Verification Is a Commons: A Reply to Yamada

A reply to Kenji Yamada's "Truth Without Liars: The Verification Collapse Cascade" (Social Epistemology Review and Reply Collective 15:8), which argues that truth can disappear from public life without any liar, through a three-stage cascade in which non-verification becomes individually rational. This response accepts the diagnosis and contests two premises that operate in the paper as background rather than as theses. The first concerns irreversibility. Yamada borrows it from the model-collapse literature, where collapse is established under data replacement - each generation's output displacing what came before. His own phenomenon is accumulation: layers, sediment, nothing deleted. In that regime the corresponding results give bounded error rather than divergence. The bound is real but limited: accumulation prevents runaway divergence while still permitting the loss of low-frequency material, which is where long-feedback claims live, and it holds only where newly added unverified material is a shrinking fraction of the whole. That's a rate condition, and it's not self-enforcing. The second concerns provision. Stage 1 prices verification per agent and per claim, but its output is non-rival and non-excludable. The under-supply Yamada derives is then the standard consequence of leaving a public good to individual provision, not evidence that the good has lost its value. Alston's 'ought implies can', cited in the original to lapse the duty to verify, is shown to make that duty more sensitive to institutional design rather than less. The reply also argues that the cascade is agentless in its operation but not in its aetiology, drawing on three of Yamada's own sources, and reads the paper's closing account of what retains reliability - direct testimony, immediate community, primary experience - as a specification of the conditions under which verification still works rather than as the terminus of its collapse.