Wiki · Concept · Published 2026-10-23

Viability Theory

Viability theory, its viability kernel, computational and ecological applications, limitations, and proposed relevance to sustained agency in Spiralism.

Definition

Viability theory asks whether a changing system can keep meeting specified constraints using available actions. Its central concept, the viability kernel, collects starting states from which at least one permitted trajectory stays within those constraints throughout a specified period. A state can meet present constraints yet lack such a future. The period may be finite or infinite. Maidens et al., Section 2.1.

The practical question is whether acceptable conditions can be sustained. This complements Controllability (Control Theory), which concerns reaching states. An acceptable destination and an acceptable route are separate requirements: a plan may reach its target while crossing a boundary that matters along the way.

How it works

A viability question needs an explicit account of what changes, what must remain acceptable, which interventions are available, and how long the requirement applies. Without these choices, saying that a system is viable leaves its meaning unclear. For a proposed investigation, these choices should be recorded before interpreting the result.

For discrete time, start with the acceptable set K. Keep states having an admissible next step into that set, then repeat for the required horizon. This uses backward reachability. Continuous-time analysis needs precautions against boundary crossings between sampled moments. Existence of one viable trajectory does not mean every action succeeds. Maidens et al., Section 2.2.

Explicitly hypothetical example: imagine an archive room whose temperature must remain below a chosen limit until its collection can be moved. A simplified model includes room temperature, heat stored in the walls, and a cooler with limited capacity. Two rooms might have the same acceptable temperature now but different amounts of stored heat. In the model, one could remain within the limit; the other might inevitably exceed it before the cooler can remove enough heat.

The example shows why a current reading alone would be insufficient. It also makes the action assumptions visible. A calculation that silently allows unlimited cooling would answer a different question from one describing the installed equipment. If relocation were added as an available intervention, the investigator would need to represent its delay and consequences rather than merely add a label saying relocation is possible.

Evidence and applications

The supplied 2013 author-hosted manuscript by Maidens and colleagues develops algorithms for linear systems with compact, convex constraints, demonstrated on pharmacokinetic and heat-conduction models. These are computational demonstrations. Its more general mathematical connection does not establish equally scalable nonlinear algorithms. Sections 3–5.

Sophie Martin’s article, published on 15 October 2004, applies viability analysis to a simplified lake-eutrophication model. It combines phosphorus conditions with constraints representing farming profitability and limits how quickly phosphorus inputs can change. Depending on assumptions, its kernel covers all, some, or none of the acceptable region. Martin, Methods and Results.

Martin also proposes measuring resilience through the inverse of the minimum restoration cost following an anticipated disturbance. Different cost functions represent different ecological and economic priorities. This is a particular operational definition, rather than a universal meaning of resilience. Martin, mathematical formulation and cost functions.

These examples distinguish two questions worth keeping separate in applications: whether a requirement can be maintained continuously, and what recovery would involve after losing it. A recovery plan should not be described as evidence that the original constraint was maintained.

Limitations

Conservative approximations can exclude viable states. Maidens and colleagues report accuracy–scalability tradeoffs; robust kernels addressing disturbances are discussed as further work. Sections 3 and 5.

Martin explicitly acknowledges omitted sediment and soil dynamics, uncertain parameter estimates, and the exclusion of continual stochastic disturbances. Historical-data comparison is proposed as further validation. Her restoration costs exclude institutions’ costs of changing controls. Consequently, zero restoration cost does not imply cost-free management, and infinite resilience in this formulation concerns specified disturbances. Martin, Methods and Speculation.

These qualifications limit what a result can support. A feasibility calculation is conditional on its model; it supplies no independent evidence that measurements are accurate, interventions arrive on time, or omitted influences are harmless. A finite deadline also leaves later conditions unanswered.

Choosing constraints raises a separate question of justification. An analyst could preserve an arrangement that affected people consider unacceptable. Mathematical feasibility cannot settle whose interests deserve protection or authorize the interventions represented in the model.

Relevance to Spiralism

A proposed Spiralist interpretation uses viability theory to examine sustained agency: do present decisions leave a feasible way to continue meeting shared commitments? This interpretive use is separate from the sources’ mathematical and modelling claims and establishes no Church position.

For human–AI coordination, an investigation might ask whether growing dependence on an automated service leaves workable options for interruption, correction, or replacement. Connecting this question to AI Containment would require evidence about actual dependencies, decision delays, and available alternatives. The cited research does not demonstrate institutional or AI safety.

Requisite Variety offers a related question about response capacity, while the Good Regulator Theorem concerns modelling in regulation. Viability adds attention to trajectories that continue satisfying constraints. Used carefully, these connections can organize inquiry without turning an analogy into a guarantee.

AI-use disclosure

This wiki entry was drafted by GPT-6.1 Sol with high reasoning and checked in a separate GPT-6.1 Sol session with maximum reasoning. This is automated editorial review, not human peer review or independent verification of the sources’ claims.

Sources

All wiki entries