Power Laws - People Don't Understand Them

Image stolen from linked article which stole it from here: Althoff, Tim, et al. “Large-scale physical activity data reveal worldwide activity inequality.” Nature 547.7663 (2017): 336. (Of course, to me that should be a linear fit and they’re using a more complicated model than they should be - doesn’t anyone use regularization anymore?)

Image stolen from linked article which stole it from here: Althoff, Tim, et al. “Large-scale physical activity data reveal worldwide activity inequality.” Nature 547.7663 (2017): 336. (Of course, to me that should be a linear fit and they’re using a more complicated model than they should be - doesn’t anyone use regularization anymore?)

I came across this article written by someone I used to workout with. After he moved away, I lost track of him, but it looks like he’s doing research in stuff that it directly related (I hope) to AGT. We’re all about saving the mitochondria.

He reports on a study that does some big data analytics (i.e statistics on done on a Mac) from the results of some wearable sensors - think fit bit or other privacy invading apps on your phone.

Some of the interesting findings were that in societies where there was lots of “activity inequality” there was more obesity. Now, before we problemitize that and say that it’s a systemic issue, let’s play with the math a little.

Just about everything in nature follows a power law of some kind. You’ve heard of the Pareto distribution or the “80/20 rule” where 80% of the value is created by 20% of the population, for example.

Another good approximation of the Pareto distribution is that the square root of a population is responsible for half of the output (or costs…) - If you have 100 authors, 10 of them will have written half of the books, and of those 10, 3 will have written half. People aren’t equally productive…

Same goes for activity - if you use your Mac to do a histogram of the data you gathered from fit bit, you’ll see that most of the activity is done by a small fraction of the people. That’s just another example of power laws in action (or is that inaction?)

They use a standard measure of inequality (I don’t like the term inequality here - it’s just a measure of non-linearity) and they discover that places with a large inequality have a large incidence of obesity. Though, my take on this is that since things scale on a log scale, just a small amount of random variation in the tails of inactivity can turn into a big inequality measure.

To me, this is much more descriptive than prescriptive. In places where folks don’t move as much there is more obesity. But where is the causal arrow? If we believe Taubes (and I do!), then it’s the obesity that’s driving the lack of activity - and that where there’s more obesity, there’s going to be more inactivity. And threshold effect are great at fattening tails and making power laws!

So regardless of the fancy data collection and fancy data science and the need to tie this to inequality (I don’t think you can even get a paper published anymore unless you talk about systemic inequality) - I still think this all comes down to poor nutrition. All you have to do is ignore the USDA, AHA, and ADA, and all will be well. I like that people are looking into how using technology we can start to quantify the problems in society, but sometimes it’s about subtraction and going back the old way of eating before we industrialized our food production.

There are lots of reasonable options to eat better than what the institutions tell us. Stay away from processed foods, vegetable oils, and eat lots of healthy natural foods in the form they’re found in nature. Randy, Amy, and I are all thinking of heading to the first ever carnivory conference which will be held here in beautiful Boulder Colorado - anybody else want to join us? It will be catered by Blackbelly!

Michael Deskevich