Twenty-five of Oyster Rank's 72 signals carry deprecation flags in the recovered binary

A string in a shipped binary proves that some code path once had a name for something. It does not give the weight or the stage, and it leaves open whether that path still executes. The deprecation flags matter more than the count: the analysis says 25 of the 72 enumerated signals are explicitly marked deprecated ^[4], which leaves 47 unmarked ^[5] and about 35 percent of the list tagged as history ^[6]. Unmarked means only that nobody flagged it.

The dev.to post attributes the underlying discovery to Search Engine Land, whose reporting it describes as a view into this architecture and not a definitive guide to a static local-ranking algorithm ^[7]. The post is silent on who recovered the binary and when ^[16].

The recovered materials reportedly describe a business or place through an entity-based system that can combine information from numerous providers and interpret many kinds of local intent ^[8], with 793 data-source providers and 446 local search intent types referenced ^[3]. Under that description, an edit to a Google Business Profile is one provider asserting a fact about an entity that other providers also assert things about. The post calls profile edits evidence or votes within the ecosystem, not commands that directly instruct Google to rank a business ^[9]. A wrong phone number sitting in a third-party feed survives a correction to the profile. It stays in the pile as another assertion, and the post says a complete profile cannot compensate indefinitely for conflicting, incomplete or weak information elsewhere in the data environment ^[13].

The expensive detail for anyone running a dashboard is the split between scoring stages. The reporting identifies separate offline or on-device scoring and distinguishes ranking from map visibility, so appearing in a particular map context is not necessarily the same process as ranking for a conventional local query ^[10]. Those are two measurements, and the post advises tracking local-search visibility separately from whether business facts stay consistent and whether profile updates are reflected correctly ^[14].

Three things would have to hold before the 47 unmarked names change how a local program spends its hours. The binary is production Maps code. The unmarked names correspond to stages that still run. And the relative weights are stable enough to plan against. The recovered material comes without public weights or a definitive list of active signals, and the post says it cannot be used to predict an individual business's placement ^[11].

There is a tension inside the source itself. Its headline leads with the 72-signal count ^[15]. Its body says the important takeaway is not a 72-factor checklist, because the signals are internal terms used across multiple scoring and retrieval stages rather than a single public formula with fixed weights ^[2].

What the document supports is unglamorous work: checking that name, address, phone details, category information, operating hours and website location pages accurately reflect the same business ^[12]. Local SEO practice has centered for years on listing completeness, reviews, categories, proximity and citations ^[17]. The binary adds a structural reason to expect contradictions across hundreds of providers to cost something.