PageRank and web scale
Brin and Page described a search engine that scores a page's importance by the structure of links to it rather than by the page's own content.
Why it matters
The question of what matters here was solved across millions of documents, creating the infrastructure the later training data grew out of.
PageRank computes the probability that a random surfer lands on a page, so a link from an important page counts for more. The paper also describes the architecture: crawling, indexing, storage. In this atlas it is an infrastructure record: the company founded the same year later produced the Transformer, TensorFlow and a large share of the computation the present field rests on.