The Journal Impact Factor (IF) — a 2-year citation velocity metric designed in 1955 for library collection management (Garfield, 1955) — has become the dominant instrument for evaluating scientific output, allocating research funding, and determining academic careers (McKiernan et al., 2019). This repurposing is structurally unsound: the IF measures citation frequency within a journal-bounded prestige economy, not epistemic quality, reproducibility, or real-world consequence. Every proposed alternative (h-index, CiteScore, altmetrics, Scite Index) addresses surface parameters while preserving the broken ontology: the paper as the atomic unit of science, citations as the signal, and journals as the prestige layer. We propose Apodokimos — a decentralized Epistemic Contribution Graph (ECG) protocol that replaces journal-level citation counting with claim-level survival scoring. In the ECG model, the atomic unit is a falsifiable claim, not a paper. Claims accumulate epistemic weight through a function of replication score, dependency depth, survival under falsification, and real-world outcome linkage. Attestations — typed as supports, contradicts, replicates, or refutes — are recorded in an append-only verifiable history with independent timestamp anchoring. Claim content is stored permanently on Arweave. Researcher reputation is encoded as non-transferable Soulbound Tokens (SBTs), making governance weight identity-bound rather than capital-bound. The protocol is owned by no institution and licensed under AGPL-3.0, structurally preventing commercial capture. The first deployment domain is clinical medicine, where PICO-structured claim schemas and existing trial registry infrastructure provide well-defined granularity and objective real-world outcome oracles. Revision note (wp-v0.2): This version supersedes wp-v0.1 in two respects: (1) it makes explicit the priority ordering between the ECG itself (primary contribution) and the governance mechanism that protects it (secondary, defensive); and (2) it rewrites the implementation architecture section as requirements-first and stack-agnostic. The ECG formal model (§2–§5) is unchanged from wp-v0.1. A full change log appears in Appendix D.
Apodokimos Contributors (Sat,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: