This morning there are two new entrants in the “real-time” search derby, run by two search veterans. They are CrowdEye and Collecta. CrowdEye is from Ken Moss, who ran search engineering at Microsoft and built the new engine himself. At the helm of Collecta is Gerry Campbell, who was a search executive at AOL and Reuters, as well as an adviser to Summize (now Twitter Search). He recently stepped into the CEO role at the company.
The two new engines join an growing field of competitors, including Topsy, OneRiot, Tweetmeme and Scoopler, not to mention Twitter Search itself. Then there’s Facebook, which is testing an upgraded internal “real-time” search capability, and Google, which has all but formally announced that it’s getting into the segment.
CrowdEye is indexing Twitter Tweets and pulling out relevant third party links and related queries. Here’s an example for Iran Election:
In the upper left are editorially selected, related queries. They will become dynamically generated over time. The tag cloud in the middle features terms pulled from Tweets about Iran Election, that initiate new searches. The chart above the tag cloud shows activity over a three-day time frame. You can click on the bars and drill down into the activity stream at a given time or hour. In the upper right, “popular links” are third party links that the Tweets reference, algorithmically selected for relevance.
Moss told me that this is “V1″ and he sees it expanding in numerous ways: more data sources, a wider than three day time window and so on. Right now Moss isn’t building an archive or index longer than three days and is thinking about what to do with older material as it accumulates.
Collecta, for its part, is already drawing upon a range of data that include Twitter but also blogs, articles, Flickr and other sources:
On Collecta, the left pane enables users to select sources and “shape results,” according to Campbell who says that he wants to add more filtering and “shaping” capabilities in the future.
The middle pane offers a real-time vertical scroll (which can be paused) that moves down the page as new content appears. The right panel is a preview pane.
Campbell makes a distinction between “real-time search” and “timely search.” He contends that Collecta is currently the only real-time search site, pushing content to users from the various sources it draws upon in less than a second after it appears online. CrowdEye’s Moss says that results appear on the site essentially as they’re being published, on Twitter via the API.
I asked both Campbell and Moss about use cases and whether “ordinary searchers” would understand how to use these engines. Campbell said that “events” and “places” were the primary use cases for the time being — Iran Election is an example of both. Yet both Campbell and Moss expressed optimism that “real-time search” will become a mainstream activity in the relatively near future.
Regardless the segment is arguably the hottest in search and there is now a pack of startups that claim to offer “real-time search” capability with Google and Facebook circling overhead. The concentration of activity in the space probably means that it is here to stay and that it will become a part of traditional search — although Moss believes it’s a vertical and won’t be building a complete web index.
Among the differences between traditional search and these new search competitors, we can expect them to evolve and “iterate” very quickly. After all, they’re playing in real time.