Counting citations rewards the loudest, not the best corroborated. When one source is echoed a thousand times, a tally can't tell it from a thousand independent witnesses. Validiti weighs evidence by how many independent sources actually back it — so an echo can’t out‑vote the truth.
Two claims, each cited twenty times. One is twenty echoes of a single source; the other is twenty independent sources. A counter scores them equal. Press the button — a live, sealed Validiti node weighs both and shows the difference, deterministically and signed.
Search rankings, reputation scores, and the data AI trains on all reward repetition. That’s exactly the opening for manipulation: flood the record with copies and the count says “consensus.” Weighing by independent origin closes it — a thousand copies of one claim still counts as one.
The same weighing also tracks consequence — a claim relied on and borne out in use gains weight; one later falsified loses it. Corroboration, provenance, and outcome, in one signed number.