The Google web search infrastructure consists of dozens of different services hitting hundreds of different databases, most of which obviously contain PII. We're talking about layer upon layer of abstraction and refinement. Thinking that there's one 'algorithm' to be open-sourced which tackles one of the most sophisticated software problems we've ever concocted is a bit juvenile, IMO :)