I used to coordinate new product safety reviews. A few types of engineers, Customer Service directors, and Operations pros would review all the ways a user might be a physical danger with the product, safeguards, and certain failures in it’s intended and “reasonably foreseeable” application. In total, the product along with its instructions, care, and warnings make for an entire system. It’s a great feeling to collaborate and be confident that a safe product is going to market. That as a system, even when it fails, nobody gets hurt. Then, people surprise you; they use your system in some unforeseeably and unreasonable kind of way that you never thought someone would even dream of.
Such was the case when a man severed/damaged his genitals from doing woodwork in his lap with a router having a blade whizzing around at 100mph or so.
Such was the case in the fabled lawnmower being lifted to trim the hedge.
I get the feeling that this is exactly the same type of thing going on with the U.S. Government. The system, as designed, was never anticipated to be used in this way. And, unfortunately, the damaged body is a whole nation.
The catastrophe is not as much a failure of the system as it is a failure of anticipated reason.
Graph 1. Graph provided by the module. Numeric totals and percentage reflected.
Looking at stats and running with an Average can be a tricky assumption. We typically envision an Average as the midpoint of a uniform or symmetrical distribution. However, frequently, that’s not what you have. These graphs represent the same data regarding Pageviews per Visit on a new client we’re teaming up with. Their existing site is in a self-serve, Website Tonight type cms. Graph 1. is the graph provided by the stats package. You’ll notice the large green bar representing 1 pageview directly followed by what appears to be a normal distribution. However, closer inspection of the groupings show one digit in 2, 3, 4, but then groups of 3, 5, and 7 digits etc..
You’ll be interested in knowing the Average is 1.104
Graph 2 is another variation on the same data, but using the bar chart to convey another alternate visual representation that’s misleading. The main reason we’re looking at thesedistributions is to better understand how mobile users vary from desktop users.
Graph 3 below provides a more realistic representation of how the data is actually distributed. It’s great to recall that the Average is 1.104. Considering the shape of some “metric” should be a part of an evaluation when goaling to improving that metric. Imagine executing a change that didn’t move the metric’s average value. However, a peek at the distribution may reveal a change that was otherwise dismissed in an Average.
Graph 2 – Another variation on same data.
As much as we like to have solid numbers, particularly in any optimization strategy, sometimes the better story is yet under the hood.
Graph 3 – Closer to representing the actual distribution.