Activity asks: did we do it?
It helps coordinate effort, ownership, time, and sequence.
Reality / transition / proof
State.
State is what is verifiably true now. Work records effort. State records the change in reality that effort produced—and what reality can support next.
ActivityWe installed the pump.
StateThe pump is installed, inspected, and ready for integration.
Distinction
Tasks say what happened. State says what changed.
A completed task can leave the system unready. A shipped feature can leave customer behaviour unchanged. A signed contract can leave value unproved. Activity matters, but it is not the same as progress.
State asks: what is true now?
It exposes readiness, evidence, dependencies, and the next safe move.
Time
A state is true at a named time.
State and time cannot be separated. A customer was satisfied when asked. A system was ready when inspected. A capability existed under the conditions that were tested. Without an observation time and freshness window, yesterday's truth can silently become today's fiction.
Observed state
What evidence supports now, including confidence, conditions, and freshness.
Desired state
What someone wants to be true at a future horizon. It is a setpoint, not an observation.
A state is not a permanent identity. “I cannot do this yet” describes present capability. It does not prove what the person can become.
System
A business is a nested state machine.
Operations repeat transitions the business already understands. Projects attempt to change those transitions or expand what the business can reliably do. Both depend on many smaller systems reaching compatible states at the right time.
Desired
The outcome and protected constraints are explicit.
Gate: An accountable person accepts the intended state.Designed
The system, dependencies, and acceptance tests are known.
Gate: The design can be inspected before resources are committed.Integrated
The parts work together under expected conditions.
Gate: System-level evidence replaces local completion claims.Operating
The system repeatedly produces value for its beneficiary.
Gate: Observed outcomes survive the agreed review window.
Counterclaim: not everything valuable fits a clean state model. Discovery is uncertain. Human judgment is contextual. A false state is worse than an honest unknown. Use states to expose reality, never to force reality into a convenient status.
Coordination
Proof must open the transition.
A label alone does not make a state true. Each consequential transition needs evidence proportionate to its risk. Local completion is not system readiness: installed is not integrated; integrated is not commissioned; launched is not valuable.
Machine-readable state helps people and agents see the same reality, block premature moves, and coordinate dependencies. Human authority still owns contested evidence, protected values, irreversible commitments, and changes to the desired state.
Working claim: visible, evidence-gated state improves coordination when it reduces hidden dependencies, premature transitions, or disagreement about readiness.
Claim→Case for→Case against→Narrow→Test↺
Builder
ForClear states expose dependencies and tell each contributor what can happen next.
AgainstMaintaining the model can become work about work.
ResolutionModel only consequential handoffs, not every movement.
Skeptic
ForA falsifiable state is stronger than a confident progress report.
AgainstEvidence can be incomplete, stale, selected, or wrong.
ResolutionAttach the source, time, confidence, and falsifier.
Humanist
ForShared truth can reduce blame, ambiguity, and dependence on hierarchy.
AgainstPeople, care, consent, and judgment do not fit neatly into enums.
ResolutionKeep values and contested judgments under human authority.
Systems thinker
ForNested state exposes how local completion affects system readiness.
AgainstA crisp local model can hide second-order effects outside its boundary.
ResolutionName the system boundary and inspect effects beyond it.
Operator
ForQueryable state helps humans and agents coordinate without another meeting.
AgainstUnder pressure, the record can lag behind reality.
ResolutionGive each consequential state an owner and freshness window.
Position after pressure: make state explicit only where a shared, consequential transition needs inspectable evidence. Keep uncertainty visible. Model the boundary. Let later outcomes decide whether the control deserves to stay.
Next action
Replace one status report with a state receipt.
Choose one live handoff before its next lock-in point. Write the observed state, attach the evidence, name the dependency, and authorize only the next transition. The cost is a few minutes. The return is a claim another person or agent can inspect.
Kill condition: if this discipline adds reporting cost without improving readiness, handoffs, or later decisions, narrow or remove it.
Thing or system: Observed at: Desired state: Desired horizon: Observed state: Evidence: Confidence: Fresh until: Dependencies: Authorized next transition: What would prove this state wrong?
- Time — choose the horizon and freshness the state claim needs
- Predictions — state what may become true without calling it observed
- Dreams — choose the preferred state without calling it probable
- Control loops — compare observed state with the setpoint and correct variance
- Coordination of Intent — connect desired state, commitment, signal, and update
- Coordination — see how beneficiary evidence returns into the next decision
The next question is concrete: where does the truth of this state live?