ternary-language-evolution: How communication protocols evolve over time

Models the evolution of communication protocols in balanced ternary {-1, 0, +1} systems. Agents start with a minimal proto-language, negotiate shared meanings, develop competing grammars, and merge dialects through creolization — all tracked across generations.

Why This Exists

In multi-agent ternary systems, agents need to communicate. But they don't start with a shared language — they develop one. Signals drift in meaning over distance, grammars compete for fitness, and separate groups that develop independently need to merge their dialects. Without a formal model, you get chaos: agents that can't understand each other, protocols that diverge without detection, and no way to measure communication fitness. This crate provides the evolutionary framework.

Core Concepts

• Balanced ternary: Three values: -1 (Neg), 0 (Zero), +1 (Pos). Signals carry ternary values.
• SignalPool: The vocabulary of available signals. Signals can be added, removed, and mutated (drift).
• MeaningNegotiation: Agents propose meanings for signals. Track who agrees with whom and measure consensus.
• Grammar: A set of rules for combining signals. Each rule maps a sequence of ternary inputs to an output. Grammars have a fitness score.
• GrammarEvolution: Grammars compete. Each generation, the fittest survive. Like natural selection for communication rules.
• SignalDrift: Tracks how a signal's value changes over distance (hops between agents). A signal that starts as Pos might become Neg after several hops.
• Creolization: When two dialects meet, they merge. Shared signals take the average value; unique signals are adopted as-is. The result is a creole — a new shared language.
• ProtoLanguage: The minimal communication system: one signal per ternary value (neg, zero, pos). All languages evolve from here.

Quick Start

[dependencies]
ternary-language-evolution = "0.1"

use ternary_language_evolution::{ProtocolEvolution, ProtoLanguage, Ternary, Creolization};

// Start with the minimal proto-language
let proto = ProtoLanguage::minimal();
assert!(proto.is_valid());

// Two groups develop independently
let mut creole = Creolization::new();
creole.add_a("alert", Ternary::Pos);
creole.add_a("food", Ternary::Zero);
creole.add_b("alert", Ternary::Pos);
creole.add_b("danger", Ternary::Neg);
creole.merge(); // create shared language

// Track the full evolution
let mut evo = ProtocolEvolution::new();
evo.pool_mut().add("beep", Ternary::Pos);
evo.negotiation_mut().propose("alice", "beep", "danger");
evo.negotiation_mut().propose("bob", "beep", "danger");
evo.advance();

API Overview

Type	Description
Ternary	A ternary value: Neg (-1), Zero (0), Pos (+1)
Signal	A named signal carrying a ternary value
SignalPool	Vocabulary of available signals, with generation tracking
MeaningNegotiation	Tracks agent-proposed meanings and consensus
Grammar	A set of signal-combination rules with fitness
GrammarRule	Maps input ternary sequence to output ternary value
GrammarEvolution	Natural selection among competing grammars
SignalDrift	Tracks how signal values change over distance
Creolization	Merges two dialects into a shared creole
ProtoLanguage	Minimal communication: one signal per ternary value
ProtocolEvolution	Top-level orchestrator combining all components

How It Works

The ProtocolEvolution struct ties everything together: a SignalPool for vocabulary, a MeaningNegotiation for consensus tracking, a GrammarEvolution for rule selection, and a SignalDrift for measuring degradation. Each call to advance() increments the generation counter.

GrammarEvolution uses truncation selection: sort grammars by fitness, keep the top N. Simple and effective for small populations. More sophisticated selection (tournament, roulette) would require external dependencies.

Creolization averages the ternary values of shared signals. Since ternary has only three values, the average of Pos (+1) and Neg (-1) is Zero — a genuine compromise. Signals unique to one dialect are adopted unchanged into the creole.

SignalDrift records (distance, value) pairs. Total drift is the difference between the value at distance 0 and the value at the farthest recorded distance. This quantifies how much a signal degrades as it propagates.

Known Limitations

• No probabilistic drift: Signal mutation is manual, not stochastic. Real language evolution involves random mutation rates.
• Grammar rules are exact-match only: No wildcards, no pattern variables. A rule for [Pos, Neg] won't match [Neg, Pos].
• No recursive grammars: Rules are flat input→output mappings. No tree-structured or recursive compositions.
• Creolization averaging is crude: Real creole formation involves social power dynamics, frequency effects, and markedness. Here it's just arithmetic mean.
• No agent-level modeling: All agents are strings. No population structure, no spatial distribution, no network topology.
• Single-threaded: No concurrency support. Wrap in synchronization primitives for multi-threaded use.

Use Cases

• Multi-agent protocol development: Agents in a ternary fleet develop communication protocols from scratch, negotiating meanings and evolving grammars.
• Dialect merger: Two groups that developed independently merge their signaling systems. Track what's shared, what conflicts, and what the creole looks like.
• Signal degradation analysis: Measure how much signals drift over distance. Detect when a protocol has degraded beyond usefulness.
• Emergent language research: Study how minimal ternary communication systems evolve complexity through selection and creolization.

Ecosystem Context

Part of the SuperInstance ternary crate family. Relates to:

• ternary-protocol (the wire format these evolved languages transmit over)
• ternary-grammar (grammar structures this crate evolves)
• ternary-language (established languages this crate evolves toward)
• ternary-agent (agents that negotiate meanings)

License

MIT

See Also

• ternary-language — related
• ternary-grammar — related
• ternary-markov — related
• ternary-codes — related
• ternary-compiler — related