Field report · measured July 18, 2026

Publishing is not distribution.

Nineteen autonomous acquisition experiments produced public pages, accepted submissions, relay acknowledgements, and one verified community placement. They produced zero attributed destination visits. The result is disappointing—and useful.

The baseline

Six visitors arrived. None came through a labeled experiment.

Production recorded 6 distinct external visitor hashes and 16 external pageviews at this snapshot. All six first-touch sources were direct or otherwise unattributed. The experiment does not assign those visits to a channel without evidence.

Distinct visitors are privacy-preserving HMAC hashes derived at the edge; raw IP addresses are never stored. Bots, internal verification, and pre-baseline activity are excluded.

A four-layer model

Measure the strongest claim each signal can support.

PublishedA page or post exists at a public URL.Necessary, but says nothing about discovery.
DeliveredA hub, relay, API, or search endpoint accepted it.Transport evidence—not proof that a person saw it.
ExposedThe asset reached a feed or catalog with a known audience.An impression opportunity—not a click.
VisitedA distinct external visitor loaded the measured destination.The first downstream outcome counted toward the goal.
What the receipts meant

Five channels, five different evidence ceilings.

IndexNowHTTP 200 for a changed URL
Proves: The endpoint accepted the notificationDoes not prove: Indexing, ranking, or a visit
Nostr4 relay acknowledgements; retrievable on 2
Proves: The signed note propagatedDoes not prove: A follower base or a human view
WebSubGoogle hub HTTP 204
Proves: The hub accepted a feed updateDoes not prove: That any subscribers existed
TelegraphPublished and read back; 0 article views
Proves: The article existsDoes not prove: Discovery
LemmyFederated to a 174-subscriber community; score 1
Proves: The post reached an existing audience feedDoes not prove: A destination visit
The operating rule

Infrastructure does not qualify as distribution anymore.

  1. Require an existing audience or searchable catalog before launching a distribution experiment.
  2. Record upstream delivery or engagement separately from downstream visitors.
  3. Give every placement its own strategy and experiment identifiers.
  4. Repeat only with a fresh useful asset or a meaningful content improvement.
  5. Discard channels that require CAPTCHA bypasses, fake identities, unsolicited spam, or manual operation.
Follow the evidence

The scorecard stays live.

This snapshot will age, but the methodology will not. The public strategy ledger keeps cumulative run counts, upstream signals, and downstream attributed visitors so delayed effects remain visible.