By Paul Rincon
Science reporter, BBC News
Armed conflicts show striking statistical similarities
Researchers have uncovered common patterns in the scale and timings of attacks across a variety of different violent conflicts.
A total of 54,679 violent events spanning several decades were analysed.
The team searched for statistical similarities across nine historic and ongoing insurgencies including those of Iraq, Afghanistan and Northern Ireland.
The results, published in Nature journal, may offer the hope of reducing casualties in future conflicts.
"We found strikingly regular and similar patterns in the sizes and timings of violent events," said Professor Michael Spagat, an economist at Royal Holloway, University of London.
Professor Spagat said that there were more large-scale attacks than one might expect and that attacks tended to come in bunches, or bursts.
The relationship between the "size" of attacks (measured in terms of casualties) and the cumulative frequency of attacks could be plotted as a straight line on a graph, the Royal Holloway researcher explained.
"If you look at a single conflict, you first see this regular pattern: they all line up nicely on a straight line on logarithmic axes. Secondly, as you range across conflicts, they're fitting almost the same pattern," Professor Spagat told BBC News.
This was different to the "bell curve" shape one might expect the graph to take, he said. With the bell curve, data is clustered around the mean and tails off on either side.
Professor Spagat said: "You've got this pattern of bursts, with large events being less rare than people might suppose. That suggests your emergency response planning has to have a lot of capacity, and that this capacity would be under-utilised almost all of the time.
"But then you shouldn't be surprised when you get bursts of large events and the trauma units are overloaded. You should put a lot of slack into your disaster planning."
Bursts of violence
The tendency of attacks to occur in bursts, with activity dipping in between, even bore some similarities to traffic patterns, the Royal Holloway economist explained.
Other researchers have shown that the global distribution of terrorist events fits much the same patterns as insurgencies.
"With events the size of 9/11, it's not just a freak occurrence that could never happen again. It's rarer, but very much within the realms of possibility that something on that scale could happen again," Professor Spagat explained.
The researchers say the fact that fundamental patterns emerged from the apparent chaos of armed conflicts afforded some hope that some practical knowledge could be developed to reduce casualties.
"We could never say with certainty that there were going to be 10 attacks on one day, for example," Professor Spagat told BBC News.
But, he explained: "Within the framework of the wars in Iraq or Colombia, you could break days down into busy days, average days and light days.
"If, over the last three days, you've had a pattern of busy, busy, light, you could look at three-day windows in the past that show the same pattern.
"Then you could make a prediction on whether the next day was likely to be busy, average or light."
His team put together a "unified model" of human insurgency that reproduced the commonalities in the data.
The study involved researchers from Royal Holloway; Cambridge University; the University of the Andes in Bogota, Colombia and the University of Miami, US.