Post by Bill Bellows, Deputy Director, The Deming Institute, featuring highlights of ongoing conversations with Ed Baker, author of The Symphony of Profound Knowledge (created in partnership with Aileron.org), and a Trustee Emeritus of The Deming Institute.
In this first episode, Ed and I share our interpretations of Continuous Improvement and Continual Improvement and why we believe Dr. Deming preferred the latter term. Also, we offer a reminder of the limitations of a focus on defects, rather than both good and bad process outputs, when striving for continual improvement.
Episode #1
Bill: While I know Dr. Deming favored continual improvement over continuous improvement, I would appreciate your explanation of the difference Dr. Deming was trying to convey.
Ed: Continual improvement implies that change is a step function. There is a measurable or observable change of state and the transformed state exists until the next change of state, like water transformed from a solid frozen state to a liquid to a gas. Continuous improvement (i.e. change) has no observable or measurable plateaus.
Bill: I like the idea of comparing step changes to plateaus. My explanation of continual improvement is the ability to find a leverage point in the system, where I could invest resources somewhere in the system to achieve a far greater gain somewhere else. The effort is not constrained with a focus on what is not meeting requirements, but rather where best to invest resources, including time, money, thought, etc. Once completed, I stop and then look for the next such situation. I have likened this to asking where does a stitch in time save 7, perhaps, wherever one sees a greater return than the investment. Next up, perhaps asking where does a stitch in time save 5. And so on. Each change begins and ends. I would then offer continuous improvement as an effort to improve a given process, ad Infinitum, which could easily lead to improvement well below the point of the return being great than the investment. This is what I think of as improvement for improvements sake.
Are these explanations of continual improvement and continuous improvement in keeping with your interpretation of Dr. Deming’s explanations of both?
Ed: Yours is also a helpful perspective about the concepts. You have offered a practical definition of the words. I think that Dr. Deming would have liked it because it kind of resembles his views on the need to consider practical significance when evaluating the importance of any statistical test results. However, I do like the idea of plateaus since improvement is a learning process.
Bill: Given this appreciation of continual vs. continuous improvement, how would you explain the weaknesses of an improvement strategy which focuses on eliminating defects? That is, one which is problem focused. What comes to mind is a classic adage from Russ Ackoff that “getting rid of what we don’t want might not get us what we want.”
Ed: Dr. Deming provided many examples of how failure to think from a whole system view leads to incomplete information and wrong conclusions. He described a situation he saw in a plant that manufactures tires. The engineers studied only the defective tires to determine the causes of defects. They also should have studied the non-defective tires in order to understand the functioning of the system as a whole. Without profound knowledge, in this case theory of variation and appreciation for a system, how could they know the source of the defective tires?
Bill: We are often asked for examples of how to apply Dr. Deming’s philosophy more broadly. How might we look at the broader implications of not focusing on defects?
Ed: This type of thinking, to look beyond defects, has broad application that we can apply every day in our lives. You might have this experience in a restaurant. You ask the server what the chef’s special is. The server tells you, and you say that doesn’t appeal to you. Then you list other things that you don’t like, including seasoning and methods of cooking. Of course, the server still doesn’t know what to do because you haven’t explained what you do like. If you give the server an idea of what you do like as well as what you don’t like, that is, you provide a sample from your whole-system of preferences, it will increase the chance that you will get what you like and will be a satisfied customer. This highlights a weakness of defining quality only as the absence of defects and the limitations of zero-defects programs that don’t define what does satisfy the customer. Eliminating defects can’t help customers if the product or service does not meet their requirements. (See Symphony of Profound Knowledge, pp. 18-19)
A recent paper by Ron Snee and Roger Hoerl, “Show Me the Pedigree,” (Quality Progress, January 2019, pp. 16-23) highlights the critical importance of knowing the origin and history of data, including the sampling method, before any meaningful analysis can be conducted and valid conclusions can be drawn. The failure to do so can have tragic results. They cite the disaster of the Challenger space shuttle due in large part to faulty analysis of the relationship between temperature and O-ring failures. The analysis did not include data for which there were no O-ring failures, which, if included would have informed NASA that launching at abnormally low temperatures would be very dangerous.
In other words, data from the whole system were not evaluated.