Wednesday, May 27, 2015

"New Study: The world's a lot more violent than reported"

On his Twitter page Harvard's Niall Ferguson calls it "hugely important."
In the paper ["On the tail risk of violent conflict and its underestimation."] Nassim Nicholas Taleb, the author... argues that "Violence is much more severe than it seems from conventional analyses and the prevailing 'long peace' theory which claims that violence has declined."
Contrary to current discussions, all statistical pictures thus obtained show that 1) the risk of violent conflict has not been decreasing, but is rather underestimated by techniques relying on naive year-on-year changes in the mean, or using sample mean as an estimator of the true mean of an extremely fat-tailed phenomenon; 2) armed conflicts have memoryless inter-arrival times, thus incompatible with the idea of a time trend. Our analysis uses 1) raw data, as recorded and estimated by historians; 2) a naive transformation, used by certain historians and sociologists, which rescales past conflicts and casualties with respect to the actual population; 3) more importantly, a log transformation to account for the fact that the number of casualties in a conflict cannot be larger than the world population.
The authors base their article on the methods of extreme value theory.

A striking chart accompanying the article dramatically shows the impact of violence on all periods of recorded history. The chart measures conflicts featuring more than 50,000 deaths relative to today's world population. (Thus, 50,000 deaths today = 5,000 deaths in the eighteenth century.)

6 comments:

Mitch H. said...

Meh, both sides of the argument seem to be abusing statistical techniques to conjure better data than the extant reliable records have to offer.

bagoh20 said...

By this population adjustment factoring, our lives today are only worth a fraction of the value of our caveman ancestors, which is confirmed by the seriousness of TV and popular periodicals today.

edutcher said...

That last parenthetical line really is the money quote.

Proportional perspective does make a difference,

PS I do wish they'd labeled some of those bubbles so we know what conflicts they deem more significant.

That, and it looks like somebody really needs some Alka-Seltzer.

Bleach Drinkers Curing Coronavirus Together said...

Eh, Niall's an interesting lecturer on monetary history - but I'll believe it when he refutes all the points Pinker's massive tome directly. For one, the time scale on that chart is a tiny sliver of what Pinker looked at.

Bleach Drinkers Curing Coronavirus Together said...

The excerpt makes explicit reference to "armed conflict", rather than to violence in general. The problem with that is the number of deaths and casualties matter. If war has gotten safer, whether through better medical treatment, more targeted tactics, or better protection, I think that plays into what people think of as "violent". FFS, you could call paintball or throwing snowballs acts of violent aggression; but the fact that no one gets hurt makes a difference.

Mitch H. said...

Rhythm, the problem is that the actual misery and killing of wars, throughout history, and almost without exception, is done out of the view of the interested historian or chronicler, and honestly, can sometimes be difficult to distinguish from day-to-day insecurity and societal violence, esp. in the premodern era and in regions where the state is weak or non-existent. Raiding and night-time mayhem almost always racks up a greater bodycount than stand-up battle, I don't care if you're talking about Waterloo or a bunch of naked tribesmen waving spears and making faces at each other over an open field.

This makes it very difficult to assess the actual mortality of any given period of warfare, except to do graveyard forensics, if you happen to find yourself a culture that favors intact burial, and even then, the palimpsest nature of old graveyards makes *that* a charming horror of a task.