When you look at how the PageRank formula actually works, you see that it isn’t necessary for Google to ever look at, or even consider, IP classes or domains. The PR formula measures the importance of a page within a web of pages.
While it may be beneficial to link to a page that is on a different IP or domain, it’s not the difference in IP nor the domain name that makes it beneficial. I believe you have this all wrong. It is the linking of pages contained within a different web that boosts the PR that is passing value into your web. Google never needs to look at your domain name or IP address, in the core PageRank formula those are unimportant attributes.
I believe that the concepts of “IP diversity” and “domain authority” are based on people looking at statistics and not understanding what is the cause and what is the effect. If you consider how Google utilizes the PageRank formula, you see that those notions of “IP diversity” and “domain authority” are based on statical probabilities, and as such are effects, not causes.
One could inject the possibility that “IP diversity” is used to calculate TrustRank, and that makes some sense. However, I believe that it would create many problems for certain types of searches. There would be many pages that would never rank due to link profile patterns that don’t match the statistical norms. Pages would be excluded, not for spam or PR manipulation, but simply for real-world statistical variances. Not a very good reason to exclude a page from the search results, is it? It doesn’t seem to me that Google’s engineers would use such a crude formula, I believe they are smarter than that.
