Uplift and Engineered Evolution

Seed idea. Sparked by Children of Time — the nanovirus meant to accelerate a species toward intelligence.

The hook

Deliberately steering evolution: directing, rather than waiting on, the path to greater intelligence or capability.

Threads to grow

• “Uplift” in SF (Brin, Tchaikovsky) vs. real tools (gene editing, directed evolution).
• Accelerated evolution as a virus/agent vs. selective breeding vs. CRISPR.
• The control problem: you seed a process you can’t fully steer (the virus finds a different host).
• Connects to Genetic Algorithms and Benevolent Viruses as the computational analogue.

Side Project / Prompt Seed

If I hand this file to Claude: let’s build something. Start by digging up relevant research, then prototype a proof of concept in code.

Premise: directed evolution toward a target capability — and the control problem when the process escapes your intent (the virus finds a different host).

What to build (POC): a directed-evolution sandbox. Evolve simple agents toward a goal (e.g. reach a target behavior), but add an unintended pressure or a second “host” niche the selection can spill into. Show how the outcome diverges from the designer’s intent — the book’s core irony, in miniature.

Research to pull first:

• Directed evolution in biology (Frances Arnold’s work).
• “Uplift” in SF (Brin’s Uplift series, Tchaikovsky) — the idea’s lineage.
• Reward hacking / specification gaming in RL — the modern “it optimized the wrong thing” literature.

Stretch: reuse the engine from Genetic Algorithms and Benevolent Viruses and add the divergence mechanism on top.