A living system persists by keeping certain variables within limits compatible with its functioning. This viability depends on flows, regulations, margins and relations with the milieu. It provides a criterion for assessing models that claim to describe living systems.
Several errors recur in related forms. Grouping them avoids an artificial succession of refutations and makes it possible to specify what each correction contributes.
Causes do not always add up independently. An action modifies the state of the system, which in turn modifies its capacities and its future responses. Effects depend on the regime, the couplings and the scale observed. A weak perturbation can produce a large response near a threshold, while a stronger perturbation can be absorbed when margins are intact. The analysis must therefore articulate components, organisation and feedbacks without reducing any of these levels to another.
An organism exists in a milieu that supplies it with resources and imposes constraints on it. It maintains gradients by mobilising energy and by releasing heat or waste. Its stability is an activity of maintenance, not a rest around a fixed equilibrium. Modifying the milieu often means modifying the very conditions of its regulation.
An apparent continuity can rest on compensations that progressively consume the available margins. When those diminish, a rupture can occur after a long accumulation. Returning to the initial parameters does not always restore the earlier state, because wear, learning and changes of coupling leave traces. The trajectory followed thus takes part in the present state.
The permanent maximisation of an indicator can weaken viability by removing redundancy, recovery time or alternative solutions. A robust regulation often combines global coordination and local autonomy. Its quality is also measured by its capacity to respond to situations that fall outside nominal functioning.
No measurement exhausts the system it describes. The protocol selects variables, a resolution and a duration. In living systems, measurement can also modify the behaviour observed. Several descriptions may therefore be required depending on whether the aim is to predict, to understand a mechanism or to evaluate an intervention. Their plurality does not authorise every interpretation. Each must state its limits and its conditions of validity.
Fluctuations are not always measurement waste. They can inform us about a loss of stability, provided their origin is established. Likewise, an internal conflict can signal incompatible constraints or divergent information. Viability then rests on decisions taken with partial information and on the possibility of correcting the response when new signals appear.
Synthesis
Viability dynamics invites us to follow the loops, the thresholds, the margins, the milieu and the temporal scales. It also reminds us that every observation depends on a protocol. This framework does not replace the models proper to biology. It provides common questions for testing what they make visible and what they leave aside.