September 14, 2005

Google Launches Blog Search

I just checked out the new Google Blog Search. I've been hearing about this for months, and now it's finally here. Of course, I checked searches I'm familiar with, to see how it does. One for my blog name (as key word) turns up many matches but also is missing a number of references I know exist. Not sure how they determine the relevance matches but on that one, for my own blog name, I come up the most. I would say that page after page of me referencing myself is not so useful, and I can't believe I'm more relevant than many of the people who link to me. But what that means is that they are likely matching terms as well as linkrank. It would be helpful if they filtered out the blog with the same name, or relevance searches for any blog that matches will just be useless lists of page after page of that blog's posts.

URL lookups are similar.. with many matches but many missing items as well. However, since I rarely link to myself, under a relevance view, I get lots of other people's posts linking to me. But wait, they are all strangely in reverse chron order, even under relevance. Date view shows the same exact list of results in the same order. Blog search is still in Beta.

The advanced search, under data range, appears to start March 1, 2005. Still, it's really useful to be able to isolate any period we want, and once there is significant historical information in Google's database, this will become far more useful.

I did about 20 searches on terms and links that I regularly monitor, and found that references created by a blog or blogger or concept associated with a blog show up as the top result(s), for those same searches. I'm not so sure that seeing blogs referencing themselves is so useful, and maybe these should be a third view, after relevance and date, where they provide *all results* including those kinds of references. Again, I'm seeing page after page of results that show the blog associated with my search term, instead of everyone else talking about that concept, blog or url. Not terribly useful as I can just go to that blog to see all its own posts. I want to see what others are saying about it. These searches are ones I have watched for years in my aggregator using 5 different feed search services, and I'm noting there are a lot of recent missing entries at Google as well.

I'll keep playing with it, but I think they have a ways to go in understanding blog search and getting the result sets right. And they need to pull in the rest of the results I know are there for all the searches I did. However, the search result layout is clean, and organized. It's a good start. And is very fast.

One more thing, Google Blog Search's FAQ says that it is searching RSS and Atom feeds only. It doesn't spider blogs. So it will only pick up what goes through a feed. Blogs that send through partial posts, will only have that part of the post in the RSS feed included in search results. This is a drawback, because the possibilities for finding topic communities are lost when a service doesn't spider blogs, because the full information is going to be necessary in order to figure out that problem.

Anil Dash has a good write up on the overall issues with this new search.

Posted by Mary Hodder at September 14, 2005 04:41 AM | TrackBack

Just tried my first search on Google's new blog feature. I loved how they time stamp the posts in the search results. Personally, I think this feature is going to make bloggers more accessible and more mainstream.

Posted by: The Nirvana Poster at October 23, 2005 08:40 PM