Evaluating AI's 'Non-Linear' Sacrifice Patterns

The Logic Behind Algorithmic Sacrifice

Modern AI (such as Katago) frequently employs sacrifice patterns that defy classical 'shape' intuition. These non-linear sacrifices often involve dumping stones that seem valuable in isolation to gain superior thickness or to disrupt the opponent's overall structure.

• The Geometry of Sacrifice: AI evaluates the board in terms of 'expected future value.' A stone that appears to be a 'cutting' stone might be discarded if its removal improves the efficiency of your surrounding groups.
• Analyzing Efficiency Gains: When AI suggests a sacrifice, focus on the 'potential' versus the 'actual.' If the sacrifice eliminates an opponent's potential base, the gain in global efficiency far outweighs the loss of 2-3 stones in the corner.
• Professional Training Drill: Select a complex joseki where AI suggests a sacrifice. Play the move, then toggle the engine to 'Human mode' to compare the evaluation difference. Repeat this 20 times with varying patterns to internalize the AI's valuation logic.

The primary error for players is the 'sunk cost fallacy.' Holding onto heavy stones during a fight is a common losing condition. Learn to recognize when a stone's function has been exhausted and use it as a probe or sacrifice to maintain sente.