The Filter - An intelligent, interactive portal for Music, Films and Web Video.

Rating: Yeah I know I like that artist, I already told you that!

True Social Networking works not because its been called Social Networking by some PR company or a tech company who have no other angle but to call their application a social networking one to raise VC.

Social Networking works for me because I learn to trust the opinion of friend, and once I do, when they recommend me a web clip, a piece of music or a movie, I WILL check it out. That for me is THE POINT.

Application and service developers have been developing automatic recommendation software for a while now. The idea that the end user can then find new music, buy new music, watch new films. All very commendable and let me say a great idea I believe. I have been using Last FM since 2006 (http://www.last.fm/user/remixboy/) and have played 41,733 tracks thorough it, so the concept isn’t new to me but like all things, new algorithms are written, different angles explored, new concepts built.

Today a new music, film and video clip site was launched. The Filter. It’s described as a unique discovery portal, “an intelligent, interactive portal designed to help users discover new music, film and web video, based on their individual tastes and moods.”

They say:

The Filter, the vision of music legend and technology pioneer Peter Gabriel, will provide an holistic approach to filtering out irrelevant content and filtering in content based on an individual’s taste and mood. The vision is to extend this technology to other content areas such as books, TV and podcasts to enable the technology to be a filter to everyday life.

“Many of us are drowning in choice, we need good tools to help us make smart decisions. Traditionally, we have had help from people who are more knowledgeable than ourselves, or whose taste we trust, and today we have expert systems to help guide us.” said Peter Gabriel adding, “The Filter integrates the best of both approaches, man and machine, and takes data learned in one area to help guide in another. For example, data about musical taste can help produce better selections in film, or someone else’s tastes - friend, celebrity, whatever - can be mashed up with your own to provide new and interesting discoveries. As well as being fantastically useful, this thing is real fun too.”

How does it work?

Users who visit The Filter (and undertake a brief taste profiling exercise) will receive a daily homepage of music, film, and web video recommendations, all personalised to their taste. Further usage of the site will enable The Filter to learn and therefore improve the levels of personalisation.

There are also dynamic filters which allow the user to discover more music and movie content that they will like, based on factors such as familiarity or era. Users will therefore be able to build their digital libraries based on The Filter’s interpretation of their tastes and moods. Embracing the movement towards data portability, the site also encourages the user to import their online profiles from other relevant sites, such as Last.fm or Flixster, to gain even more evidence about their entertainment tastes.

At The Filter’s core is a recommendation and discovery engine derived from a branch of Artificial Intelligence, called Bayesian mathematics. When The Filter’s engine is supplied with one or more items of interest, such as a song, actor, movie or web video, it delivers a pick-list of items that are statistically relevant by order of probability.

The Filter is unlike most tools, however, that simply render a raw pick-list of recommendations. Instead, the pick-list is fed through a number of filters to deliver the customised recommendations. The Filter then incorporates simple filters to bias results by personal taste, popularity, era, or genre. More complex filters can also be applied to capture mood, the influence of friends, third party experts or reviewers.

We Say:

So I signed up with The Filter today. Like the PR says, the first thing you are asked to do is choose genres of music and films and then rate artists/films based on your taste. There is a little slider which starts at 50% and you can pull it left or right to rate higher or lower. It doesn’t take long, and I have to say I liked the initial sign up. It felt like my choices were important. How wrong I was!!!!!

I was asked how do I rate Good Charlotte, U2, Air, plus others. All artists picked by the software from the genres I had selected. By now I was looking forward to finding new music. I LOVE finding new music, in fact I’m renowned for it amongst my friends. As I went through the sign up process I thought it only fair to give the application as much info as possible, so instead of just picking 5 artists I like as recommend by the instructions I picked more.

At the end of this process and having entered an email address and password my home page appeared with my personal recommendations. Imagine my surprise when I was recommended artists I had already said I liked! Surprise surprise, I like Air and got recommended Air, I rated U2 above 50% and got recommended I listen to U2!!!!!

I was not impressed. I could have easily have stopped at this point but I have a techy background, and a sneaking admiration for PG and his songs, (after listening to the Steam Remix me and a friend spent several days programming 50 odd tracks of percussion on an Atari Pro 24!!)

Anyway so I thought I would give the system more help, and told it more of my choices. No Change. Ok, so then I thought I would import my Last FM preferences, this is supposed to help the system suggest new music. Still got recommended U2, Air, Good Charlotte etc etc.

So then I downloaded the The Filter widget, which basically integrates with your iTunes, Windows Media Player or Winamp and tracks what you play and the guys n gals at The Filter have the software use that data to improve your recommendations. Now as I write this, the widget is doing its work..its a very slow process. Much slower than other Widgets I have used for this purpose.

It hasn’t even got a quarter distance yet but guess what? My recommendations are still the same. Green Day, Blink 182, U2, Air etc. All artists I had already rated. No new music, what so ever.

I am hugely disappointed. It can be slightly argued that these things take time to analyse your choices and over time build recommendations and I have some sympathy for this approach, but I had already loaded in my preferences from Last FM with a history of 40,000 odd tracks already played and it didn’t propose one new artist to me.

The Filter.com lives or dies on its recommendations, after all that’s it’s main purpose. Like all the other companies doing this, the idea is solid, but the results from The Filter are poor and on this test, its going to be a long time before I find any new music via The Filter. Ill stick to personal recommendations and the time old process of listening first at this stage. Humans 1 Software 0.

Related News:

1. The Filter: extends mobile music recognition to the Nokia E60, N93 and N80
2. The Filter: mobile music search, recommendation and discovery
3. Vodafone UK to deliver cheaper off-portal music tracks via SMS
4. Nokia World 2007: Mobile Music - Nokia comes with music?
5. T-Mobile takes advantage of Last.fm directory for mobile music search service