The Replicators

Core idea

Before there was life there was chemistry. Before there were genes there were replicators. Dawkins's account of the origin of life is not a detailed biochemistry — it is an argument about what kind of thing had to come into existence first, and what properties that thing must have had.

Dawkins's argument: Once a molecule arose in the primeval soup that could, by accident, act as a template for copies of itself, natural selection began — automatically, with no further ingredient. Any entity that makes copies of itself will, by simple logic, come to outnumber entities that do not. From that moment onward the story of life is the story of competition among replicators.

The first replicator did not need to be alive in any rich sense. It only needed to do three things well enough: persist long enough to be copied (longevity), make many copies (fecundity), and make those copies faithfully (copying-fidelity). Variations between molecules in any of these three would produce differential survival of variant lineages — that is, evolution.

Why it matters

The accident that started everything

Dawkins is careful that the topic is not a theory of how life began, just a requirement-list. Some kind of molecule in early Earth's chemistry — fed by lightning, ultraviolet, volcanic heat — must by chance have come together with the property of acting as a template. The Miller-Urey experiments of the 1950s had shown that amino acids and other building blocks form spontaneously under early-Earth conditions; what the topic postulates is one further chance step: a molecule whose shape persuaded surrounding building blocks to line up against it and become a duplicate.

The chance is, considered for any one molecule on any one occasion, vanishingly small. But the soup was vast, the time available was hundreds of millions of years, and "vanishingly small" multiplied by "vast" and "long" can be "more than once." That is all the topic needs.

Stability is the engine

The deepest move of the topic is to recast natural selection in the most general possible terms: it is the differential survival of stable forms. A pile of pebbles on a beach is "selected" because the round ones roll to the bottom and the flat ones stay up the slope. No magic — just geometry plus waves. The pre-biotic chemistry version of the same logic is that some molecular configurations last longer than others, and the longer-lasting ones come, over time, to be more common.

A replicating molecule simply takes this further: it not only persists, it makes more of itself. So now selection is no longer just about which static forms are stable, but which forms are stable as lineages. Lineages with more accurate copying, more frequent copying, and more durable molecules outgrow lineages with less of those properties.

The three properties define the rest of biology

Longevity, fecundity, copying-fidelity will recur throughout the book. The bee's suicide sting, the alarm call, the parental sacrifice, the meme — all are about which of these three properties is being maximized for which replicator, and at what cost. The topic establishes the vocabulary the entire rest of the book uses.

Survival machines, in embryo

Towards the topic's end, Dawkins describes how replicators must, sooner or later, have begun building containers around themselves to defend against degradation and to monopolize resources. These containers — initially simple protein shells, later cells, later whole organisms — are the precursors of what the book will call survival machines. The argument is teleological in form but not in fact: replicators that happened to build better containers were copied more; their descendants inherited the container-building trick; the result, scaled up four billion years, is us.

Key takeaways

Mental model

Copy sourceZoom

Practical application

Use the three-property test on any "evolving" system

Any time someone claims a system is "evolving" — software ecosystems, languages, religions, businesses, ideas — apply the three-property test. Are there entities that copy themselves (or are copied by some external machinery)? Do those entities vary? Is the variation heritable? Are some variants outliving others in identifiable ways? If yes to all four, the system supports an evolutionary process and the formal tools of evolution will help you reason about it. If any one is missing, the metaphor is just decoration.

Distinguish "stability" from "design"

The topic trains a habit that recurs throughout the book: distinguish features that exist because they were selected for from features that exist because they are simply stable. Many things look designed but are only persistent — they last because nothing easily destroys them, not because anyone (or any process) preferred them. Asking "why does this exist?" begins with the cheaper hypothesis (stability) before invoking the expensive one (selection).

Hold the time-scale in your head

The single largest barrier to understanding evolution is human intuitions about probability. "It's so improbable it can't happen" is almost always wrong when the relevant time-scale is hundreds of millions of years and the relevant population is in the trillions. The topic does not say this in those words, but it is the lesson. Practice scaling: a one-in-a-trillion event over a billion years on a planet-sized chemistry set is expected, not miraculous.

Example

Consider a chain email — the prototypical low-tech meme. It carries instructions saying "copy me and send me to ten friends." Those that lack such instructions die out: nobody forwards them. Those that include the instructions but also include something boring die out: nobody bothers. Those that include the instructions and a hook (a prize, a curse, a guilt-trip) propagate.

Over thousands of generations of forwarding, the surviving emails are not those that are true or useful — they are those that have hit on the best combination of fecundity (compelling forward instructions), longevity (memorable enough to be re-read), and fidelity (short enough to be copied without error). This is exactly the three-property frame Dawkins lays down for the first molecular replicators. The chain email and the proto-RNA are doing the same dance, four billion years apart.

Related lessons