/ writings ·

Every track is a tiny operating system

Composition is design under the deepest constraint a builder can take on: time you can't pause.

By Abhishek Krishna

Tempo, key, and texture are state. Surprise is the only metric. An audience that can quit any second is the cleanest user-research panel a builder can have.

People who know me from software ask, occasionally and politely, why I keep making music. The honest answer is that they’re the same job.

I don’t mean that metaphorically. I mean that on most days I cannot tell which one I’m doing.

What a track actually is

A track is a finite-resource system designed to hold a listener’s attention across a bounded interval. It runs on a clock you cannot stop. The state is small (tempo, key, texture, dynamics, timbre), but it changes constantly, and the user can quit at any moment.

That is, structurally, a piece of software. The constraints are sharper than most code I’ve shipped:

  • Latency budget: the listener decides whether to keep going inside the first eight seconds.
  • Memory model: working memory is about three to five elements; if you put more in the mix, something else has to drop out.
  • Error budget: surprise is a feature; unexplained surprise is a bug.
  • Compatibility: the track has to render on phone speakers, car stereos, studio monitors, and earbuds, without ever telling the listener which one it was tuned for.
Fig. 04 · Attention budget for a four-minute track peak silence 0:00 1:00 2:00 3:00 4:00 +8s · ship-or-skip BUILD HOLD RELEASE motif in tension peak surprise window resolve
A track's macro envelope is the same shape as a well-designed product loop: a ruthless first interaction, a tension build, a sustain at peak, and a deliberate release. The mint hairline is the latency budget. If the listener isn't onboarded by then, the rest of the curve never plays. Working backwards from that constraint changes how you write, the same way it changes how you ship.

You don’t get to ask the user to read the docs.

The craft transfer

The instinct that helps me ship products is the same instinct that helps me ship music. Three things, in particular:

Editing is the actual job. What you cut from the arrangement is more important than what you write into it. Same with code. Same with a deck. Same with a paragraph.

Tension is what audiences pay for. Build it, hold it, release it. Any track that doesn’t do this dies; any product that doesn’t do this never starts.

Constraints are gifts. Eight bars. Six minutes. Four chords. One key change. The same is true of seed budgets, deadlines, and team sizes. Constraint is the soil quality is grown in.

Why it shows up in the work

Some of this is autobiography. Muzix Tech was the music platform Ojam, which grew to a 100K-person community before we wound it down. Wild Monks was the live era. Abhi Cris is the present tense: tracks released as software releases, versioned, dated, public.

Some of it is method. The strongest products I’ve worked on have all started from a single repeated motif (a phrase, a behavior, a feeling) that the rest of the system was designed to support. That is exactly how a track gets written. You find the figure that has to exist, then you build the smallest possible engine that lets it exist.

The reason I bring it up

Founders are often advised to specialize, to be legible, to pick a lane. The advice is not wrong, but it’s incomplete. Some lanes are the same lane.

If a discipline trains the same muscle as the work (same constraint shape, same feedback loop, same audience trust), then doing both is not a distraction. It’s compounding.

The shipping cadence I trust most is the one I built making music. The audience research I trust most is the one I built making music. The taste I bring to product decisions came, in large part, from making music. None of this is a side project. It is the same project, in a different format.