I’ve recently signed up for the beta of a new discovery service, The Filter, which tries to figure out what music, films, TV and web video you like, and then recommend other stuff you might want to try. It’s an early, pretty buggy version of the service, and although it looks reasonably interesting, it’s a bit early to judge whether it will really help or not.

There are quite a few services out there trying to do similar things: iLike, Last.fm, and the recommendation engines of online stores such as Amazon.com all try to figure out ways of making it easier to find stuff that will be to your taste, using a variety of different approaches. Amazon.com looks at what you’ve purchased; Last.fm allows you to graze the music of people with similar interests and uses its software to build up a picture of which artists its users consider alike; and both Last.fm and The Filter allow you to download little applets that monitor what you’re playing in iTunes or Windows Media Player and so build up a picture of how you relate to the music you’ve bought.

All these applications are nibbling away at modern culture’s biggest problem: infinite digital choice. It’s really very hard to get your head around the scale of the pickle we’ve created for ourselves by digitising all the stuff we care about. Digital culture involves the sorts of numbers most of us can’t possible relate to. I have a few friends who are into their music who have thousands of CDs in collections that they’ve built up over time. I know a few more that have very large libraries. When you confront such collections in the real world they’re exciting, intimidating, and a bit overwhelming. How then can we possible relate to the millions of books, DVDs, CDs and games available via a single click online?

It’s not just about the overwhelming nature of the choice: it’s also about the terrible quality of the interface to that choice. When you prowl around your friend’s books or admire a tottering stacks of CDs, you’re in a marvellous 3D interface called reality which gives you huge amounts of information about the products: how well-read and dog-eared certain books are, which CDs are in box sets that have never been opened, which books are in pride of place by their bed. In contrast, the web is an incredibly limited and ineffective interface.

I pretty much hate Blockbusters for its narrow selection and onerous late fees, but if I walk into my local branch and spend three or four minutes scanning the aisles I can comfortably consume data on hundreds of movies. The experience of trying to rent movies from Lovefilm, which I pretty much love, is that it has a comfortingly huge selection of films that it is depressingly difficult to explore. Surfing tiny thumbnails of DVD artwork and book covers just isn’t that great. Using the web is like using a cash machine: great if you know what you want to buy, rubbish, ironically, for browsing if you don’t.

Many smart people have noticed this, hence ventures like The Filter and Last.fm which use the power of software, and the social capital of human networks, to try to help people discover stuff. It’s a tough problem. The experience of signing up to something like The Filter, in which your first task is to make a bunch of choices to train the software about your likes and dislikes, can be a bit embarrassing for everybody. These engines get better over time, so their initial suggestions are always a bit weak. (“You’re a sci-fi fan and you liked Alien? Well, you’ll probably like Aliens!”)

There always issues of self-image and disclosure around matters of taste: I say that I like Goddard and Woody Allen, you recommend À bout de souffle and Manhattan, but I’ve seen them both a hundred times and in fact I want to watch anything with Uma Thurman in, although I don’t really wanna scream about the fact.

There’s also something a bit creepy about seeing too much weight being placed on your previous purchases. Amazon.com’s recommendation engine sometimes feels like an overenthusiastic shop assistant fixating on the till receipt from your last purchase (“I see Sir has recently bought a book on Second Life and a pair of headphones. Would Sir like to see ten more books on Second Life and ten more pairs of headphones?”)

These discovery engines are trying to build a software bridge between the human and the digital. I hugely enjoy a regular editorial feature in the Observer Music Monthly called The Record Doctor, in which interview subjects confesses their musical tastes, and are then prescribed a bunch of new music designed to expand their frames of reference and introduce great new artists they’ve not yet encountered. As readers, we get to go along for the ride, and hopefully expand our frames of reference and meet some new artists too. That’s how recommendations work in the human world, and it’s that human power which social networking software solutions are trying to tap in to in their reference engines. On the other end of the spectrum is something database driven which records the hard data of what we’ve bought and listened to, and has the machine’s ability to accurately record and compare our choices to the vast existing databases of digital culture.

I suspect that recommendation and discovery software that works will find ingenious ways to exploit the power of both human and machine networks, that it will work quietly in the background without helper apps or much in the way of initial training, and that it will succeed virally, undramatically, and under the radar, as a discrete helping hand on the elbow. The Filter hopes to white label its service over time to other businesses, and that sounds about right. As we start to crack the problem the invisible hand of discovery software will nudge us as we browse and purchase online, compensating for the weakness of the computer interface and helping us tackle the problem of digital overabundance. I look forward to seeing how The Filter evolves. In the meantime, know any good Uma Thurman movies I’ve missed?

---

Michael Parsons, now Editorial Director, Consumer Media, for CNET Networks UK, spent five years working in Silicon Valley and worrying about technology. He can be reached at michael.parsons@cnet.com

Explore Personal Tech

• InGear