We were promised Strong AI, but instead we got metadata analysis

A 2021 essay arguing that structured metadata consistently outperforms AI/machine learning in practice. Using Google Search as the primary case study, it traces how crawling relies on Sitemaps, content understanding relies on OpenGraph and JSON-LD tags, and ranking relies on backlinks (PageRank) rather than true language understanding. The pattern extends beyond search: self-driving cars use GPS-mapped speed limit databases, fraud detection uses merchant whitelists, and government surveillance prefers communication metadata over message content. The core thesis is that once a task's value is proven via AI, society eventually provides the necessary data in machine-readable form, making the AI redundant. The essay criticizes Google's 'focus on content, forget metadata' messaging as self-serving mythology that discourages webmasters from using simple, effective structured data.