Suppose you wanted to prevent future urban riots or, at least, reduce their likelihood. You would want to reduce the frequency of the potential spark as well as reduce the proportion of people who are ready to respond to the spark (or join with others who have responded to the spark). Let us consider the second challenge first.
There will always be a rage of responsiveness to a spark, where responsiveness derives from some as yet undetermined combinati0n of individual proclivity to criminality and frustration or anger at social conditions. Suppose you assume that there is some threshold of responsiveness below which people do not riot. Suppose also for the moment that responsiveness is not 100% individual proclivity to 0% social conditions. That means you could imagine reducing the causes of frustration and anger. To do so would be equivalent to shifting the distribution of responsiveness from the top curve to the bottom. (The horizontal axis depicts responsiveness on a scale from 0 to 1; the vertical axis depicts the proportion of the population having this responsiveness on an arbitrary scale.)
In the top plot 1 person in 146 is above the threshold (the cross-hatched lines on the right hand tail); in the bottom, 1 person in 1526. If the spark is not experienced by everyone, but by say 100,000 people, then this is a reduction from 685 people to 66.
Of course, the numbers are made up, for purposes of illustration only, but the point illustrated is that shifting the population is an alternative to framing the issue in terms of controlling the individuals who are highly responsive to a spark. (Control may be policing, curtailing civil liberties, preventative detention, control over social media in times of tension or unrest, and so on.) Under that alternative, paying attention to the frustration/anger side of responsiveness corresponds to an interest in reducing future violence even if politicians and the media polemically equate understanding the causes of responsiveness with absolving individuals of responsibility that thus condoning violence. Indeed, an emphasis on control is likely to have the counter-productive result of increasing the frequency of sparks given that there are inevitably mistakes made by some of those who are given the power to control more heavily. The most obvious mistake is when an individual is treated according to the group that they are (or presumed to be a) member of, where the range of responsiveness for that group is disproportionately to the high responsive side.
What would it mean to pay attention to the frustration/anger side of responsiveness? (This question is a precursor to the question of what would it mean to shift the distribution of responsiveness.) Answering that requires serious social science, but it should be possible if there is variation among regions in how deeply budget cuts have affected services, racial profiling, job loss, and so on.
What would it mean to pay attention to the individual proclivity side of responsiveness? (This question is a precursor to the question of what would it mean to control the individuals who are highly responsive to a spark, especially if you want to do that without increasing the sparks.) Answering that requires serious social science, but it might be possible if there is variation in incidence of riots among regions with equally deep budget cuts, racial profiling, job loss, and so on. Once you have some answers, however, it is difficult to convert that knowledge into individually focussed control. In practice, individuals are treated according to the group that they are (or presumed to be a) member of, where the range of responsiveness for that group is disproportionately to the high responsive side.