Causal responsibility and moral responsibility
On a different topic, there was a decent Metafilter comment a while ago that highlighted the difference between causal responsiblity and moral responsibility. I think that's an important point to start with, because a lot of the articles about the riots are getting these things confused.

Let's suppose that a woman is attacked and raped while walking home from a nightclub at 3AM. Let's also suppose that if she'd taken a black cab home instead of walking, she would not have been raped.

Now, the decision to walk instead of getting a taxi is part of chain of causes. Breaking that link would have meant the event did not occur. However, moral responsibility for the event lies entirely with the rapist. His decision to commit the crime was also an essential link in the chain of causes, but it was his decision that was the moral decision. Therefore he is the one who is morally to blame. Furthermore, the moral blame is wholly, entirely, 100% his. It is not shared with the victim.

Similarly, with the riots, the moral responsibility for the riots lies wholly, entirely, 100% with the rioters. Saying that does not mean that there are no elements of causal responsibility which lie elsewhere. With different police strategies, parenting styles, weather, distributions of wealth or role models; the chain of causes leading to the riots might have been broken. But none of those things change the fact that the moral responsibility for a riot lies with the rioters.

I think this is worth mentioning because the media, and TV news in particular, are enjoying the ability to conflate causal and moral responsibility. TV news anchors like to ask people for the causes of the riot, then when they reply, angrily accuse them of "making excuses" for the rioters. This doesn't necessarily follow: it's possible to consider that there are causes beyond the rioters themselves, without shifting the moral responsibility from the rioters.

Since it's very difficult to change the moral character of other people, it's often more sensible to look at the practical causes instead. If you think your car may be stolen and choose to fit an alarm to it, that doesn't mean you're accepting moral responsibility for the car theft. It just means that you've looked at the chain of causes leading up to the theft, and have decided that's a more practical link for you to break.

Background
For some context, it's worth looking at the Wikipedia list of London riots in history. From that, it should be pretty clear that social media sites are not essential to riots. For instance, the Bawdy House riots of the 1660s show that rioting can be widespread across different parts of London without advanced technology.

The riots were triggered by different causes: price rises, taxes, racial tensions, political reform, resentment of foreigners and so on.

Beyond London, there is some research looking at the causes of riots in general. Marginal Revolution mentions several studies. DiPasquale and Glaeser find that per capita GDP is negatively correlated with rioting, unemployment is correlated with riot occurrence, but that poverty in the community has little effect, ethnic diversity makes rioting more likely, as does the opportunity cost of time and the potential costs of punishment. A study by Bohlken and Sergenti of Hindu/Muslim riots in India finds that a 1% increase in economic growth leads to a 5% decline in riots.

In a paper called "Austerity and Anarchy" Ponticelli and Voth (author article, gated paper) examine the link between budget cuts and unrest. They find:

Once expenditure is cut by more than 2% of GDP, instability increases rapidly in all dimensions, and especially in terms of riots and demonstrations. Severe cuts – of 5% or more, as in Greece today – are associated with the highest level of instability...

Tax changes do not have the same effect. While they are also associated with increasing unrest, the link is weak and could be due to chance. We also test if the relationship between cutbacks and instability that we document simply reflects that both are more likely in economic downturns, and conclude that this is not likely.

English Riots of 2011
With the background in mind, we can start looking at specific causes and solutions to the English riots of August 2011 in particular. When looking for causes, we should bear in mind that these riots happened in 2011 after more than a decade without major riots, and that there have been periods with intense rioting at many times in the past.

Moral explanations
Several commentators have put forward primarily moral explanations for the riots. Two articles in particular have been widely circulated Melanie Phillips says: "a liberal intelligentsia has smashed virtually every social value".

Peter Oborne says:

...the criminality in our streets cannot be dissociated from the moral disintegration in the highest ranks of modern British society. The last two decades have seen a terrifying decline in standards among the British governing elite.

I think that both of these fall into the trap of confusing moral responsibility with causal responsibility. Yes, the rioters are morally responsible for the riot. But that moral choice is just one link in a causal chain. Someone may be morally willing to commit a crime, but choose not to do so for many years, until he sees the chance to get away with it. When the crime happens, it doesn't mean that he has suddenly become morally bad. He was always morally bad, but fear of the consequences restrained him until now.

So, the fact that the riots have broken out now, doesn't necessarily mean that the riots are a symptom of moral decay in society now. Clusters of riots do occur in history, and we can see from the studies above that riots can start to happen for non-moral reasons. So while I can't disprove the assertion that moral decay in society is responsible for the riots, there doesn't seem to be any actual positive evidence that it is responsible either. Until Oborne or Philips produce at least some correlations between rioting and liberalism or moderate parliamentary corruption in history or worldwide, I'm inclined to reject their theories. It seems to me more that they're latching onto the riots as a convenient consequence of their usual bugbears.

Another set of explanations looks at ethnic divisions as a source of the riot. This seems more plausible. The DiPasquale and Glaeser finds an actual correlation which supports the basic principle. The first riot started at a protest over the death of Mark Duggan, by a largely black crowd.

However, as the riots spread, this connection seems to grow weaker. The rioters in subsequent days expressed no interest in Mark Duggan: didn't chant any slogans or cite him as a motivation. The rioters in other parts of London were more racially mixed. In Manchester and elsewhere in the UK, the rioters were predominantly white.

Ethnic explanations
The chief motivations of the later rioters as individuals seem to have been the fun of destroying stuff, and the desire to loot goods. From pictures, some of the groups of rioters were racially mixed.

There was some degree of tensions between ethnic groups. For instance Asian or Turkish shopkeepers defended their shops against white or black rioters, the notorious car-crash killing, and the EDL massing as a "defenders" in Eltham. However these seem to have been relatively late and not occurred in a majority of incidents. While there was immense property damage, there was relatively little fighting between groups of rioters.

Overall, there seem to have been a couple stages to the ethnic development of the crisis. In the first stage, the initial Tottenham riot had a racial dimension. The second stage was an almost colour-blind day of cross-ethnic mass looting.

We may have been lucky that the riots ended when they did. If we had reached a stage where there were clashes between different groups which entered into a cycle of violence where each group retaliated in revenge, it could have proven very difficult to end that cycle. Northern Ireland shows how hard that can be be. If the EDL had ended up fighting black gangs on the streets, we could have ended up with a long-running feud with a lot more human violence than actually happened. However, if riots erupt again, this remains a possibility.

Police and the media
The initial trigger for the Tottenham riot was the protest over the shooting of Mark Duggan. The media initially reported that Mark Duggan opened fire on the police and they returned fire, the evidence being a bullet embedded in a police radio. The Tottenham crowd did not believe this story. It later turned out that the embedded bullet was a police bullet, and the gun found had not been fired. The official IPCC statement didn't mention a shootout, but they later stated:

The IPCC's first statement, issued at 22:49 on 4th August, makes no reference to shots fired at police and our subsequent statements have set out the sequence of events based on the emerging evidence. However, having reviewed the information the IPCC received and gave out during the very early hours of the unfolding incident, before any documentation had been received, it seems possible that we may have verbally led journalists to believe that shots were exchanged as this was consistent with early information we received that an officer had been shot and taken to hospital.

Any reference to an exchange of shots was not correct and did not feature in any of our formal statements, although an officer was taken to hospital after the incident.

If this misleading information was just an isolated mistake, it wouldn't be a big problem. However, it seems to fall into a pattern. When the police shot Juan de Menezes, all the press reported that he was highly suspicious because he was wearing a bulky jacket and vaulted the station barriers. Neither of those things was true, but they dominated the reporting. When they killed Ian Tomlinson, the police initially claimed that they were attacked by "a number of missiles" which later turned out not to be true. The police also have a somewhat unhealthy relationship with the media.

It's important to remember that the Metropolitan police employ 45 press office staff. This is not a small, amateur operation that's likely to make elementary mistakes. It seems to me they have a deliberate policy of controlling media spin by putting out unreliable, favourable information, on the assumption that this will dominate the front pages; while the retractions will be buried on the inside pages after the story has ceased to be big news.

If so, this needs to be changed. It may be a good way to shape the story in the short term, but over the long term it's disastrous. It's very difficult for older members of the community to debunk wild rumours, when everyone knows the official line is likely to be a lie.

Police tactics
There have been a lot of calls for the police to be equipped with water cannon and rubber bullets. Water cannon have some use in some riots: they can disperse a crowd, and they can protect a building in a protest. However, as basically a tanker truck, it's not that easy to move them around to deal with roving groups of people, and it seems unlikely that they could be deployed to handle multiple crowds in multiple locations. So it seems to me unlikely that they would have been much use in these riots.

Baton rounds or "rubber bullets" may have some use, but they can still be fatal, and in Northern Ireland the resentment when they killed people was a significant problem. I think it was probably sensible not to use them.

When large numbers of police were deployed onto the streets of London, the rioting ceased completely that night. I don't think this was a coincidence. Putting enough police on the street seems to work. The question is: is it affordable to keep enough police deployed overnight in the long term?

16,000 police were deployed that night, by cancelling leave and using police from outside London. The Metropolitan police themselves employ 52,111 people of whom 33,258 are police officers, 4,226 are Specials, and 4,520 are PCSOs. Now given everyday duties and leave, it's certainly not possible at the moment to keep 16,000 police on the street every night. The cost of the riots is estimated at £200 million, which works out at £12,500 per cop. So if these riots are only likely to happen once a year or so, it's probably cheaper to let them happen than hire this many police and have them on duty all the time.

However, the police bloggers have been complaining for a long time about a lack of response officers at night. Given that we know things like these riots can happen, and that police officers can help, it seems like a good idea to increase the number of police officers on duty at night to some degree. Implementing the current plan to actually cut police numbers seems like a pretty bad idea.

There have been some reports, which are officially denied , that the police were ordered to "go soft" and simplu stand and observe the rioters. If that happened, it seems like a bad move. I'm not convinced that it did happen. More likely they either didn't have the numbers, or they weren't willing to tackle aggressively hostile crowds.

While there have been some calls for the British police to be permanently armed, I don't see much evidence it would have helped in this case. There are plenty of firearms officers already. If the problem was that there weren't enough of them, if they'd been frantically rushing from scene to scene, that would be evidence that the police need more guns. But none of them used their guns in the riots at all. So, I don't see how having more of them would help.

Who are the rioters?
It's still hard to get a picture of who the rioters were, though some details are starting to emerge from the arrests. Also, it's not certain that those arrested are representative of the actual rioters. It's been suggested by some that the arrested group will have more middle-class members than the rioters as a whole, sin the "underclass" would be better at concealing their identity. This is possible. But it's also possible that it's the other way around: that the middle-class rioters were more likely to conceal their identity, having more to lose in terms of career and reputation. The latter seems more likely in view of DiPasquale and Glaeser's result that costs and opportunity cost are a factor: if so there is some rational risk assessment going on.

The Guardian has a datablog and BBC has some early data. As expected, the rioters are generally young and male. The BBC says in 70% of cases the crime was in a different postcode to the one they live in, so they weren't necessarily on their doorsteps. Many papers have commented on the middle class rioters including "a millionaire's grammar school daughter, a ballet student and an organic chef." While they are presumably a minority, this is not purely a riot of the "underclass" or career criminals.

Economic explanations
Every riot is made up of individual rioters. Each individual has their own mix of motivations. These include greed for looted goods, the adrenaline rush of breaking the law, anger and resentment at the rest of society, herd instinct to go along with the crowd on one side; fear of consequences and concern for harming victims on the other. But an individual can't riot on his own. Whether the riot happens depends on common factors that can influence many individuals at once, and information cascades within the group.

These riots happened in 2011. If we want an explanation of why they happened now, I think we need to look for common factors that are happening now. I don't think think ethnic divisions work as an explanation. They haven't shifted much in recent years except in terms of East European immigration, who were generally neither perpetrators nor victims in the riots.

The other major class of explanation is economic. Ponticelli and Voth find that budget cuts are correlated with riots. DiPasquale and Glaeser find that unemployment is correlated with riots, and GDP negatively correlated. Bohlken and Sergenti find that growth is negatively correlated with riots. Right now, we have budget cuts. We have very slow growth following an outright recession of negative growth. We have high unemployment. These factors, which we know to be correlated with riots, are present now. So it seems to me that these are likely to be the factors that explain why the riots happened now.

Yes, there is a moral element. But I don't think it's likely that morality suddenly dropped. More likely people had much the same morals in 2011 as 2008; but back then enough people had careers to lose and earnt disposable income that there wasn't a critical mass for a riot.

Responses
So given all this, what's likely to happen, and what can we do about it?

I think we have to be prepared for more riots in future. While some rioters have been caught, many more are likely to have got away. Everyone knows now that on a night with a normal police presence, it's possible to have a riot. The underlying causes of the riot are likely to remain. This time, in terms of human casualties and inter-ethnic conflicts, we got off fairly lightly as large-scale riots go. There is the potential for future riots to get very much worse.

In the short term, if we look at the initial trigger for the Tottenham riot, I think we need to make sure the Metropolitan police press office thinks a lot more long term. Rather than try to put an immediate spin on the story with inaccurate briefings, they need to stick to either the truth, or lies that can be sustained over the long term.

In terms of policing, we need to try to arrange things so we can deploy greater numbers of police at short notice after dark. It's possible that through changing shift patterns, having more on-call police, it could be done without increasing overall police numbers. Decreasing police numbers should not be done.

DiPasquale and Glaeser's study suggests that greater punishment is a deterrent. Tough sentences are probably a good idea. However, from the resentment of internment and other measures in Northern Ireland, we have to be aware of the risk of making things worse by letting a community perceive itself to be a victim. So basic principles of justice should not be violated.

Over the longer term, this shows it's more important than ever to provide jobs, growth and a good standard of living. Assuming that we want to continue current plans to cut the deficit, it's worth looking at different ways to do that. For instance, a land value tax might allow us to put a greater burden on taxation of the rich. This would mean we could cut public services to a lesser extent, and possibly cut VAT which disproportionately hurts the poor.

It's sometimes said that regardless of fairness or justice on pragmatic grounds we can't afford to tax the rich or they will harm our economy by leaving the country. If we want to be purely pragmatic rather than fair and just, then on the basis of these riots we should perhaps pragmatically conclude than neither can we keep cutting services for the poor, or they'll harm our economy by rioting.

In terms of the view that riots represent a decline in morality, I don't think there's evidence to support it, or a mystery that needs to be explained by moral factors. From polls moral explanations seem to be more widely accepted than economic by the public; and politicians and pundits generally seem to favour them. If true, it's not clear what can actually be done to improve public morality. I suspect that's part of the appeal of these explanations: if nothing can be done, you don't have to do anything.

I think that's a dangerous approach to take. Whether the causative factors are moral or economic, they are still there, and the possibility of more riots remains. While five people died in these riots, in the 1992 LA riots 53 died. The 2005 Paris riots lasted for longer and left an undesirable "new normality" behind. Riots have the potential to be far worse than what we experienced. We need to be serious about preventing and handling future riots. We need to look at the evidence and find the best strategies we can in all the areas: police tactics, prosecutions, and economic and social factors too. Leaping immediately to the most comfortable conclusions won't help.