Similarity versus Compatibility →
A few weeks ago I met a new person that has a big passion for music. We had an interesting conversation and we connected our profiles on last.fm (if you don’t know already, last.fm is a social network which revolves around its music recommendation engine).
When you visit a user’s profile on last.fm, the interface displays an indicator that shows your musical compatibility with that user. If the musical compatibility is high, the indicator turns green and if it is low, it turns gray. When I visited this person’s profile, the indicator told me that our musical compatibility was very low.
Image credits: njekaterina
Even though the indicator promises to show your compatibility with the other user, what you really see is the degree of similarity between your profile and his. It “confuses” the things that two users have in common (i.e: their similarity) with their compatibility.
The funny thing is that it’s not only the indicator that does that, we also do it. We often (unconsciously) think that being compatible with a person means to have things in common with that person.
By definition, to be compatible means to be able to exist and perform in harmony. Having things in common may increase the likeliness of compatibility but it does not guarantee it. That’s why, the efforts of those trying to achieve compatibility by strictly seeking to match the things that they have in common are usually useless.
While similarities allow us to fire up quick conversations with other people, a strong focus on compatibility can strenghten our courage and desire to learn from them. It doesn’t really matter that we only have a few things in common: a few artists, a few books, a few interests. What matters is that there is an opportunity for us to learn from each other.
PS: I wonder how this applies to movie/music/etc recommendation systems. If most of them are only based on a mechanism that identifies similar features between movies that I saw and the past and movies that I didn’t see yet, they’re missing a lot. I’d be curious to play with a great recommendation system combined with a great discovery system.
