While studying Google's recently granted Historical Data patent [patft.uspto.gov], I noticed that the language helps to separate two concepts that we tend to use casually at times: trust and authority.
Clearly, Google has two different metrics going on. As you can see from the reference to Larry Page's original patent, authority in Google's terminology comes from backlinks. When lots of other websites link to your website, you become more and more of an authority.
But that isn't to say you've got trust. So what exactly is trust? Here's an interesting section from the same patent:
So we've got two references here, government documents and high traffic! From other reading, I'm pretty sure that trust calculations work like this - at least in part. Google starts with a hand picked "seed list" of trusted domains. Then trust calculations can be made that flow from those domains through their links.
If a website has a direct link from a trust-seed document, that's the next best situation to being chosen as a seed document. Lots of trust flows from that link.
If a document is two clicks away from a seed document, that's pretty good and a decent amount of trust flows through - and so on. This is the essence of "trustrank" - a concept described in this paper by Stanford University and three Yahoo researchers [ilpubs.stanford.edu].
This approach to calculating trust has been refined by the original authors to include "negative seeds" - that is, sites that are known to exist for spamming purposes. The measurements are intended to identify artifically inflated PageRank scores. See this pdf document from Stanford: Link Spam Detection [dbpubs.stanford.edu]
To what degree Google follows this exact approach for calculating trust is unknown, but it's a good bet that they share the same basic ideas.
So let's all work to keep these two concepts distinct - trust and authority.
[edited by: tedster at 6:14 am (utc) on Dec. 3, 2008]