an exposed king, active pieces, a passed pawn, weak squares in the enemy camp. These gains are compensation.

Compensation is evaluated qualitatively, not just by counting pieces. A pawn sacrificed for two active bishops and an open file may be excellent compensation; the same pawn for a passive piece arrangement is bad compensation. Strong players develop an intuition for what compensation is sufficient and what is not — an intuition often summarised in opening literature as sound or unsound sacrifices.

Engines have changed how compensation is judged. Lines that human masters once played for compensation are now sometimes shown to be objectively bad — the compensation was real but insufficient. Other sacrifices, considered dubious for decades, have been rehabilitated. The Petrosian Variation in the King’s Indian and several Marshall Attack lines in the Ruy Lopez have seen their evaluations move in both directions.

The word compensation is also used to describe positional concessions for non-positional advantages. Doubled pawns are a structural concession; the half-open file beside them is the compensation. The bishop pair is compensation for a slightly worse structure. Most middlegame decisions involve a small trade of one kind for another.