1. Digital Organisms and Evolutionary Computation

Core concept

A digital organism is a self-contained computational entity that exists in memory, subject to the same evolutionary pressures as biological life: mutation, selection, and inheritance. Unlike traditional AI models that are trained once and deployed, a digital organism lives continuously, adapting to its environment in real time.

Genome Structure

The genome is the organism’s complete hereditary information. In this implementation, each genome consists of:

• Gene arrays: Fixed-size vectors of floating-point values in range [-1.0, 1.0]. Each array represents a distinct phenotypic trait. The number of genes and the length of each array define the organism’s complexity.
• Generation counter: Monotonically increasing integer tracking how many mutation cycles have occurred.
• Fitness score: A scalar in [0.0, 1.0] representing adaptation to current conditions. Higher is better.
• Species tag: A string identifier for lineage tracking across peer exchanges.
• Birth timestamp: Time of genome instantiation, used for age-based decay or pruning.

Mutation Operators

Each tick, the genome undergoes mutation with a base probability (e.g. 5% per value). Three mutation types exist:

1. Point mutation: val += uniform(-0.2, 0.2)   // small drift
2. Scaling mutation: val *= uniform(0.8, 1.2)   // multiplicative shift
3. Replacement: val = random(-1.0, 1.0)         // complete reset

Each mutated value is clamped to [-1.0, 1.0] to maintain stability. The mutation rate can itself evolve through meta-parameters or adapt based on environmental stress.

Fitness Function

Fitness is not computed from a single objective function. Instead, it is shaped by multiple pressure sources:

• User feedback: Accepting a proposal increases fitness (+0.05). Rejecting decreases (-0.03).
• Environmental pressure: Extreme weather (temperature >35C or <0C, humidity <20%, precipitation >5mm) imposes small negative deltas.
• Remote pressure: Feedback from peer organisms on the mesh can broadcast selective pressure signals, influencing local fitness.
• Fitness decay: Without positive feedback, entropy gradually reduces fitness through the natural mutation process.