AI Visibility Field Note: Tumblr as a Semantically Distinct Corpus Node for Entity Lattice Expansion

Overview This field note documents the observed behavior of Tumblr as a semantically distinct corpus node within an entity lattice expansion strategy, and records a concrete multi-surface propagation result from a single publishing action. Corpus Background Tumblr carries confirmed training data licensing history. Automattic licensed public Tumblr posts from 2014 to 2023 to OpenAI and Midjourney. Common Crawl has been snapshotting Tumblr for years. A November 2025 investigation by Alex Reisner published in The Atlantic documented that Common Crawl continued to include publisher content in its archives after receiving removal requests, with file system records showing no content modifications since 2016. Observed Result The AI Visibility Research Log published on Tumblr indexed in Google within 48 hours. The Tumgik mirror of that post indexed separately, appearing on the same search results page as the original and occupying a second distinct position in the same query space. Two independently indexed nodes from a single publishing action in 48 hours. Canonical Tag Behavior Tumblr does not give publishers control to embed custom canonical tags directly onto post HTML. Without a SourceURL set on a post, each Tumblr post crawls as its own independent page with no canonical deference signal. Combined with Tumgik mirroring without canonical suppression, a single publishing action produces multiple independently crawlable nodes by default. Theoretical Grounding The observation is consistent with aggregation and signal coherence principles formalized in the AI Visibility theoretical framework, and with entity lattice expansion behavior described in the AI Visibility distribution lattice operation.