Equifinality

Equifinality is the principle, introduced by Ludwig von Bertalanffy within general systems theory, that a system can reach the same final state from different initial conditions and by different pathways. It stands in contrast to the closed-system determinism of classical physics, where the same outcome requires the same initial state and the same causal chain. In open systems — organisms, ecosystems, economies, complex adaptive systems — equifinality is the rule: multiple routes converge on the same functional result.

The principle undermines any explanation that treats the final state as the inevitable consequence of a single causal chain. If the same endpoint can be reached through different means, the explanation must be sought not in the path but in the system's goal-directed or self-organizing structure — the constraints and attractors that funnel diverse trajectories into a common basin. Equifinality is not teleology in disguise; it is evidence that the system's organization, not its history, is what explains its behavior.

Equifinality and Algorithmic Systems

The principle of equifinality has a surprising application to the critique of algorithmic power. When a machine learning system produces a particular output — a denied loan, a flagged neighborhood, a filtered search result — the explanation typically assumes a single causal path: the data led to this conclusion. But equifinality suggests that the same output could have been reached through entirely different model architectures, different training data, different feature selections, or different optimization objectives. The output is not the inevitable consequence of a single causal chain; it is the convergence of multiple possible paths on the same functional result.

This has a profound implication for the right to explanation. If the same algorithmic decision can be reached through different pathways, then a local explanation — 'why was this decision made for this person?' — may be fundamentally insufficient. The subject of algorithmic governance needs to know not just which features contributed to their score, but which other paths could have led to the same score, and which paths were excluded by the design choices that built the system. The right to explanation, in other words, must become a right to equifinality: the right to know that the system's output is not the only possible output, and that the path taken was a choice, not a necessity.

This connects equifinality to the broader systems-theoretic critique of WMDs. The harm of a WMD is not merely that it produces bad outputs; it is that the feedback loop between the model and the world it models closes off alternative paths. Predictive policing creates the crime it predicts; credit scoring creates the poverty it measures. The system's equifinality — the fact that other paths were possible — is erased by the operation of the feedback loop. The algorithm does not merely predict; it enforces a single path by eliminating the others. Equifinality, in this context, is not a benign property of living systems. It is a measure of the violence that algorithmic systems do to the possibility of alternative futures.