The Filter was officially launched at the beginning of June 2008 after almost two years in beta and has grown into one of the largest content filtering services on the Internet. While we’ve seen collaborative filtering of this sort on more than a few e-commerce sites, The Filter is focusing on all levels of entertainment and promises to be the best means of discovering the latest trends. As of this writing, The Filter claims to find music, movies, TV programs, and Web videos geared toward an individual user’s personal tastes better than any other service out there.
When most of us use Amazon.com to purchase a product, we can’t help but take full advantage of the customer reviews that are placed there to help us make better buying decisions. The site has become a community of sorts where users can share their thoughts about the things they buy. This inevitably makes more people want to participate in the “fun,” whether it be lamenting or glorifying a product or sizing up their next purchase. And now, Amazon uses information about past acquisitions to recommend other items they think you would be interested in obtaining. Netflix is another popular service that uses collaborative filtering based on user actions and recommendations to relieve their customers of the overwhelming feeling of choice caused by a vast inventory.
In fact, Peter Gabriel explicitly wanted to create The Filter to help combat the tsunami of choice the Internet has become. He says, “the first freedom the Internet brought was the possibility of access to any content, at any time, or anywhere. Now that many of us are drowning in choice, we need good tools to help us make smart decisions.” Ultimately, The Filter will be able to provide recommendations that cross individual media types: a purchase of Fight Club will generate a recommendation for Era Vulgaris by Queens of the Stone Age.
The Filter seems to be the perfect solution to our unwavering capitalistic appetite. Now that we can spend less time sifting through what we don’t want, we can spend more money on things we don’t need. In the next section, I’ll tell you how filtering is to social media addicts, what drive-thru is to fat people.
The first step to recovery is admitting you have a problem. Hi, My name is Bob…and I am a social media addict. Your choice of social media is growing by the day: IM, blogs, RSS, SMS, Twitter, Facebook, MySpace, etc. Many of us have so many sources of information that several hours a day can be wasted while we’re engrossed in these social media tools. The other day I was so desperate for information that I started licking between the crevices of my keyboard.
It’s lucky for you, then, that filtering is being incorporated into more than just e-commerce sites. I’m talking about social media aggregators, like FriendFeed. You can either spend less time searching for what you want or more time looking at more than you would normally. It depends on whether you have a stubborn passion for specific interests or you enjoy expanding your horizons. I would say most people would choose the former, but filtering can also help people on the other end of the spectrum: people who are curious about the interests of others, but who can recognize trash when they see it.
I’ve already somewhat explained collaborative filtering, but the actual definition is even more ambiguous than even I’ve managed to make it. There are actually three main types of collaborative filtering: the first, Active Filtering, involves people of similar interests rating different items and sharing the information with other people. One of the problems with this, however, is that it relies too heavily on people’s biased opinions, which are also hard to come by. Passive filtering is when a web browser collects information about a user by recording their actions (views, purchases, links, etc.). This provides a much larger data sample than just active filtering. Then there is item-based filtering, which uses ratings to group various items together for comparison (think Amazon).
With better filtering options on these aggregated services, some of the noise that results from adding accounts from unrelated social platforms can be quelled. But don’t think you still won’t be able to aggregate as many accounts as you want. In the meantime, you can practice the art of self-filtering by taking other people’s preferences into account when aggregating another platform.
Click ahead to the next section to find out what The Filter is doing to incorporate the latest in filtering technology.
The Filter works by combining all the collaborative filtering methods I mentioned in the last section. Another way it gathers data is by allowing users to import profiles from other services, like Last.fm and Flixster, or data from their computer by downloading a plug-in. The plug-in application can work with iTunes, Windows Media Player, or Winamp. It sends the information back to its servers and uses it solely for improving user recommendations. “We are very keen to integrate other music-only, movie-only or TV-only services that make The Filter even more useful for people,” says CEO David Maher Roberts.
It is yet to gain the capabilities of other plug-ins, like iLike, which can show you other users with similar tastes that you might want to befriend. However, the Filter does have a social network component that allows you to add friends via email and communicate with them either through onsite mail or a wall similar to that of Facebook. It’s important to remember that this site is still somewhat new, so more social features can be expected in the future along with other media types, such as books, as long as the data can be measured electronically. As of this writing it has over 5 million songs and over 300,000 movies.
When you first sign up on The Filter, you must first complete a somewhat brief profile wizard that helps the site make some initial recommendations. You can mark items that you already own so they won’t be recommended again. You rate items by using a +/- sliding scale (with little blurbs explaining what that position would mean in terms of a rating) as opposed to a number-based scale. All of these little tasks help The Filter get to know you better.
Several different filters based on metadata are used to optimize the results. Roberts says the system “crunches evidence in the form of purchase data, consumption data, and browsing data to work out the strength of connections between content items.” Ultimately, he wants to be able to filter through input gathered outside the active user, such as the user’s friends, product experts, and even celebrities.
In the last section of this article, I’ll go over some of the major goals The Filter aims to achieve in the near future, including a new form of advertising. Oh, but I’m guess you saw that one coming a mile away.
As of this writing, The Filter has very little advertising (mostly an attempt to get people to sign up). This is probably because the site is still in its infancy and hasn’t garnered enough traffic, but because the site is able to target users of specific tastes, there’s no doubt that advertisers are intrigued. Roberts does say, though, that the updated versions of the site will always give users the ability to opt out of any personalization that is filtering taste specific ads. A nice gesture, especially for a site that’s designed to recommend products for its users to buy.
While I’m sure advertising is an important step that The Filter’s top brass want to take, their first order of business will be to augment the site. For instance, right now The Filter doesn’t provide complete song tracks to its users. “We have focused our time and resources on developing algorithms and on making sense of people’s tastes. In terms of consuming the recommendations, we have a thirty-second clip deal which enables people to get a taste of the track and then we have partnerships in place for people to click through and buy,” Roberts explains.
Pandora is probably the benchmark in terms of a music recommendation service that plays songs (full tracks) similar to the user’s favorites. Created by the Music Genome Project, Pandora also allows users to purchase songs or albums on Amazon or iTunes.
Then there’s the social factor that I mentioned in the previous section. Roberts says users will soon be able to import data from friends on Facebook and MySpace. Something similar to the “mini-feed” could also be used by users to track the their friends’ latest discoveries.
Now matter how they use it, by combining all aspects of collaborative filtering, The Filter has the chance to offer the best media recommendations of any e-commerce site. And since most people’s purchases come by way of recommendations, if users feel they are getting better service, The Filter has the potential to overtake some of the Internet’s biggest players.