The Ancestry of Belief is the foundational theoretical statement of the epistemic-lineage research program. It argues that for any bounded intelligence — a person, an institution, a language model, an agent, a robot, a civilization — that must act on belief it cannot personally verify, the operationally fundamental primitive is not truth but lineage: the durable, attributable, externally-inspectable ancestry of how a belief came to be held and how it changed. The central move, in one line: classical revision theory (truth-maintenance systems; the AGM tradition) treats revision history as disposable internal bookkeeping for a single, privately owned belief base; epistemic lineage treats revision history as a durable object of reasoning shared between minds. Truth does not stop mattering — it remains the target. But where truth cannot be directly established, which is the default condition for any mind acting on inherited belief, lineage becomes the operational mechanism for allocating trust. The shift is not from truth to ancestry; it is from blind inheritance to inspectable inheritance. From the primitive, two theorems are developed as the paper's signatures: coordination through traceable disagreement (a third way to hold a society of minds coherent, beside consensus and authority) and accountable compression (the requirement that lineage be not merely preserved but traversable). The same primitive operates at three scales without changing: it lets an individual inspect inherited belief, lets institutions preserve reasoning across turnover, and lets societies of minds coordinate without consensus. The work is deliberately precise about its bound: lineage is an accountability primitive, not a truth oracle — a permanent substrate preserves a lie as faithfully as a fact. It observes that civilization has hand-built partial, institution-dependent lineage systems for millennia (citation, peer review, the archive, legal precedent, the footnote), each surviving only as long as a keeper behaved, and identifies what a neutral, append-only, permanent, attributable substrate adds: a move from institution-dependent lineage (a social guarantee, contingent on conduct) to substrate-dependent lineage (an architectural guarantee, contingent on properties that hold). Arweave/AO is used as a single existence proof of those properties, not as a product; the argument is about substrate properties and is built to survive if any particular implementation disappeared. This paper is the philosophical spine connecting Metavolve Labs' empirical and systems work on data provenance, model faithfulness, and trust infrastructure. It is offered as a position/theory paper whose civilization-scale framing is the horizon while its load-bearing claims are kept deliberately tractable.
Tad MacPherson (Sun,) studied this question.