Plista: Ad-hoc social networks for product recommendations
Early demo of a collaborative filtering engine shown at TechCrunch party.
Of the 20 or so demos set out to bake in the afternoon sun on the August Capital patio for the TechCrunch party Friday, my award for the most interesting goes to Plista, a social recommendation service that follows what you like and don't across sites.
Plista currently uses a Greasemonkey script. Once you install it, when you go to a site in the Plista system--CEO Dominik Matyka told me there are about 30 so far--you'll find the page has new Plista pieces on it: A rating box on each element at the least, and possibly more content additions too.
By way of example, Matyka showed me Plista working with IMDB. I could rate movies on the site, and once I did, the service would show me other things it thought I would like based on the preferences of other people who liked what I did. What's cool about Plista is that the recommendation engine works across sites, so it'll recommend (in theory) books on Amazon based on movies you've rated on IMDB. And--also cool--it shows you your ad-hoc network of Plista members who have compatible tastes, so you can explore their recommendations.
I like the concept, and I like how easy it is for site managers to implement the system: They don't have to do anything. Plista does the lifting.
Unfortunately, I don't get how this service is going win wide adoption. There's a chicken-and-egg problem, for one: Getting a site Plista-enabled is a manual process, and while Plista can create the scripts for major sites, it can't reach into the long tail to code them all. Plista will need site owners to do some work themselves. And there's the egg: You can't use Plista unless you have Greasemonkey installed. While it's a popular engine for browser modification scripts, it's not a mainstream add-on, so the potential audience is limited. And who wants to bother coding a site for such a small number of users?
Worse, many of the most important sites that Plista would work on--blogs, commerce sites, and databases--already have their own rating systems that Plista would compete with. I don't see Amazon or Netflix adopting this system, for example.
I still like the concept here. Plista lets you rate products and content, gives you a lot in return for your rating activity, and it makes all your rating data yours; it doesn't lock it away deep in some secret database on the sites you visit.
Plista does have a fighting chance, I believe, if it drops the Greasemonkey tactic and instead focuses on developing a way for retailers and content managers to port their rating data to the service, in return for cross-site recommendations and the affiliate revenues that would come with them.
Plista isn't available yet, but you can sign up for the beta on the site.