The Death of Contempt

Chess, Risk-Averse Engines, and the Art of Never Winning

The rise of neural-net engines has produced a paradoxical “net result”: not just stronger play, but an epidemic of draws. The cause is an insidious shift in solving logic, where the network is trained to avoid loss at all costs, not to embrace risk for victory.

This evolution is epitomized by the removal of the “Contempt” parameter. Once a lever to inject human-like ambition, sacrificing a few superficial Elo points to play for wins, its disappearance is symbolic. We have optimized the soul out of the algorithm. In the pursuit of flawless, binary efficiency, the draw has been redefined as the optimal “participation award,” and romantic complexity has been systematically engineered out.

The consequence is a lifeless, algorithmic pragmatism that insists we all sprint down standardized openings to a pre-ordained drawing margin: a conclusion dictated not by a chess oligarchy, but by a risk-averse neural network whose ultimate logic is to never lose, even if it means never truly winning.

So be it. If the neural net’s “strength” is a preference for clean, drawn certainty, then its weakness is a fear of the messy and profound. I will exploit that. Let engines polish their drawing lines; I will forge positions that demand more than an evaluation bar can give.

The goal is simple: force the machine into the abyss of true complexity, where its quick-study arrogance falters. There, in the silence of its deepening search, the romance of the game, and the human edge, returns.