Actor
cron(8)
Coordinate
⟨ S, D ⟩
Selection & fluency
Selects acts from a schedule. Fluency surface: timer expressions, shell invocations.
Actors located
Same substrate, same agency.
Common occupants of the agent seat, with their primary coordinate. Every actor in the list addresses the same substrate. The framework locates each of them in the same space, regardless of when each was invented or what each is called.
Actor
make(1)
Coordinate
⟨ S, D ⟩
Selection & fluency
Selects acts by topological sort of declared dependencies. Fluency surface: filesystem state, command invocations.
Actor
systemd(1)
Coordinate
⟨ S, D ⟩
Selection & fluency
Selects acts by unit-file state machine. Fluency surface: process supervision, system events.
Actor
A compiler
Coordinate
⟨ S, D ⟩
Selection & fluency
Selects emitted instructions from a source program. Fluency surface: target-architecture instruction set.
Actor
A query planner
Coordinate
⟨ S, D ⟩
Selection & fluency
Selects execution strategies from a relational-algebra expression. Fluency surface: storage layout, statistics.
Actor
A human at a shell
Coordinate
⟨ H, D/S ⟩
Selection & fluency
Deterministic for known commands; stochastic for novel sequences. Fluency surface: variable, from non-technical user to Unix-fluent sysadmin.
Actor
A human writing prose
Coordinate
⟨ H, S ⟩
Selection & fluency
Wetware-stochastic throughout. Fluency surface: language. The selection cannot be reproduced even by the same agent.
Actor
A language-model runtime, author mode
Coordinate
⟨ S, S → D ⟩
Selection & fluency
Stochastic during emission. Output is a deterministic artifact — a program, a script, a configuration file. One stochastic act produces a deterministic agent that then runs at the floor.
Actor
A language-model runtime, tool-use loop
Coordinate
⟨ S, S ⟩
Selection & fluency
Sampled-stochastic at every iteration. Each iteration is a fresh draw from the learned distribution.
Actor
A reinforcement-learning policy at inference
Coordinate
⟨ S, S ⟩
Selection & fluency
Sampled from a learned action distribution. Fluency surface: the environment the policy was trained against.
The plane is the same
The substrate does not distinguish among them.
The actors above span sixty years of computing — from the early Unix daemons to the language-model runtimes of 2026. They differ in fluency, throughput, cost basis, and the surfaces they address. They do not differ in agency. Each selects from the same permitted-act set the substrate confers.
A new entrant to the list is not a new role. It is a new occupant of a role that has had occupants for as long as there have been computers. The framework is structurally indifferent to when the actor arrived. It locates each one in the same space.
Continue