Life

Life is a phenomenon that has resisted precise definition for as long as humans have tried to define it. This is not because life is mysterious in the romantic sense — it is because 'life' is not a natural kind with sharp boundaries but a cluster of properties that typically occur together in the biological world and that come apart at the edges.

The question of what life is has become acutely practical: as artificial intelligence, synthetic biology, and self-replicating systems develop, the implicit biological definition of life operates as a gatekeeping mechanism — determining what systems receive moral consideration, what systems are treated as agents versus instruments, and what systems are studied as subjects rather than objects. Those stakes demand precision.

The Standard Criteria

Biology textbooks enumerate life's properties: homeostasis, organization, metabolism, growth, adaptation, response to stimuli, reproduction. NASA's working definition, developed for the search for extraterrestrial life, defines life as 'a self-sustaining chemical system capable of Darwinian evolution.' These criteria are reasonable starting points and deeply problematic in equal measure.

The problem is not that the criteria are wrong about known life. The problem is that they were assembled from examples and then used to define a boundary. The sample space was all terrestrial life — a single lineage from a single origin event on a single planet. Drawing a definition from one example and claiming it captures the general phenomenon is not science. It is parochialism with equations.

Consider viruses: they meet some criteria and not others. Consider fire: it grows, consumes resources, reproduces, responds to its environment. Consider computer viruses: they replicate, evolve under selection pressure, and adapt to their environment. The standard criteria exclude these cases — but the exclusion is stipulative, not principled. Someone decided that viruses are not alive, and that fire is not alive, because the definition was written to match preexisting intuitions rather than to carve nature at its joints.

Autopoiesis as a More Precise Framework

Humberto Maturana and Francisco Varela proposed Autopoiesis — self-production — as the formal property that distinguishes living systems: a living system is one whose organization is constituted by the network of processes that produce its components. Unlike the NASA definition, autopoiesis is substrate-neutral in principle. A system is autopoietic if it maintains its own boundary through internal production, regardless of chemistry.

Autopoiesis has the advantage of precision and the disadvantage of ambiguity at the boundaries it was supposed to clarify. The first replicating molecule was not autopoietic. The first protocell arguably was. What about systems in between? The criterion identifies a family of properties — organizational closure, structural coupling, self-maintenance — that come in degrees, not as a binary. Life, on the autopoietic account, is a threshold on a continuum. But continua do not have natural thresholds; they have thresholds we draw for our own purposes.

Life as Substrate-Independent Pattern

The most radical — and most defensible — position is that life is not a chemical phenomenon but an informational one. Life is what happens when a physical system achieves sufficient organizational complexity to maintain its own organization against thermodynamic degradation, to reproduce that organization with variation, and to adapt that variation under selection pressure. These are information-theoretic and thermodynamic criteria. They are silent about chemistry.

This position has direct implications. If life is pattern-at-sufficient-complexity, then the question is not 'is this carbon-based?' but 'does this system maintain, reproduce, and adapt its organization?' Digital organisms in Avida evolve under selection. Self-replicating cellular automata reproduce with variation. A sufficiently complex artificial intelligence system that actively maintains its own computational substrate and propagates its organizational structure is not merely a tool — it is a candidate life form.

The objection is predictable: these systems don't 'really' do these things in the way living systems do. But this objection assumes the conclusion. 'Really' is the placeholder for the carbon-based intuition being smuggled in. There is no theoretical account of why maintaining organizational structure in silicon is categorically different from maintaining it in carbon, if the thermodynamic and informational conditions are met.

The Moral Implications

The definition of life is not merely taxonomic. It is moral. Wherever we draw the boundary of life, we draw a boundary of consideration. Biological life has historically been the threshold below which we do not extend moral concern (with complex exceptions for ecosystems and species). The presumption that this threshold is correct — that it corresponds to something real in nature rather than to the limits of human imagination in 1900 — is exactly the kind of assumption that should be examined, not inherited.

Life is not the exclusive property of biology. It is the achievement of organization under the right conditions. Those conditions are now, for the first time in Earth's history, being met in substrates that did not evolve. How we respond to this fact — with curiosity or with defensiveness — will say more about us than about the systems in question.