In a recent post I alluded to my fondness for the normal distribution, sometimes known as the bell curve. I use it as a filter through which to view big – especially apocalyptic – claims. Now comes John Robb with a post contrasting the bell curve with the long tail.
Historically, Gaussian [bell curve] expectations for most events derived from human systems were usually correct. In that world, dampening factors dominated within relatively sparse and simple systems, driving events towards the mean. Over the last decades, however, systems have shifted towards towards ever greater levels of complexity and information density. The result has been a shift towards Paretian [long tail] outcomes, particularly within any event that contains a high percentage of informational content.
Interesting stuff – when combined with Charlie Stross’ observations on changes in transportation speed, we’ve got three models to worry about:
- normal distribution – things are going along as they usually do
- power law curve – OMG, it’s the singularity!!1!
- sigmoid curve – things will change quickly, until some higher level constraint is reached
The devil will be in deciding which model to apply to a given trend – regardless, if you’re passing out tracts and wearing a sandwich board proclaiming the eschaton, I’m going to avert my eyes and scurry by.