I'm not sure to understand what you mean here. By directly, I guess you mean by crawling the web ? You're right about the fact that it would give access to more data, but, as Klout measure is people-centric, a social graph API is a more straightforward data source. Associating web pages to unique identity of persons is a challenge of its own.

> Klout were never going to gain the level of access the needed to the graphs on the various networks they utilized

Given their scale, access to twitter graph data should not have been a problem (they have enough users not to be bothered by the API's rate limits). Other social networks are much more trickier indeed.

Full disclosure: I'm co-founding an social media analytics tool providing more granular view of the interest graphs, using twitter as a data source.