Natural Selection: The Algorithm That Needs No Designer

The human eye has a blind spot. The nerve fibres carrying signals from the retina run in front of the light-sensitive cells, then bundle together and punch backwards through the retina to reach the brain — leaving a hole with no receptors at all. No engineer would design this. The octopus eye, evolved independently, does it the sensible way round.

This is not a minor imperfection in an otherwise elegant design. It is a clue about the process that produced the eye, and it tells you something the elegance of the finished organ would hide: the eye was not designed. It was accumulated, one viable modification at a time, with no ability to go back and fix an early architectural mistake.

Natural selection is the simplest powerful idea in science. It requires no foresight, no goal, and no designer — and it is not, fundamentally, a theory about animals.

The mechanism, stated precisely

Natural selection follows necessarily from three conditions. Where all three hold, it must occur; it is closer to a logical consequence than an empirical discovery.

Variation. Individuals in a population differ from one another.

Differential survival and reproduction. Some of those differences affect how likely an individual is to survive and reproduce in its current environment.

Heredity. Offspring resemble their parents more than they resemble random members of the population.

Given these three, the traits associated with greater reproductive success necessarily become more common over generations. Nothing is striving. Nothing is being optimised toward. The distribution simply shifts, because the individuals carrying certain variants leave more descendants.

The single most common misunderstanding is what "fitness" means. In this technical sense it does not mean strength, health, or dominance — it means reproductive success in a specific environment. A parasite that cannot survive outside a host and would lose any fight is enormously fit if it reproduces prolifically. A magnificent specimen that leaves no offspring is, in this exact sense, unfit.

Why it produces compromises, not optimal designs

Three constraints explain why evolution yields the blind spot rather than the octopus solution — and why nearly everything biological looks jury-rigged when examined closely.

It cannot plan. Selection operates on the current generation, judged by current conditions. It cannot accept a temporary disadvantage in exchange for a superior long-term architecture, because the organism carrying that temporary disadvantage dies before the payoff arrives. Every intermediate step must itself be viable.

It cannot start over. The vertebrate eye's wiring was established in early ancestors, and every subsequent improvement had to work with that inheritance. Ripping it out and rebuilding would require passing through non-functional intermediates that selection eliminates. This is path dependence, and it is why history constrains biology so heavily.

It optimises for multiple contradictory pressures at once. A peacock's tail is a liability for escaping predators and an asset for attracting mates. The observed tail is a compromise between opposing pressures, not an optimum for either. And the pressures themselves shift — an adaptation finely tuned to current conditions becomes a liability when conditions change, which is precisely why the Red Queen hypothesis matters: the environment includes other organisms that are also evolving.

The algorithm outside biology

Here is why this concept earns a place well outside a biology syllabus. Natural selection is substrate-independent. The three conditions say nothing about genes or organisms. Wherever you find variation, differential persistence, and inheritance, you get the same dynamic — and you can reason about it with the same tools.

Markets satisfy all three. Firms vary in strategy and structure; some survive and others do not; surviving practices are copied by new entrants and imitators. What emerges is not a set of optimally designed companies but a population shaped by whatever happened to persist — including plenty of path-dependent absurdities inherited from earlier eras, exactly like the blind spot. This is the machinery underneath creative destruction.

Ideas satisfy all three. Variants of a claim circulate, some spread further than others, and successful phrasings get copied. Crucially, the selection pressure on an idea is transmissibility, not truth. Selection optimises for whatever correlates with persistence — which is why an environment that rewards outrage reliably produces outrage, regardless of anyone's intentions.

This yields the most useful diagnostic question the concept offers. When you see a system producing outcomes nobody seems to want, do not start by looking for a culprit. Ask instead: what is being selected for here? What variants survive, what gets copied, and what quietly dies? The answer usually explains the outcome better than anyone's stated intentions — because selection does not care about intentions, only about what persists.

The blind spot in your eye is not evidence of poor design. It is a signature, left by a process that never designed anything, and it is worth learning to recognise that signature — because it is stamped on far more of the world than biology.