The Funnel Experiment

Each of the rules of the funnel have been simulated and charted. The charts are shown below.  Rule 1. Exhibits the classic picture of a statistically stable process.  It is predictable within limits and, unless the system changes or is changed, is likely to keep on this way.      Rule 2.  Shows the beginnings of a zig zag pattern as the funnel moves back and forth.  It is not statistically stable as stability is defined as varying randomly between limits.  It can also be seen that the variation (size of shifts from point to point) is greater than under Rule 1.  If predictability is the aim, the enemy is uncontrolled variation.    Rule 3.  This rule is similar to Rule 2 and produces a similar result.  The primary difference being that variation has increased fairly dramatically and the back and forth zig zag is more pronounced.   Rule 4.  The plot starts like Rule 1, but after a few iterations, the process begins to drift.  In essence the process begins a random walk off in one general direction.  Sometimes it takes a while to begin it’s drift, but eventually off it goes, as Deming would say, “To the Milky Way”.     We can see from this simple experiment that before improvement efforts begin it is important to understand what constitutes improvement and what can make things worse.  Adjusting a process that’s in statistical control after every occurrence of the process is likely to make things worse.