What's wrong with 'social cleansing'? This is the question raised by the hostile reaction to Newham Council's efforts to move housing benefit recipients to Stoke.

There's something to be said for this. It helps market forces to do what they should do - signal value. Houses should be occupied by the people who value them the most. If the state is paying for occupancy willy-nilly, the result is inefficiently high rents in some areas. Refusing to pay such rents might help force them down; remember, housing benefit is at least in part a landlord's benefit.

The argument against encouraging* HB recipients to move to Stoke rests upon the value of social capital. In moving from Newham to Stoke, ties with family and friends are weakened. But is this a good thing or a bad thing?

One might argue that it's an intrinsically bad thing. But I'm not sure. Yes, some people suffer from being parted from supportive family, but others might benefit from having an excuse to make a new start or move away from domineering fathers or ex-partners who harass them.

And economics tells us it's ambiguous too.

It's a bad thing to the extent that social networks can help get people into work. The more friends and acquaintances you have who are in work, you more chance you have of hearing about job openings. If you have to move away to where you don't know anyone, you lose such chances.

However, this is only true if you have friends who are in work. If most of your friends are out of work. You might lose hope, reduce job search and so have less chance of getting a job.

There's another potentially important effect - upon crime. It's well-known by now that our decision to commit crime or not is strongly influenced by our peers - hence 'gang culture.' One might imagine, therefore, that removing people from a criminally-inclined peer group reduces the chances of them falling into crime. It does. But there's a flipside to this. Peers also keep us honest; law-abiding peers, or admirable role models, can turn youngsters away from crime and towards education (pdf). Taking people away from their networks risks removing this positive influence, as well as the negative one.

If all this sounds ambiguous, that is precisely the point. Things such as crime, getting a job or pursuing education are examples of emergent behaviour. Whether people do them or not is sensitive to initial conditions - the precise structures of their social networks - small changes in which can have large, and unforeseeable, effects. This is one message of Mark Granovetter's threshold model (pdf). And it is why riots are so hard to predict.

If there is an objection to 'social cleansing', it lies in this - that, in breaking HB claimants' networks, it is a form of blind social engineering - of making changes which have potentially big but unforeseeable effects. I suspect that many people are relaxed about this, because they believe that the networks of HB claimants have bad effects rather than good. What worries me, though, is that this belief might owe more to class prejudice than to hard evidence.

* I want to avoid the issue of whether HB recipients will be 'forced' to move to Stoke, as this is a standard left-vs.-right question of semantics.