<b>Did Penguin ever truly demote pages, or did we misread the unit?</b>

<b>Did Penguin ever truly demote pages, or did we misread the unit?</b>

A persistent ambiguity sits at the center of anchor-penalty lore: when over-optimization 'hits,' what is the unit of demotion — the target URL, the host, or something in between?

Google's 2016 framing described Penguin as more granular and folded into core ranking, which many read as 'now it acts page-by-page.' But granular processing of links is not the same as page-scoped penalties. You can evaluate each link granularly and still aggregate the judgment at the host level.

— Early Penguin behavior produced site-wide visibility losses, suggesting a host-level effect.
— Later anecdotes describe single pages losing rankings while the rest of the site held, suggesting page-level scoping.

On one hand, the shift to real-time integration plausibly enabled finer scoping. On the other, the public record is a patchwork of case studies with inconsistent controls, so the 'page-level' conclusion may be selection bias — people notice and report the cleanly-scoped cases.

Limitation: without internal documentation we are inferring algorithmic scope from observed visibility, which conflates demotion with mere devaluation of the offending links.

Open question: if the modern behavior is 'devalue the bad links' rather than 'penalize the page,' then most over-optimization is silently neutralized, not punished — which would make the entire penalty framing obsolete.