The problem with social software as a recommendation network has its roots in the problem of social software itself. "Friend" is a pretty blunt instrument when it comes to describing relationships, especially in matters of taste. The sad reality is that most of my friends have rotten taste in music (I don't hold it against them), while the music recommendations I actually follow are mostly from people I've never met, be they Rhapsody editors or MP3 blogs. Same for virtually every other narrow category where I need advice; odds are that the real subject matter experts aren't anyone I know. In other words, the assumption that there's a correlation between the people I like and the products I like is a flawed one. To use an analogy, Bill Joy, the co-founder of Sun Microsystems, famously uttered this truism (now known as Joy's Law): "No matter who you are, most of the smartest people work for someone else." The same might be said of recommendations. No matter who you are, someone you don't know has found the coolest stuff. Compounding the problem, the people whose recommendations I trust in music are different from those whose recommendations I trust in movies. Gadgets are yet another group of mavens, as are games and books. Indeed, although I have dozens of "trust networks" (usually formed by reputation and experience, not personal relationships), most of them have nothing in common with each other, and almost none of them I consider friends. Some of them aren't even human--they're software. In a sense, you can think of all your filters as being part of orthogonal trust networks, often with the only common member being yourself. They rarely, if ever, overlap. Thus any service that tries to condense all of your different planes of influence into a single dimension is going to fail, at least as far as useful recommendations go. That isn't to say that such services shouldn't offer playlist sharing and Amazon wishlists, only that I'm likely to find better advice elsewhere.
Thursday, June 23, 2005
Quote from The Long Tail blog. It's all about using referrer networks and working out good "recommendations" (eg. per Amazon product pages) based on other peoples tastes, p2p networks are related too by their nature. Very thought provoking.