I have some great friends, no doubt, but sometimes I prefer strangers. When being given recommendations online for anything from books, to restaurants, to baby furniture, I often heed the advice of complete unknowns. Thanks to various tools like collaborative filtering used by Amazon (i.e. recommendations based on others with similar purchasing habits), rating systems used on retailer Web sites like Home Depot and Toys “R” Us, and wisdom of the crowd approaches like Rotten Tomatoes for movies, I now benefit from the collective opinions of thousands of people that I have no direct relationship with. This may seem like an obvious point to many, but many Web 2.0 services being launched are very much focused on relationship capital and friend networks.
In my mind there are two approaches that companies seem to be taking to generate recommendations and help users navigate various types of information: trust networks and collective intelligence. Trust networks are based on your social graph (i.e. your online relationships) and therefore use your friends to help recommend content. Facebook Connect is a popular method of leveraging trust networks by porting your contacts into other applications (or other applications to your contacts). Collective intelligence, on the other hand, is a much more data-driven method of rating and recommendation.
The Twitter vs. Facebook debate provides another example of the relative merit of these two approaches. People often lump both social networks together because they both have status updates that detail the minutia of our everyday lives. But how these services are used is very different. Twitter is a platform for weak ties (i.e. more suited to collective intelligence) where Facebook is a platform for strong ties (i.e. by definition, a trust network). As Denis Hancock notes, “Facebook if for people I’d actually let into my house.” My Twitter contacts, on the other hand, are dominated by people I have never met and will likely never meet.
The important point is that both types of platforms are useful. While Facebook lets me keep in touch with friends and organize social events, on Twitter I find all kinds of interesting links to articles that I never would have found by polling my friends. Similarly, Twitter is a valuable platform from a market research standpoint because companies with the appropriate listening tools in place can now mine the collective opinions of millions of customers, prospects, and influencers. The argument that Twitter is a waste of time because, “why would I care what everyone is doing at every moment of the day” is valid, but the emergent trends and memes resulting from ‘what everyone is doing at every moment of the day’ can be extremely valuable.
This brings me to an interesting article I read on TechCrunch a few days ago: Location’s social paradox, which states that the main problem with location services is, “The more people you follow on them, the less useful the service is.” This is essentially recognizing the fact that for certain services—particularly those where you give up your physical location—the exclusivity of your close network of friends trumps the benefit gained from having access to many individuals. As such, most of the recent offerings (such as Foresquare and U.K.-based Rummble) rely on trust network – login and we’ll tell you what your friends are up to and what they recommend. In these cases, I would agree with MG Siegler from Tech Crunch who suggest the more friends you have on this type of network, the less useful it can be.
However, the potential to use these same services for collective intelligence applications is huge. For example, when you are considering a particular restaurant for dinner, you would be remiss to only consider the opinions of the few friends in your network that have been there. More powerful would be real-time access to the aggregated reviews of as many patrons as possible. Similarly, when driving on the highway, it may not be terribly useful to know that three of your friends are on the same highway, but it might be extremely useful to understand the traffic patterns of all the other cars on the road via the GPS signals from anonymous drivers’ cell phones. In short, social network applications that incorporate elements from both trust networks and collective intelligence (perhaps with the ability to toggle between the two or shut one off based on context) will be most valuable.