The previous post, Time-in-Process and Entropy, was somewhat technical. Its concepts, however, deserve a less technical explanation.
Suppose you’re a manager trying to improve some work or business process you’re responsible for. Or you’re an executive leading this manager or a consultant trying to give this manager some advice. You may be making observations, recording some data about what’s going on, analyzing, looking for patterns, and deciding on the basis of such information.
When looking at information about what’s going on, there are some good questions to ask. Is the information good enough? Is it sufficient? Do we really know what’s going on? How do we know?
In his book The Black Swan, Nassim Nicolas Taleb described two fictional countries, Mediocristan and Extremistan. Mediocristan is populated by phenomena such that our observations of them can never stray too far from the average. Extremistan is populated by phenomena allowing the probability, if small, of extreme events — black swans. If the world’s largest or smallest person entered a stadium where 50,000 fans were already seated, how would that change the average height or weight of these people? Hardly at all. Now what if Bill Gates entered this stadium? As the joke goes, the average fan in the stadium is now a millionaire. This is because a person’s size lives in Mediocristan, while wealth lives in Extremistan. (By the way, telling this joke as “Bill Gates enters a bar…” doesn’t properly communicate the extremes.)
Sometimes people describe Mediocristan as Gaussian and Extremistan as Pareto, referring to two somewhat popularized technical terms.
In the heart of both Mediocristan and Extremistan, it takes a lot of observations to really know anything. In the Gaussian world, you’ll see something outside the mere plus-minus three-sigma control limits, on average, once in approximately 370 observations. In Extremistan, you may be going through thousands of white-swan sightings before seeing the first black swan.
Mediocristan and Extremistan do share a border. I’ll spare the technical terms for how this border is demarcated. What’s important is, unusual things happen in the border region much more often. The value of each observation is much higher. And we don’t need too many observations or too many data to really know what we’re observing.
Time-in-process and time to market live in this border region. Good news if you’re a manager with responsibility for delivering some products or services to your customers! You can create a very good model of what’s going on in your business with very few data points. You need less data thank you think!