If I search for "java string api" (no quotes) on Google, Bing, or Qwant, they give me a link to the documentation ... for version 7. We're now on version 19, which doesn't show up on the first page.
Why? Why would a search engine do that?
@peterdrake it looks likse in version 8, the jdk documentation has a lot of inter-documentation links in pure html, unhidden. They are all visible on every page without having to Interact with the page at all. Search engine algorithms have ways to compensate for this nowadays, by day running a JavaScript engine against the page to make sure every component of the page loads. But this is more cpu intensive to do than simple html parsing
1/
@peterdrake * sorry above comment should read "Java 7", not 8.
@peterdrake so it could be the spidering algorithm favors the site that is easier, less cpu intensive to parse. That favor could be passed on to the ranking system to encourage sites that parse out faster
@peterdrake This is all pure speculation. But search engine optimization experts are paid a bunch to make these kinds of tea leaf readings all the time. But the truth is, no one really knows. I suspect even the engineers working on Google's search term don't even really know why it works this way as it's probably all buried under layers of machine learning honed by years of feedback loops from user interaction to ML model refinement which changes user interaction in subtle ways, etc.
@peterdrake The older docs likely have more links to them than the new versions. At it's core, the pagerank algorithm will rank sites higher the more pages there are that link to them.
That said the algorithms used by modern search engines are vastly more complicated than the algorithm discussed in the original pagerank paper and likely have ways to compensate for this. I'm not sure how that might work and how it might go wrong in this case.