Social Network Fuzzy Logic (ie here comes FuzzBook)

Some days ago Paul Clarke and I were talking about how we were sure that people who followed us on Twitter were being dropped, and he came up with the interesting observation that it may be deliberate - here is his blog post explaining this, summarised below:

Real relationships aren’t binary. They’re analogue. You can like someone not at all, a bit, or a lot. That can change from day to day – sometimes from hour to hour. Independently of how much you like them there are other factors involved like distance and frequency of contact. You might adore each other but only communicate once a year.

Social networks can (so far) only provide the palest echo of this rich texture. You’re either someone’s Facebook friend, or you’re not. Twitter’s a bit more subtle in its branding of the relationship, but we’re humans. We’re tempted to attach emotional significance to everything to some degree.

It all comes to a head with the Unfollow Problem

Unfollow me? You mustn’t like me any more. I’m sad. I don’t enjoy this experience much. Best keep away from it.

And to those who do the unfollowing and reap more than they bargained for, this brings its own problems. Retaliation. Icy silence. Worse. People will interpret the same fact in countless different ways. We don’t all operate according to the same textbook of emotional responses (mercifully).

So if you’re a savvy social designer, you want to design the sadness and badness out where you can. You want to keep your community happy. You want to keep your community there. So you need loopholes. Get-outs. And you quietly introduce a random unfollow ‘bug’. Just a small one. Perhaps 0.1% of relationships ‘accidentally’ broken in a month. Not enough to reduce confidence in the integrity of the system.

But enough to offer a face-saver to the unfollower. And a hope-giver to the unfollowed.

Now a few years back we (as well as many others, I'm sure) looked at how to design Nuance in a social network (I just hate the term "Social Graph", with all its geometric arcana so I'm not going to use it).

In essence, how do I design a choice architecture so that:

- I can differentiate between Mother, Lover, Work Colleague and Friend (say)
- The system is capable of following layers of simple rules to decide what content is desirable for which category
- The system is very easy, ie intuitive, to use

You don't go far before you realise these are opposites in design (like cost, quality and time) - too many categories and the rules get complex and the system is no longer intuitive. Too few and the rules can't be nuanced. Too "intuitive" (ie simple) and the nuance is gone.

What we also found was thet you needed Fuzziness - ie the acceptance that the edges are blurred. Paul has hit this with his observation about Unfollow as well.

The resultant thought is that these systems will have to be designed with Fuzzy Logic in their next generation