Monthly Archives: March 2013

Sampling Theory – Measuring the difference between what we know and what we think we know.

My story starts with sampling theory.  It winds through W. Edwards Deming and working in Quality Assurance at Boeing.  And on a personal level, it takes me through psi and skepticism and higher mathematics and god and mental illness.  Not necessarily in that order.

But it starts with sampling theory.  Sampling theory is a technical area in the intersection of mathematics and quality assurance.  It is concerned with measuring the difference between what we know and what we think we know; more specifically, it is concerned with using samples to make strong conclusions about the true nature of a larger population.  I was studying this in the 80’s to apply in the ‘real world’ a.k.a. the Boeing Wichita Manufacturing Facility.

I started work at Boeing in January of 1984, age 25, a freshly minted college graduate with a bachelor’s degree in mathematics in the QA R & D organization.  Those initials stood for ‘Quality Assurance Research and Development’.  Suddenly I was an engineer!  That was my job title.

The most important thing I learned while working in Quality was the difference between what was actually happening and what people thought was happening. They inability to make this distinction lead to many heated disagreements between engineers and shop workers, including their first level management, at Boeing about what was and wasn’t a reasonable description of the ‘as is’ and whether or not the ‘to be’ was remotely possible.

The ‘real world” is the perennial favorite term to refer to someone’s own tribal perspective as opposed to that of their rival. It’s always appropriate no matter what the technical jargon and generational differences. The ‘real world’ as opposed to the academics in their ivory towers,  the rich in their bubbles, the religious in their delusions, youth in their naivete, or engineers in their offices with desks and chairs and the opportunity to sit while working.

At one point relatively early on, I had an assignment to help a software engineer develop a computer program to provide Mil-Std-105D sampling plans.  I advised him to program it as look up tables.  He wanted a formula because memory was expensive. Since there were a few hundred different tables in the document, it would require a lot of resources to input and store the tables.  A program to compute the value would take less time and save hardware costs.

Sampling theory back in the 80’s was not significantly changed from the military standards created during WWII to get maximum production from US manufacturing plants.  When I dug into the equations that were used to compute the values in those tables, they turned out to be a kludge of techniques.

Let me note here that this compendium of sampling tables produced for the military was actually a major achievement for that time, before the advent of cheaply available computing power. Apparently they had used a variety of techniques and basically done a hell of a lot of hand computations independently and then compared answers.  There was no single formula or even a consistent set of formulas.

After duly researching the matter, I had to report back that I couldn’t do it. I could find no documentation of what formulas the original authors had used.*

A few years after that assignment, I was in a meeting with some Boeing folks from Seattle.  I told them that we would be moving away from Mil-Std-105D towards better methods of acceptance sampling.  They laughed in my face.  It was humiliating.  Not least of all because they were right.  A decade later, Boeing hadn’t changed at all in that regard.

I feel my years spent working there were for naught.  Despite my best efforts, which included trying some seriously crazy off-the-wall stuff I called Guerrilla Quality, I was unable to make any sort of dent in the Boeing corporate culture.  It was hard core theory X management when I started and it never moved far from that position.

I have always wanted to contribute something lasting to my culture.  At this point, I feel my best choice for accomplishing that is through my idea for computing generic basis values and the accompanying specification limits for composite materials. I think that this idea will be of significant value to my society because it will allow a much wider array of individuals to make use of these materials. This availability will have a significant impact on our ability to go into space cheaply.  It could help open up space travel to many more people.  That would make me very proud.

Except I can’t seem to actually write a paper explaining my ideas.  Or rather, I can’t seem to revise my rejected paper.  I’m calling it writer’s block. I don’t know if that’s accurate.   I can’t bear to open it and look at it, much less work on it.

I think the reason I’m not working on it is because I have issues I need to face surrounding it.  This includes revealing embarrassing stuff I don’t really want to talk about, but feel that I need to because they form the blockage in my head that’s  preventing my generic basis values paper from getting out.

That’s why this narrative will meander around many different issues that don’t seem to have anything at all to do with quality assurance or statistics. Mental illness, god, psi, and other things I want to talk about relate to the issue of what we know and what we think we know.  That’s how it connects up with sampling theory.  At least, that’s how it connects up in my irregularly functioning but sometimes brilliant brain.

* I now realize that what I needed to do was figure out what formulas would have been appropriate and set up my own, but such an endeavor was beyond my professional capabilities at the time.  Of course, it would now be much cheaper to buy memory and put in look-up tables than pay professionals like myself to sort out the right formulas and make sure the code produces the right computations.