Jenny, the Shifted Librarian says today:
I think librarians love Google, but we’ll never trust pagerank enough to use it to verify authenticity. Of course, we take almost nothing on the web for granted, and that’s one major difference between us and your average web surfer. Librarians need our own pagerank system that we can control. [The Shifted Librarian]
I started thinking about this, and the best model I can see is the old PGP “web of trust“. The idea was fairly simple, but the problem it solved was complex: how can you safely rely on the digital signature of someone whom you’ve never personally met? If a signature was a fake (i.e., someone misappropriated another’s public key), then it was unreliable. Trust, in the context of the PGP web of trust, established both your certainty that the individual was who they said they were, and that the individual was reliable in taking care of her digital info.
If I trusted you, and you trust others, than my trust of you would be imputed to the others. By trusting just a few well-connected individuals, you would have a trust network on which you could rely. Pretty elegant.
In the context of the issue Jenny raises, it goes further. She points out that librarians need a system they control. This isn’t that much different than weblogs: how do I know the information that you’re presenting is reliable, authentic, etc? What we need is a decentralized approach – one in which I can establish my own trust parameters (i.e., “if Jenny links to it, I’ll trust it. If Bill links to it, I’ll never trust it.”) Google, as good as it is, is centralized – and the lack of independent validation of their methodology means that it’s inherently flawed when it comes to individual valuations.
In response to today’s article in The New York Times about Google’s search for a business model, John Robb suggested this morning that Google should move to the desktop. I wonder if what John had in mind was something that would address Jenny’s question: what if Google let me keep my own roster of trusted sites – so that Google’s own search results would be skewed according to my rankings?
Now that would be cool.