In 2011, our then-head of policy, Omar Khan, wrote on the term 'gang' and the UK riots. As the issue is again in the news, we're republishing the article unamended.

Runnymede has been urging caution on speculating on the causes of the riots that are now hopefully over in London and other English cities. Politicians have been somewhat more careful in jumping to conclusions than our media, but today the Prime Minister showed less compunction in identifying the causes of the riots, an incaution that unfortunately spread across the government and opposition benches.

According to Cameron: "gangs were at the heart of the protests and have been behind the coordinated attacks". While there is some anecdotal evidence that some criminal gangs may have taken advantage of the riots on the second and third day of rioting, the majority of rioters were obviously not gang members.

We have warned in the past about the explanatory usefulness of 'gangs'. As we put it in a recent publication, 'Rethinking Gangs', '"the gang" provides a potent shortcut to understanding youth conflict, offering Hollywood style images of urban chaos and random violence, threatening to spill out from inner city ghettos, in the place of more complex explanations exploring the realities of this phenomenon.'

The suggestion that US police familiar with handling gang violence will be asked to advise on responding to these riots goes against consistent findings that there is no evidence of US-style gangs in the UK. But the idea that policy should tackle gangs more directly is not limited to importing a US police official, nor is it simply a Conservative fascination. Indeed, policymakers' likely definition of a 'gang' will probably flow from that defined in a 2009 Act passed by the previous government. It defines 'gang-related violence' as follows:

'Violence or a threat of violence which occurs in the course of, or is otherwise related to, the activities of a group that:

a) Consists of at least three people;

b) uses a name, emblem or colour or has any other characteristic that enables its members to be identified by others as a group; and

c) is associated with a particular area.'

Under such a capacious definition, it's hard to see how any rioter could escape the charge of being a gang member. So have we woken up in Britain with tens of thousands of young gang members? Are there hundreds or thousands of localized gangs of 3-10 people, or scores of interconnected gangs in various parts of London, Birmingham, Liverpool, Manchester, Gloucester and Bristol? Such an outburst of gang activity is unlikely. More charitably, perhaps the Prime Minister was referring instead to criminal 'gangs' involved in more serious and planned activities. Yet from the beginning of the media's coverage of Mark Duggan's death, papers have referred to him as a 'gangsta', likely a cagey reference to his race as well as his putative responsibility.

Given these multiple, and arguably competing, definitions of 'gangs', Runnymede would again urge caution in using this term to understand the riots or the identity and motivation of rioters. As we put it in Rethinking Gangs, "the gang" has developed a public life independent of any empirical foundation or conceptual exploration, full of its own sound and fury, but signifying very little.'

We can and must do better in trying to understand why these riots occurred than using such slippery language: we owe it to the people who died, the businesses that were destroyed and the communities that have been torn apart by these events.