self-sorted arrays
Sorting as a collective behaviour
Self-sorted arrays show how global order can emerge from simple local rules. Each element is an autonomous cell following a basic sorting algotype (bubble, insertion, or selection) and acting only on its neighbours, with no central view of the array. Order arises bottom-up through distributed swaps.
Sorting as agency
The framing follows Michael Levin and colleagues' work treating classical sorting algorithms as agential material: reframed as populations of agents, they show unexpected robustness, clustering, and delayed-gratification behaviours a monolithic algorithm never displays. The cells here can also mutate, convert, divide, and aggregate.
The calibration panel checks the exact order metrics (sortedness and aggregation); the assumptions panel separates those metrics from the basal-cognition interpretation.
Model changelog
- array elements modelled as autonomous cells that swap with neighbours by local rules.
- per-cell algotypes (bubble, insertion, selection), swap policies, and mutation strategies.
- order metrics: sortedness (fraction of increasing adjacent pairs) and aggregation (fraction of like-algotype adjacent pairs).
- fixed the sortedness metric to compare cell values rather than cell objects.
- calibration checks sortedness and aggregation against hand-built distributions with known values.