Daily Archives

One Article

Posted by Tom Cheesewright on

Today It’s Phone Numbers, Tomorrow: Relationships. Why Our Computers Will Know Our Friends Better Than We Do

A couple of weeks ago I was on BBC Breakfast talking about the stress that can be caused when our Facebook circles — mostly work and life — overlap. It reminded me of a paper from a recent conference I attended at Lancaster University where an academic was trying to address precisely this issue.

Overlapping online relationships is not a new problem: Google was trying to solve it with the introduction of Circles in Google Plus. If you’re not yet a user of this particular entry in the pantheon of social networks, Circles are a way of circumscribing the different segments of your life into multiple overlapping groups of contacts. You can choose which of your circles receives each status update/piece of content. Your family might see things your friends don’t, or you might have different circles of business colleague who are privileged with different levels of information.

This isn’t a bad solution but it is a little clunky. I could try to define my relationships with contacts in the rich, minute detail that truly represents every day life, but this would take a long time to set up properly, and make every post on the network painful as I tried to decide which circles ought to see everything.

Instead I — like many people — have historically maintained just three ‘Circles’, albeit using an even less sophisticated approach than Google Plus’s Circles: Facebook is for friends and family, LinkedIn is for work, Twitter is for friends and work. Anything I wanted to post in any of those networks I have mentally edited for appropriateness to the relevant network.

There are of course finer controls over access within the social networks — particularly Facebook with its own version of Google’s Circles — but the reality is that few people use them. Particularly when they are having a good time: the posts that happen late at night after a couple of shandies are likely to be the ones that cause the most issues with both your work and social lives.

So we either all need to become a lot more open, or we need a more sophisticated solution.

This is where the academic comes in. Dr Jose M. Such at the Security Lancaster research group is working on a system that can automatically determine our level of ‘intimacy’ with different contacts based on our previous interactions. And based on data from Facebook the group has already produced a software tool that can model intimacy with 90% accuracy.

This intimacy score enables a system to automatically determine the proper privacy policy between you and each other member of the network.

For example, it could decide whether I can access an item — say a photo — based on my relationship with the owner of the item. It could also take account of the intimacy scores between me and all the people tagged in the photo before it made a decision. I’d probably never see this calculation — the picture just wouldn’t appear in my timeline if the compounded intimacy scores didn’t justify it.

In the early days I think it’s unlikely we would allow such a system to make all the decisions for us. At the minimum we might want ‘Pull’ and ‘Push’ buttons alongside the ‘Like’ to adjust the intimacy level between ourselves and others. And we may need more than one dimension in the scoring: there are different levels of intimacy in the office environment that are entirely separate to those in our social lives. Apart from the occasions that the two overlap, which is where this whole conversation started in the first place…

Tom Cheesewright