A topic in the Open Knowledge Graph — a free, open map of 15,290 topics and the order to learn them in.
Stochastic composition uses probability distributions to generate or organize musical material. Rather than deterministic rules, composers like Xenakis used Markov chains, Poisson distributions, and other probabilistic models to create complex musical sequences that balance structure with apparent randomness.
The key insight behind stochastic composition is that probability distributions produce statistical *shapes* — and shapes are perceivable. If you draw pitches uniformly at random from the chromatic scale, the result sounds chaotic and undifferentiated. If you use a Gaussian distribution centered on middle C with a narrow standard deviation, the pitches cluster around middle C with occasional outliers — you hear something that fluctuates around a center. If you use an exponential distribution for note durations, you get many short notes and rare long ones. Iannis Xenakis, the central figure in this approach, recognized that by choosing distributions deliberately, a composer does not surrender control — they delegate it to a defined probabilistic process whose statistical character is entirely predictable, even when the individual events are not.
Markov chains add memory to this picture. From your prerequisites, you know that a Markov chain defines transition probabilities between states: given the current state, the probabilities of all possible next states are fixed. In a compositional Markov chain, states might be pitch classes, rhythmic values, or timbres, and the transition matrix encodes musical grammar. A matrix that makes neighboring pitch classes likely produces stepwise melodic motion; a matrix with equal probability to all states produces random leaps. The chain can be designed to favor cadential progressions, to avoid repetition, or to navigate between tonal centers according to a statistical "grammar" that the composer specifies. This differs from both deterministic rule-based composition and pure randomness: the chain has a characteristic *style* defined by its transition probabilities, even though individual outputs are unpredictable.
Xenakis formalized the macro-level use of stochastic processes in his "stochastic music" works, using the Poisson distribution to control the density of sonic events per unit time and the Gaussian distribution for pitch clouds. His piece *Pithoprakta* (1956) distributes bowing gestures across a string orchestra by treating each instrument event as a particle in a statistical ensemble — the score was generated by mapping physical probability models onto musical parameters. The listener hears a constantly shifting texture of density and register rather than melodic lines, because the musical material is defined at the level of the statistical ensemble, not the individual voice.
The tension between structure and surprise is stochastic composition's central aesthetic claim. A purely deterministic piece is fully predictable to anyone who knows the rules; a purely random piece has no pattern to perceive. Stochastic processes occupy the space between: they have a definite character (the distribution's shape, the Markov chain's tendencies) that gives the music a recognizable identity, while their randomness ensures the specific unfolding is always fresh. This connects to minimalism's interest in process-over-result — like phase music, stochastic works make the generative procedure itself compositionally legible — but replaces deterministic phase relationships with probabilistic ones. When analyzing stochastic music, describe the process (what distribution? what parameters?) and the perceptual result (what texture, density, and character does it produce?) before asking how that character serves the work's larger formal arc.
Topics in reflective domains aren't scored by quiz answers. Read, reflect, and mark when you've thought it through.