Monday, July 1, 2013

Social Recommendations, Discover-ability of the Truly New, and the Aha!

I'm sure you see a lot of "people who bought this also bought this" or "people who read..." or "friends who listened..." or "friends who watched...." nowadays on sites. That is social recommendation for a lot of businesses. The assumptions are clear. We are sheep. No, that sounds bad, but we're some similar creatures.

I kind-of agree. Most of us are not early adopters, and we seek sanction in the actions of other people. Why should we try untested waters unless we know people who jump in them come out alive, or better, enjoy it? The friends part makes even more logical sense. If we're sheep, we follow sheep and now flies. We make friends based on some commonality, and if my friends like something, maybe I like it too. Better chance than trying out a product none of my friends like. Right?

Yes, except that most of us are also unique. Sometimes, actually most of the times, we are sheep. But some times we are deliberately not. We don't want to wear the same shirt as everyone else. We also try to be different. We want to seek our own New. We want the thrill of being the first to discover something cool which then, other people also like. Some times we want to hold on to what we discover and don't want others to know about the small cozy cafe we discovered because once they do, it'll stay neither small nor cozy. Also if you think about it from the other perspective, that of someone / something wanting to be discovered, you have a terrible cold-start chicken-egg problem.

No one knows you, no one likes you, so no one follows any one to your door. Seems like a dead-end.

Not on the Music Genome Project, not on Pandora.

Anyone who's used Pandora is a fan, or I haven't had the fortune to come across any other type. Pandora is a radio-station for those who still don't know, that will ask you for a couple of songs you like and then will go on to predict, based on what you like and what you skip, to progressively suggest songs that you love, but have never heard of. Pandora doesn't do the 'like' business. The Music Genome Project, in the background, describes each new song, through an algorithm on about 400 attributes (genes), and each new song is rated by a musician on those attributes again, and that is what it uses matches to match your taste. This means that if Pandora has figured out what kind of music you like, and there's someone who's written a song like that, you discover it, hear it, love it - without ever needing to hear about the band, or the genre, or the country or century it was written in - and without any of your friends having heard of it. For those who can (it's legally available only in US / AU / NZ I think), you must try it for genuine aha moments.

What I find even more amazing is that this stuff is possible in as complex a realm as music. What I find amusing, however, it how much simpler it might be for, say, books - and you don't see this happening. T-Shirt fans could be following the herd in styles and trying to stand out a bit in terms of the slogan, but book fans really look for that unheard of book that gives them the aha.

Recap, recommendations work. Social also works. Attribute-based cataloging also works. There are categories (like the undiscovered), and occasions (imagine gifting someone what everyone else is gifting), and people (the cult of the non-conformists) for whom a Pandora of things is a crying need.

It is more difficult to build this out, but this could be well worth its while. I'm sure, at least hoping, that there is stuff being built as we speak.


No comments:

Post a Comment

Thank you for sharing your thoughts with me. Be sure to check back again because I do make every effort to reply to your comments here.