The Periodic Stack of State

Abstract

A running system holds state. The shape of that state — how it is modeled, how it composes, how transitions are advanced, where the boundary around it lies, what the machine knows of itself, what each transition costs, and what can be checked of it — is a finite set of independent design commitments. The tuple of those commitments is a coordinate, and a coordinate predicts what the system can express, what it can verify, what survives a restart, and what it pays in joules per step. This note defines nine axes that span state, the discrete coordinate space they form, the structural constraints that prune most of it, and the consistency that ties this stack to its siblings.

1 · Position and lineage

The genealogy of state in computing is unusually well-formalized. Turing’s machine made state explicit in 1936. Finite automata gave it the simplest formal shape. Petri introduced the first formalism for concurrent state in 1962. The actor model arrived in 1973 and turned state into an architectural boundary. Zeigler’s DEVS gave simulation a formal account in 1976. Hoare’s CSP and Milner’s CCS gave concurrent state an algebra by 1980. Harel published statecharts in 1987 — hierarchy, orthogonality, broadcast — and gave the field its language for compositional state. Lamport’s TLA arrived in the 1990s and put temporal reasoning on a model-checkable foundation.

Each of these contributed a vocabulary; none of them organized those vocabularies into a single coordinate space. This stack is that organization: a finite set of independent axes that places every formalism, and every running system that instantiates one, at a measurable position.

2 · What state is

State is the part of a running system that persists between transitions and constrains what the next transition can be. Five consequences follow:

• It has a model. Whether a finite-state machine, a statechart, a Petri net, a process calculus, an ECS table, an actor mailbox, a behavior tree, or a TLA⁺ specification — every running system commits to a formal model that decides what is expressible.
• It composes. Atomic states stand alone; compositional states are built. The compositional structure is its own axis: sequential, AND-parallel, OR-hierarchical, dynamic-restructurable.
• It has a boundary. State is either shared, encapsulated, message-passed, process-isolated, or machine-isolated. The boundary is the architecture.
• It advances. A transition runs on a clock, an event, a continuous stream, or the system’s own self-model. What drives the step is a design choice, not a property of the universe.
• It can be seen, or not. Reflection — what the system knows of itself — is a structural axis. Opaque states behave; reflective states reason about how they behave.

3 · The nine axes

Each axis is finite and independent. The full enumeration with value-by-value descriptions is the subject of its own page.

• F — Formalism. The model the state is held in.
• K — Composition. How simple states assemble into larger ones.
• C — Concurrency. How transitions advance in time.
• I — Isolation. The boundary around state.
• R — Reactivity. What drives the step.
• A — Self-reference. What the machine knows of itself.
• T — Thermodynamic class. The cost floor of a transition. Inherited from the Periodic Stack of Computation.
• V — Verifiability. What can be checked of a transition. Shared with synthesis and data.
• τ — Lifetime. How long the state lives, and what it survives.

Read the nine axes in full →

4 · The coordinate space

The product of the axis cardinalities — 10 · 5 · 5 · 5 · 4 · 4 · 4 · 5 · 5 — is 2,000,000 discrete cells: the addressable map of state-space. Structural constraints prune most of it; what remains is the admissible map on which every running system has a place.

5 · Impossibility analysis

Constraints between axes eliminate large regions of the coordinate space:

• Composition × concurrency. AND-parallel composition (K₂) requires a concurrency model that supports it (C₂ preemptive or higher). Atomic composition (K₀) is incompatible with autonomic reactivity (R₃) — there is no internal structure for the system to reason about.
• Isolation × concurrency. Shared-memory isolation (I₀) is incompatible with distributed concurrency (C₄) by definition.
• Self-reference × verifiability. Reflective state that modifies its own structure (A₃) cannot be statically model-checked end-to-end (V₁) without further structure; the running system must carry its own checkable invariants.
• Formalism × composition. An ECS (F₅) does not natively compose OR-hierarchically; a behavior tree (F₇) does not natively compose AND-parallel without explicit constructs. The formalism determines which compositions are idiomatic.
• Lifetime × isolation. Ephemeral state (τ₀) cannot be durable across machines (need I₃ or I₄ for τ₃+ to be meaningful).

Applying these prunes the space to a structured remainder. A sparse cell flanked by occupied cells — the Mendeleev face — is a prediction. The predictive face presents ten such cells.

See the predicted cells →

6 · Cross-stack consistency

Two of the nine axes are shared with the sibling projections, and the shared axes must agree:

• T is the thermodynamic axis of the Periodic Stack of Computation. A transition’s cost class is the dominant class of the compute primitives it invokes; the state-system T must match the compute-stack T of its transition kernel.
• V maps to the verifiability axis of the synthesis and data stacks — what can be checked of an answer, of a record, of a transition. The same five values appear in each.

Because the shared axes agree, a single primitive carries consistent coordinates across all of the stacks. They are projections of one structure, not separate taxonomies.

7 · Using the stack

As a design tool: name the coordinate before naming the implementation. A choice of cell predicts what a design can be good at — and what it cannot — before a line of code is written.

As an audit tool: read a running system across the axes. Where the coordinate disagrees with the system’s claims about itself, the impedance is the gap. Most production state systems sit at a coordinate softer than the one they advertise.

As a research tool: a sparse cell whose flanking cells are all well-occupied is a target. The predictive face is the first-pass enumeration; the structure generates more.

8 · Summary

Nine axes span a discrete space of 2,000,000 cells. Structural constraints prune it to an admissible remainder on which every running system has a place and the sparse-but-flanked cells are predictions. The method classifies how state is held, makes the design space legible, and lets a running system be reasoned about before it is built — position predicting behavior, which is what a periodic structure is for.

References & citations → · A genealogy of state in computing →