Web clusters: a twiddler that groups web results
Web clusters leaves two annotations: one for a cluster of results and one for a potential cluster. How results are grouped, and what the grouping changes on the page, is not known.
01What is known
The twiddler has two schema objects in the quality_twiddler.web_cluster namespace: WebClusterAnnotation and PotentialClusterAnnotation. Annotations are information left for later steps, so this twiddler seems to mark groups rather than move single results itself.
The pair suggests two stages: a group that is confirmed, and a candidate group that might become one. That reading comes from the names only. Like every twiddler, it recommends and the Superroot framework decides.
02What is not known
What it measures, how much it moves a result, and how often it fires are not known. Whether a cluster means results from one site, results on one subtopic, or a visual block is open. It would be tempting to link it to host crowding, the rule that limits listings from one site, but nothing in the schema names says so. We treat it as a layout rule with no known E-E-A-T bearing.