Spawning more agents on the same evidence collapses posterior coverage from 0.94 to 0.26.

This paper formalizes an epistemic Sybil problem: another agent is not another observation, because apparently independent reports can descend from the same evidence root and no report-only aggregator can generally distinguish replication from corroboration. Over 20,000 controlled agent report and extraction calls confirm it. Holding one evidence root fixed while report multiplicity rises from 1 to 32 drops naive posterior coverage from 0.940 to 0.263; raising evidence roots from 1 to 16 at fixed report count closes the gap entirely. Replicate extraction errors from a shared base model correlate at gamma 0.719. Fanning out N agents over one source buys confidence, not information.