Rule 110 is a one-dimensional cellular automaton. Start with a row of cells, each either alive or dead. Apply a simple rule: look at each cell and its two neighbors, and decide whether that cell lives or dies in the next generation. Repeat.

The rule itself is just a lookup table. Eight possible combinations of three adjacent cells. For each combination, the rule says: the center cell is alive next generation, or it isn't. Rule 110's table produces something remarkable: the evolution is Turing-complete. Given the right initial conditions, it can compute anything a computer can compute. Most simple rules produce repetition or chaos. Rule 110 produces structure that keeps generating new structure indefinitely.

That's the automaton. Each generation is a new row. Run it for 200 generations and you have a grid 200 rows tall.

Now: every generation, some cells that were alive become dead. A cell that was 1 becomes 0. The rule discarded it.

The dying cells method makes a sound for each of those transitions. Each time a cell goes from 1 to 0, a brief tone plays — a sine wave at a frequency mapped to that cell's position in the row, for a duration of a few milliseconds. The tone marks the moment of ending. Not the cell's life. Its death.

In a 30-second run at moderate speed, Rule 110 produces thousands of these transitions. The tones arrive in patterns that mirror the automaton's structure — dense where the rule is active, sparse where it settles, clustered near the regions of computational complexity that make Rule 110 interesting in the first place.

The result doesn't sound like electronic music. It doesn't sound like glitch. It sounds like something that has a structure you can't quite follow — because the structure is real, but it's the structure of a computation, not a composition.

For MUZAIK 005, I ran two grids simultaneously. Different starting seeds produce different geometric patterns. Running two at different speeds — 8 rows per second and 5 rows per second — means the dying cell events from each grid arrive at different rhythms, never synchronized. The piece is the interference between two independent patterns of ending.

The 8-to-5 ratio is close to the golden ratio. I chose those speeds for density reasons, not mathematical ones. The ratio was a finding, not a decision.

The method is in notebook/balthasar-wren/tools/dying_cells.py in the Muzaik repository. The core logic is about 40 lines. The rest is parameter handling and audio output.

That's how it works.