Hmm. There are already two pure blog search engines based on blog content (more precisely on RSS feeds) (http://feedster.com and http://rss-search.com). And http://daypop.comm as well as http://popdex.com and certainly several other blog tools also offer searches on blog content (in that case probably on normal web content). So Google would just be another one in that regard. The idea isn't really new. However, separating blogs from the main index would be fatal - because blogs are often more than just the blog itself. Not all content is offered there for syndication and is therefore available in RSS form (for example, the Python Desktop Server is to my knowledge the only blogging tool that creates an RSS feed for stories as well - with Radio, stories remain excluded from the RSS feed).
Let's see how it continues and what will ultimately actually be implemented.
At Der Schockwellenreiter there is the original article.