I am not going to subject you to another rant about comparing Google search and Twitter search. Seriously. I’ve foamed about it enough; I’m sure you’ve gotten the idea that I consider them to be two very different data sets requiring two very different data strategies and I would never say that one was absolutely better than the other.
That doesn’t mean I’m not interested in search engines that take advantage of the speed and freshness of Twitter; I am. So I was intrigued by Topsy, a new search engine powered by real-time community data. Amazingly enough, Topsy does NOT claim to be in beta; it’s at http://topsy.com/ .
At first glance Topsy looks like a Twitter search engine, and it can indeed give you a lot of data about Twitter usernames. But it describes itself as more of an indexer of data from the “living social Web,” which includes things like blogs, Digg, etc. This first iteration is looking at Twitter only according to the About page.
Not having a good example search on hand, I did a search for hangover movie. The search results is a set of links, ordered by the number of comments which include that link. (In this case the “official Web site” for the movie was the first result on the link list.) Click on the number and you’ll get the comments about that particular link, all tweets of course.
Also on the search result page you’ll get an interesting breakdown of the number of results over time. In the case of Hangover, and at this writing, there are 1831 all time results, 810 this month, 308 this week, 28 today, and 1 in the past hour. This would be an interesting way to determine if something had peaked or not. On the right side of the search pages, you’ll also get a list of the most popular authors for this particular topic. This might work better for other searches — in this case the most popular author was Blake Rhodes, who was talking about The Hangover in the context of his own search engine, Icerocket.
Topsy does give you some trending results on its home page, but mostly you’ll have to do some experimenting yourself. It might be too much to ask for but I’d love to see Topsy provide related words which have a major impact on the number of search results. For example, Steve Jobs was noted today as one of the trending topics but the result count over time looks very pedestrian. However, if you change that to search to Steve Jobs Liver or Steve Jobs Transplant the search count gets a lot different, telling me those are major words, current words.
That might be harder to implement as Topsy starts to index more content. How will the real-time quick shots from Twitter hold up to blog posts? Is blogging really dead? Personally I think both content types will be able to peaceably coexist…