Heuristic algorithms pair computational power with expert judgment to guide composition, unlike borrowed formal models

Roads proposes a hybrid formal/informal approach that combines the computational power of algorithmic control with heuristics — ‘judgmental knowledge, the knowledge that comes from experience, the rules that make up the art of good guessing.’ Unlike formulas imported whole-cloth from physics or mathematics, heuristic algorithms are domain-specific, context-dependent, tested by experiment and refined by human perceptual judgment. Xenakis used stochastic processes heuristically, modifying and rearranging results to suit the piece. For granular synthesis this means borrowing concepts from scientific models of granular processes but reworking them so the physical model serves as a metaphor for musical organization rather than a strict simulation.

Examples

Harold Cohen’s forty years of highly specific heuristic drawing algorithms that do not try to draw like a human; Xenakis rearranging stochastic output; using a granular-physics model as a musical metaphor, not a literal simulation.

Assessment

Distinguish a formal algorithm from a heuristic algorithm in composition, and describe one way to apply heuristics to improve a purely generative granular process.