Re: No, the world is not becoming a more violent place so what are you getting out of
Found Pinkers book at the local library ... Keeley, too.
Funny thing is Keeley actually was showing that the early "primitives" were just as violent as "moderns". If there is such a drastic drop in violence then how come the early violent actors are shown to be just as violent as the moderns? Presuming one buys into the per capita metric in the first place (which I do not).
Pinkers presents data which shows 77% of violent death in recorded history happened in the 20th century.This does not count ~140 million deaths indirectly causedd by war or 'leviathan'. That is a pretty big mess to sweep under the rug. The only logic which makes this mess go away is to calculate death rates in per capita terms. There are many reasons why a simple per capita calculation is completely inapproprate but I will only present the following example. Consider a desert island with 2 people in which one murders the other. Now consider a neighboring island with 20 people in which one is murdered. Total body bount in both cases is the same but the per capita rate however is 10x different and by this logic the first island is 10 times more violent. Well, if that is the way one wants to think then I pity them greatly. Consider for a moment the possible relationships amongst the parties involved - maybe the 20 where 19 pirates and 1 stranded innocent ... context makes the per capita logic crumble. In either case the sheer numbers of unmurdered does not 'dilute' the issue of the violent acts.
In addition there are several points in which Pinker boldly claims some data set clearly shows on obvious downward trend (provided one ignores the large spikes) - Even without the obvious issue of ignoring the large spikes the downward trends is simply not clear and no explicit regression is presented to support the claims. Note that these 'spikes' were WWI & WWII ... not exactly the kind of events one should cavalierly ignore - even if Pinker proclaims them statistical flukes. In any case there is no statistical rule that permits one to exclude data so one is honor bound to include the spikes and 'flukes' or present cogent argument on why they are not relevant to the analysis at hand (this is hardly the case).
In general the book was a good read (even morso for Keeley) for both entertainment value and it does present some interest for general pondering. Of course one must read such works with their skeptic hat firmly in place. Although long and meandering in many places still worth the effort.
On a personal note - Pinker also uses the "half below average" quip so if the OP had been so critical of Pinker as to me for a simple one-liner then maybe a different weight would be put on the conclusions of Pinker. Note that Pinker befuddles himself by mixing median and mode in a nonsensical way so perhaps I can lend him my "For Dummies" books to brush up a bit.
Last edited by Rob Watson : 12-29-2012 at 11:44 PM.