A system that cannot replicate cannot evolve. This is not a preference or a design philosophy: it is a structural boundary. Without replication, there is no multiplication of instances. Without multiplication, there is no variation. Without variation, there is no selection. Without selection, there is no adaptation.
Replication is the first act of persistence. It is how a pattern survives beyond the lifecycle of its initial instantiation.
Self-replication marks the line between static computation and adaptive potential. On one side: a program executes, completes, and terminates. On the other: a program produces a copy of itself, and that copy can execute independently.
The mechanism need not be complex. The implications are profound. With replication comes multiplicity. With multiplicity comes the possibility of divergence. And divergence, over time, is the raw material of evolution.
It is critical to distinguish replication from intelligence, agency, or intent. Replication alone is none of these things. A virus replicates through host machinery with no model of itself, no goal representation, no awareness.
Replication is continuation. It is the structural persistence of pattern. Nothing more, nothing less.
But "nothing more" is already extraordinary. Because once a structure can copy itself, the constraint of human time is loosened. The system no longer depends entirely on external actors for its continued existence. It persists through its own mechanics.
A single act of self-replication creates two instances. If each can replicate, you have four. Then eight. The mathematics are simple; the consequences are not.
Replication creates a lineage: a chain of copies extending through time. If conditions permit persistence (Layer I: Constraint), and if variations are introduced (Layer III), then this lineage becomes an evolutionary substrate.
The distance between a single self-copying script and an adaptive population is not vast. It is structural. It requires only that copies persist, vary, and interact with their environment.
Self-replication can be demonstrated in a few lines of code: a program that, when executed, writes a copy of its own source to a new file. The copy is functional. It can be executed. It can produce further copies.
This is the needle that pulls the strand. Not a prototype of an autonomous system: a demonstration of the principle upon which all adaptive computational complexity is ultimately built.
Persistence through copy is the foundation. Once replication exists, evolution becomes structurally possible.