I don't think that this paper is the nail in the coffin that NNT or his groupies present it as. There are a lot of arbitrary transformations in the data that,

I don't think that this paper is the nail in the coffin that NNT or his groupies present it as. There are a lot of arbitrary transformations in the data that, while acceptable and in many cases enlightening, are by no means the only way of analyzing it. For example, having a bounded function between the population min and max and then a dual function that goes to infinity may help, but it's not immediatly obvious why. I find the rescaling of the data to be a fraction of the population (as Pinker did) to be more natural than their functional normalization, in spite of their objections. I also think that analyzing data in a cumulative fashion, rather than simply as events that occurred in a point in time (with uncertainty) is far more appropriate in the ease of simultaneously handling both short-duration events (WWII, US Civil War) and protracted violent conflicts (i.e. Mongol invasion, Hundred Years War). Then, breaks in slope in the cumulative w/ time function can be much more easily discerned. (Also do this with your budgeting/accounting instead of your monthly expenditures or whatever to see if your spending habits are changing). But again, looking at the cumulative function is great for getting a running tally through time but has problems with normalization. The point here is not necessarily that they did anything wrong, but that it's an ambiguous situation with lots of what Andrew Gelman refers to as 'researcher degrees of freedom'[1], and it's necessary to read all of it critically, and realize what researchers are doing and why. edit: [1]: http://andrewgelman.com/2012/11/01/researcher-degrees-of-freedom/ http://andrewgelman.com/2012/11/01/researcher-degrees-of-fre...