This has the potential to be a wildly useful service. How many of you have had professional or personal reasons to seek a list of the top blogs on a new topic? I know I, and many people I talk to, find themselves in such need frequently. How do you access the new search? How well does it work? Read on.
How to Search Google for Blogs by Topic
The Google Blogsearch service has for a long time surfaced a small number of blogs related to any search query, above the list of results from a search of blog posts, or entries. This new search is different, though, and surfaces different results.
Do a search on the general web search interface, google.com. Then, click on the "more" link in the left-hand sidebar if "blogs" doesn't already appear as an option. Once you're looking at blogs from this perspective, then a new option will appear in the sidebar: search for posts or blog home pages related to your query.
I searched for ceramics blogs, semantic web blogs, cloud blogs, social media blogs and more. All those searches brought up pages and pages of clearly topical sources. I imagine that different searches may have different degrees of success, but this is Google - so it ought to work well.
How Good Are the Search Results?
The search results in this new search by blog feature look pretty good to me. The ranking of those results, however, seems questionable. The more closely tied the title of the blog is to your search query, the higher the blog shows up in search results. That's not the best indicator of quality or authority.
That said, in as much as I know about the topics I searched for, the top blogs in those fields definitely peppered the search results, to greater or lesser degrees.
What would I do with these search results? To be honest, I'd ask our fabulous staff developer (whom you may not steal from us) to whip up some scripts that pulled down the first 300 search results (or more) from any query I searched for, grab the RSS feeds from those sites using another script and autodiscovery, turn the feeds into an OPML file, upload them to Postrank, sort them by degrees of reader engagement, then chop off the top number from that list as suits your needs. Your mileage may vary, of course. I'd likely then use some other processes to make sure I wasn't missing too many obvious sites or including too many false positives. (And then I'd do some other stuff with it too.)
Now that the sun has set of the Technorati blog directory, and no one has done as good a job in its place as it once did, it's great to see a new option at all. The fact that full-text search is the method employed here, along with some patented analysis of the sites, is great. The patent that Slawski points to, (Indexing and retrieval of blogs, filed September 2005), "describes how it may create a 'hybrid document' about a blog out of information from both XML feeds, blog posts, and pages linked to from those feeds and posts such as profile pages," he writes.
That great. Add some ranking, some OPML export, and then we're really talking.
Long live blog search!