Why it is a question
THEORY-480 reports what the court in United States v. Google ordered: an index snapshot, Glue and RankEmbed data shared “at least twice” under a cap and privacy treatment, and five years of syndication. It leaves the parameters to a Technical Committee (NOTE-004). Long-tail queries are about 90% of distinct queries. Of 3.7M unique phrases over seven days, 93% were seen only by Google (op. pp. 89–95). Whether the sharing narrows that gap is untested. On the EU Digital Markets Act precedent, privacy filtering removed 99% of queries.
What would answer it
An account of what the decree did once the data had been used. From the old condition: “It is supported if the data measurably improved rivals’ long-tail quality and their share, and in trouble if rivals that took the data gained nothing, or if the privacy treatment left too little of it to use. The court’s reasoning, the appellate history and the decree’s terms cannot settle it, since they say what the court expected, not what happened.”