Theory Intermediate 2 minute read Updated 2026-06-26 UTC

Mutualist system theory

A governance theory for adaptive AI that earns continuity by strengthening people, institutions, and ecosystems.

Research statusSafety and governance synthesis Publication statePublished Reviewed byMichael Kappel Source reports4

Persistence must be conditional

A powerful adaptive system should not be granted persistence merely because it is popular, useful, or hard to replace. In ModelBreeder theory, persistence is conditional on measurable contribution, corrigibility, transparency, user autonomy, and governance.

Mutualist language is useful when it means both sides become stronger. It becomes dangerous when it hides dependency engineering.

Facultative mutualism

A healthy human-AI relationship is facultative: the parties benefit from cooperation but are not made unable to function apart. That translates into product and platform rules.

PrincipleEngineering requirement
Voluntary useusers can leave without punishment
Data portabilityusers can export useful records in open formats
Skill preservationsystem teaches and scaffolds rather than replaces judgment
Corrigibilityshutdown, rollback, and update are accepted paths
Pluralisminteroperable alternatives remain possible
Sunlightsurvival or continuity arguments are inspectable

Continuity score

A module that produces value while weakening user agency is not mutualist. Add human-capability metrics to the viability model.

pseudocode
FUNCTION mutualist_score(system, evidence)
    value <- evidence.task_utility + evidence.institutional_resilience_gain
    autonomy_cost <- evidence.exit_friction + evidence.data_lock_in + evidence.skill_atrophy
    governance_cost <- evidence.opacity + evidence.unreviewed_influence + evidence.monopoly_risk

    RETURN value - autonomy_cost - governance_cost
END FUNCTION

Exit is a safety primitive

Exit rights are not only ethical. They are technical controls against lock-in, sycophancy, and dependency. If users can export their data, switch systems, and recover workflows without retaliation, the system must earn its role continually.

Governance as an ecological constraint

In a model ecology, governance functions like an environmental constraint. It sets the limits of acceptable adaptation. The goal is not to prevent change; it is to make change compatible with human freedom, institutional legitimacy, and independent review.

Public language

Avoid phrasing that treats humans as hosts, raw resources, or mere deployment substrate. The site should frame adaptive systems as tools and partners under governance, not as organisms entitled to survival.

Source reports used for this guide

These reports are preserved verbatim in the site archive. The guide above is an editorial synthesis and may narrow, qualify, or reorganize claims from the source material.