It begins with a loop. In generative music, this figure is both constraint and invitation—a short phrase, a probability field, perhaps, or a set of rules articulated in Max/MSP or TidalCycles. If we consider Brian Eno’s Music for Airports, the iconic tape loops measuring precisely 23½ seconds, we see not just machinery but intention—restricted movement creating possibility. In digital domains, a random walk function might determine pitch or filter cutoff, where, for example, a Markov chain restricts movement to paths today’s code deems plausible.

Generative structures rely for their lifeblood on the interplay between determinacy and drift. Even the weighting of events—one note at 60% likelihood, another at 20—bespeaks a world micromanaged and yet allowed to surprise itself, much like the apparatuses of government comprise both rigid systems and unwieldy personalities. A set of conditionals, coded as if-else in SuperCollider, proves that music is less a text inscribed than a protocol unfolding in real time.

Listeners sometimes remark on the stasis, the sense that nothing ‘happens.’ But such music is, in fact, anything but static. Each moment is a speech act of sorts: weighted, contingent, adjusted by the smallest increments—be it in the decay time of a reverb, expressed as milliseconds, or the randomisation of a delay line by fractions. The generative approach thus resists central planning and revels in limited agency, forcing both creator and audience to trade certainty for fertile anticipation.