A short rant about keeping your eye out for “noise” in your observations

This thought came to me a few days ago while listening to some pretty flimsy data analysis from a marketer and a journalist…

All modern philosophy is a philosophy of doubt. You can’t explain things away forever. At the same time, when, historically, a LOT of your data can be explained away as crap or “noise” (as in “signal versus noise”), then you get an eye for that sort of thing. If you run into someone who has absolutely NO eye on the noise, how can you not help but seriously doubt their conclusions?

When tracking down a difficult IT problem, parsing millions of lines of log files can often yield nothing but a stack of false leads. There are innumerable variables and we are just poking around trying to isolate something useful. A clever person with a lot of experience can hopefully come up with something solid to work from.

On the other hand, should major policy decisions, involving hundreds of thousands of dollars and good people’s jobs, should these things be decided upon by milling around the food-court and collecting some anecdotal evidence from passerbys? Should one angry email from a customer cause you to ignore hours of careful data collection to the contrary?

We use the latter method all the time though. We must, given the constraints of life. However, I propose that we should ALWAYS keep a humble and healthy eye on the noise and the potential for noise. There are so many ways in which our observations can turn out to be crap. Watch out.

What philosophers say about actuality is often just as disappointing as it is when one reads on a sign in a secondhand shop: Pressing Done Here. If a person were to bring his clothes to be pressed, he would be duped, for the sign is merely for sale.

-Soren Kierkegaard, Either/Or, EK p.42

Photo credit