, , , , ,

Back in Feb 5 2019, Prof. Stephen Robertson published “A Brief History of Search Results Ranking” (https://ieeexplore.ieee.org/document/8634887). I finally got a complimentary copy from Robertson via ResearchGate. This is a historical article all IR researchers and search engine marketers (SEOs/SEMs) may want to read. It is part of the IEEE Annals of the History of Computing (https://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=85).

His take on PageRank? He writes: “As a secondary aim, I hope to debunk the myth of PageRank, which remains in widespread currency.”

In the paper, Robertson sets the record straight regarding the myth of Google’s PageRank. He wrote:

“So how important was PageRank in the Google ranker? It was one of a large number of features, and contributed something to the overall effectiveness of the ranker. But in my view, it was much less important than doing all the other things well. And in particular, the advantage claimed for PageRank (that it quantifies the authority of a page as viewed by other web page authors) can already be obtained from matching the query against anchor text. Anchor text is a little less subtle than PageRank as a quantitative measure, but on the other hand it is query specific, which PageRank is not. This fact has been recognised in the information retrieval
research community, supported by some evaluation work by Hawking and colleagues, for example — work which was published nine years before McCormick’s book.

Matching anchor text well is vital for a good web search engine; using PageRank is useful, but nothing more.”