The Filter Bubble Within Social Media

During the 2010 U.K. general elections, I, like thousands of others, was glued to my various screens as I kept up to date with one of the closest contest in living memory. This included the first ever televised leadership debate in a British election, during which Twitter acted as a default running opinion poll.

At the end of the election, I came out with the feeling that Twitter had generally been on my side, that is, supportive of either Liberal or Labour candidates and issues. And yet, when research was carried out, it turned out that Twitter had been like the election itself, just in favour of the Conservatives.

Why was my experience of Twitter different to the reality? The answer is simple, and obvious with hindsight: I had filtered my social media feeds and networks so that they were predominantly populated with people like me. This is what a recent book described as a filter bubble, and the effects of such filters are becoming more and more pronounced.

Such filters can be built or triggered by any number of things:

• Google builds filters based on your social graph and your click data, personalising results to such an extent that one +1 or tweet can bump a page from 40 to 40 in the SERPs.
• Facebook’s EdgeRank pushes content and people it doesn’t think you’ll be interested in to the ticker, leaving the feed free for the most “relevant” information.

All of this is done in the name of relevance, but it has major implications for our understanding of the world around us, not to mention how marketers connect with consumers going forward.

For a start, it’s likely to create a self-perpetuating echo-chamber where, like I did in the election, we surround ourselves with people and things that match our existing worldview and belief structure, not just because those are the people we know, but because those are the types of people and things that social networks and search engines recommend to us.

When you add this to the move in advertising whereby engagement is used as a success rate, and therefore a targeting tool, it could have massive implications for customer service and brand stature issues.

Imagine that I have a problem with brand X and spend some time complaining about it. Social network Y might decide that I am most likely to respond to similar status updates. And whilst it’s unlikely that a company like Twitter would purposefully push negative comments toward me because they match my own sentiment, it could certainly start allowing advertisers to target disgruntled customers of their competitors.

As Facebook and others start to build entirely personalised versions of the Web for us, based on our tweets, likes, clicks and friends, it could entirely strip the serendipity of discovery from our Web experience (and our wider lives, as everything we do becomes digital in some way), as advertising will almost certainly move in the same direction, as described above.

But despite all of this, there are still obvious opportunities for brands and advertisers to dare to speak from outside the echo chamber. As I described some time ago, in many ways what we need are social graphs that aren’t built around people we like, but people like us.

However, this won’t entirely remove the problem of the filter bubble. But if the net, so to speak, could be spread wide enough, it seems likely that an element of surprise could return.

Take Hunch; it’s a service that, after you have answered a load of seemingly random questions about yourself, can make predictions about your tastes and preferences with an almost spooky degree of accuracy. But what it can also do is predict that you’ll like things you’ve never heard of before. And as the data set, i.e. user base, it bases these on gets larger, the more likely it is that serendipity creeps back into our lives.

If advertisers are willing to be as brave, and dare to promote or suggest things that users aren’t 100% guaranteed to like or understand, then the role of curator, which is likely to be an important one as we enter the age of information overload, is available to them. And consumers are always likely to place more trust and value in people and brands that know them better than they do themselves, rather than in people who just parrot back things we already know.