Occam’s Razor, Value Added, and Waste

The manufacturing people I know are not fond of theories. They prefer to be on a shop floor with oil mist in the air and machines banging away, observing and acting on the actual situation, than in a conference room, drawing boxes and arrows on a white board, calling it a system and arguing about it. Theories, however, are not a luxury. We need them to structure, organize and focus our thinking about the reality we perceive. Pascal Dennis calls them mental models, and, as Georges Matheron used to say, nothing is as practical as a good theory. A good theory is sophisticated enough to solve real problems, yet simple enough for people  to understand and apply. This is a tall order. To fulfill it, you must not only be inclined to abstract thinking, but you must also have the time to do it, and the ability to communicate.

And there are theories on how to make theories, such as Occam’s Razor, which says that,  among competing theories, you should use the simplest until you have evidence to prove it false. Scientists, engineers, medical doctors, and cops do this all the time. If you make a theory more complicated than it needs to be for its purpose, you make it more difficult to confirm or refute, apply, and teach. About Lean, Occam’s Razor says that we should not have more hypotheses or concepts than strictly needed, as they would be unnecessary, or  muda. If we had a list of seven categories of waste for building theories, Occam’s Razor might provide the first: Unnecessary complexity.

When you read Jim Womack’s theory of Lean, as expressed in Lean Thinking and replicated in much of the American and European literature, you find that it is based on a concept of value-added (VA) that is more complex than the concept of waste (muda) used in the Japanese literature. Jim Womack’s VA is defined as what “the customer is willing to pay for.” I found no reference to this notion either in the works of pioneers like Taiichi Ohno and Shigeo Shingo, or academic observers like Yasuhiro Monden or Takahiro Fujimoto, whose works are available in English. Even a recent author like Mikiharu Aoki, whose 2009 book on the heart of introducing TPS is only available in Japanese, manages to explain the entire system without this notion. It is also absent from Jeffrey Liker’s Toyota Way. You do find occasional references to “value-adding work” in the Japanese literature, but always to designate activities that physically change materials, which is not the same as Womack’s VA. If they don’t need this concept, do we?  or is it a violation of Occam’s Razor?

The confusion about VA is reflected in the large amount of questions and requests for clarification that every mention of it generates in discussion groups. VA sounds deep, and everybody wants to make sure they understand what it is,  how to recognize it, and how to separate activities among “Value-added,”  “Non-value added but necessary” and “Unnecessary.” In particular, the distinction between “Value-added”  and “Non-value added but necessary” would be useful if there were differences in the actions taken about these two categories of activity. Let us take two examples:

  1. Fastening two parts with a screw is usually put in the “Value-added” column. About this kind of operation, first you make sure it is effective, meaning that you use the right screw on the right parts, thread it truly, and apply the right torque. Then you work on making it efficient, so that, for example, no operator has to walk six feet back and forth to pick one screw or fetch a screwdriver.
  2. Revision control on process specs usually goes into the “Non-value-added but necessary” category. A customer would not be thrilled to pay for it, but the plant could not operate without it. First, you want to make sure it is effective, so that the right instructions are posted at each workstation and the right programs downloaded to machine controllers. Then you want to make it efficient, so that engineering changes are either rejected or approved and implemented promptly, without using an army of people to do it.

The two F’s, Effectiveness and Efficiency, are what you pursue in both cases, in the same order. Several people have tried to explain to me how fundamentally different these cases were, but I still don’t see it. I really don’t care how different the activities may be from a philosophical standpoint. All I am interested in is what you do about them, and if the actions are the same, I see no point in having two categories. One argument I heard is that you do “non-value added but necessary” activities only because you have not yet found a way to eliminate them. But the same is true value-added activities. The fastening operation could be automated, the screw replaced by a rivet, or the parts could be welded together, or the entire assembly could become a single molded part… Just because the operation is “value-added” doesn’t mean you are not trying to eliminate it.

Using just two categories — what you need to do and what you don’t — is not only simpler but these two categories are also easier to tell apart. As we all know, customers’ willingness to pay is only proven by actual payments, and therefore cannot be clearly established for a fastening operation halfway through an assembly line. On the other hand, an operation is unnecessary if and only if its elimination would not degrade performance in any way, and we have Ohno’s list of waste categories to help us locate them. If we eliminate any kind of waiting, for example, we know that we will improve quality, reduce costs,  accelerate delivery, without jeopardizing safety or hurting morale. We don’t need to look anywhere outside the workstation to establish this.

Womack’s VA/NVA analysis bestows the “value-added” label on a small number of activities while branding everything else as “non-value-added.” By contrast, Ohno’s list of waste categories is finite. and everything that is not explicitly identified as waste is accepted as necessary. Whenever this kind of classification is made, it is only valid under current conditions.