Is the Theory of Constraints (TOC) a Theory?

The tag line Theory of Constraints Exposed in IndustryWeek article from March of this year got me thinking about TOC. Not a bad article by the way, although I’m still waiting for the Lean-TOC software system sales pitch shoe to drop.
It got me thinking, is Theory of Constraints really a theory? By theory I mean scientific theory, since any efforts toward improvement should be scientific, in my opinion. The glossary at What is Life tells us:
A scientific theory is an established and experimentally verified fact or collection of facts about the world. Unlike the everyday use of the word theory, it is not an unproved idea, or just some theoretical speculation. The latter meaning of a ‘theory’ in science is called a hypothesis.
So is TOC a scientific theory, a theory in the latter sense in that it is a hypothesis that can be proven or disproved, or neither?
If it is neither, we are at risk of practicing something akin to “belief” as a management and improvement system by using TOC, which is a thing fraught with risk. If so, perhaps it should be called DOC – Dogma of Constraints.
In the philosophy of science there is something called “falsifiability” which is extremely important. Falsifiability is logical quality of empirical statements such that they allow for logical counterexamples. Stated simply, a scientific theory must be falsifiable – there is a way that it can be proven wrong. In contrast, formal and mathematical statements can be universally true by definitions, axioms or proofs. These are more like tautologies.
Although management and improvement systems such as the Theory of Constraints or the Toyota Production System can and do contain mathematical statements or formulas which fall into this latter category, as theories or operating models, they must be scientific in order to be testable and therefore practically applicable.
The philosopher Karl Popper taught that no empirical hypothesis, proposition, or theory can be considered scientific if it not falsifiable. It must admit the possibility of a contrary case. We don’t need to observe or demonstrate that a theory is false, merely that it is logically possible for the theory to be false.
A good way to test for unfalsifiability (or untestability) is to show that a particular theory does not make a prediction, or that the predictions it makes cannot ever be wrong, even if the theory is false. It is then unfalsifiable, and therefore not scientific.
An example that is often given is that a divine being created everything that exists. While this may be true, it is not scientific by the above definition, since there can be no possibility of evidence to the contrary. Under divine creation as a scientific theory, if we did find evidence that would contradict it, according to the theory that evidence was also divinely created. There is no logical possibility in this theory that it could be false.
Is Lean a scientific theory? As a management system, Lean is scientific in terms of the thinking process. Lean has been experimentally verified in many cases over decades. The theoretical premise, or basic hypothesis of Lean, can be stated as:
1. The customer defines value by what they are willing to pay for
2. Whatever is not value is waste
3. Eliminating waste reduces cost
Waste is specifically defined in the seven types (overproduction, transportation, motion, inventory, defects, waiting, processing) as well as safety losses, wasted space, energy losses and environmental harm. If reducing inventory, defects or motion did not in fact reduce cost, then this theory could be proven false.
It is worth noting that Taiichi Ohno, the person who is credited with developing and advancing much of what is known today as Lean management, often spoke out against TPS being a theory of any kind. The words “practice, not theory” or “practical, not theoretical” were his rejoinder to managers and professors alike who poked and prodded at the workings of TPS. So perhaps “is Lean a scientific theory?” is the wrong question.
Lean appears to pass the test of falsifiability. But what about the Theory of Constraints? The steps to managing through the Theory of Constraints are:
0. Identify the goal (that which is being constrained)
1. Identify the constraint
2. Decide how to exploit the constraint
3. Subordinate all other processes to the constraint
4. Elevate the constraint
5. If the constraint has moved, return to Step 1
My concern here is that it appears to say in step 5 is “if the Theory proves false, repeat the test until it is proven true”. Does TOC admit the possibility that identifying and exploiting the constraint and then subordinating all other to the constraint, and elevating the constraint will fail to achieve the goal?
Is the Theory of Constraints falsifiable? Is it useful for making predictions? Is the Theory of Constraints a theory, or in fact a belief? You TOC experts will need to speak up to answer this question.
Even if not a theory, TOC remains useful, but we need to be careful what we call things.

10 Comments

  1. Pete Abilla

    August 19, 2007 - 8:41 am

    In response to the scientist, William James argues that judging nature when we are, in fact, a part of nature, is a subjective activity: in other words, scientist are not wholly objective — it’s impossible.
    James, instead, argues for pragmatism — rather than looking at “first things”, as he calls them, he promoted looking at “last things” to see if a theory works. He also call this the “fruits test”, taking it from the scriptural phrase “by their fruits ye shall know them.” In other words, if a theory makes sense, then it’s results will be pragmatic.
    I am a pragmatists, despite my deeply-rooted background in science. Your last sentence captures exactly the position of Pragmatists: despite it’s unclean-categorization or not-totally appropriate title, the Theory of Constraints remains useful — it works.

  2. Ron Pereira

    August 19, 2007 - 1:35 pm

    Excellent post Jon.
    I am by no means a TOC expert but do want to comment on step 5.
    From my understanding, Goldratt is saying that if (and almost inevitably) the constraint moves we must go back to step one. This is to say that attacking these constraints never ends much like attacking muda never ends for TPS proponents.
    So in this light I don’t think it is correct to say the theory is ever proven false when we need to go back to step 1.
    As an example, the constraint may be the assembly cell today but once we exploit it and it is no longer an issue the constraint could become the market in which case we need to exploit it.
    I am very interested to hear from the TOC gurus on this topic.

  3. Jon

    August 19, 2007 - 11:32 pm

    Hi Pete,
    I suppose it’s a question of precision then. There are many things that you could argue are useful, if you argue backward from the end result. Yet they are not predictive, or repeatable to any degree of precision.
    In addition to “last things” there should also be the “next things” that a theory allows for or predicts.
    As a pragmatist, it’s hard to argue with results. As a Lean thinker, I have to argue for process and predictability.

  4. mike

    August 20, 2007 - 7:40 am

    The predictability comes in the exploiting part and the math not in the end continuous improvement/kaizen like part. You cant predict that ahead of time.

  5. Pete Abilla

    August 20, 2007 - 10:14 am

    Not quite, Jon.
    Pragmatism teaches that, if a theory works by looking at last things, not axiomatically, then the components of repeatability, reproducibility, and predictability will come to bear still.
    Those elements are important, even from the perspective of Lean Thinking: a lean thinker is always evaluating — whether or not something worked, then going backwards to see where it could have been improved.
    During my short time at Toyota, I learned this lesson well: we were always encouraged to “try and see” — which meant that we should try new ways of doing things to see if they “worked.” At bottom, Lean is quite pragmatic — not axiomatic.

  6. John Hunter

    August 20, 2007 - 8:08 pm

    There are strong ties between Deming’s ideas and the pragmatic philosophy, one paper: Deming and Pragmatism.
    I like George Box’s quote “All Models Are Wrong But Some Are Useful” This can also be dangerous when people don’t understand the limits of usefulness.
    I do believe I understand your concern, Jon. The pragmatists were concerned with the theory of knowledge – how we know what we know. They were very concerned with evaluating thought and beliefs. And believed in testing to determine wether theories were correct. This thinking underpins the Shewhart/Deming/PDSA cycle.
    I believe this is very similar to the struggle Shewhart went through in developing the control chart and Shewhart cycle. He wanted to address the exact issue you raise of finding things that not only appear to be useful (which includes many instances of things that appear to be useful but in fact are not – we people are prone to this in many ways) but are predictably useful.

  7. Henrik Mårtensson

    August 21, 2007 - 3:17 am

    The TOC Chain Theory says that a system will always have a constraint. If a system has not, then it will be nothing that limits its Throughput.
    For example, an unconstrained company would grow until it owns everything on Earth. Then, of course, the company would be constrained by existing on a single planet…
    The TOC Focusing Process is a method for finding and eliminating the current constraint. TOC predicts that when that has been done, something else will be the constraint. That is the reason for step 5. The process must be neverending, or inertia will become the constraint. TOC predicts that over time, a company that does not change, will become more and more out of sync with its environment, which will first impede it, and eventually kill it off.
    I have published a webcast on the Chain Theory of TOC. You might want to watch it.

  8. Mindosan

    August 21, 2007 - 10:14 am

    > My concern here is that it appears to say
    > in step 5 is “if the Theory proves false,
    > repeat the test until it is proven true”.
    On the contrary, step 5 means “improve continuously”.
    A chain of links is only as strong as the weakest link. That’s your constraint. Once you have fixed that problem, it is no longer the weakest link. However the link that was the second weakest is now the weakest. The constraint has moved.

  9. Jon

    August 21, 2007 - 10:25 am

    Thanks Henrik.
    I really like your comment:
    “TOC predicts that over time, a company that does not change, will become more and more out of sync with its environment, which will first impede it, and eventually kill it off.”
    This positions the thinking of TOC in a very similar way to the TPS philosophy of harmony with society, rather than rampant growth regardless of social and environmental cost or social cost. Systems thinking requires that we recognize these sorts of impacts and consider them constraints.
    Based on your explanation, TOC says that there will always be a constraint. If this theory does not allow for a situation in which a system is not constrained, I will still say that it is not a theory. It is more like entropy, a physical law. TOC is actually the Law of Constraints.

  10. Henrik Mårtensson

    August 21, 2007 - 5:28 pm

    I’d still go with “theory”. While I believe that the Theory Of Constraints is essentially correct, and very dependable, the theory does have some fuzzy edges, much like the theory of relativity, and Newtonian mechanics.
    For example, production flows are often classified as being of type A, V, I, and T. While the TOC model covers all four flow models, dealing with T-flows is more difficult than dealing with the other types of flows. It is not inconceivable that someone will construct a better way of modeling T-flows some day. (Car manufacturing plants usually have T-flows, so Lean also deals with those. However, I would judge Lean and TOC being about equally effective.)
    It is conceivable that there is room for improvement in another area. CCPM, the TOC based project management methodology assumes errors in time estimates follow a normal distribution curve. There is strong evidence (Slack by Tom DeMarco; The Black Swan: The Impact of the Highly Improbable by Nassim Nicholas Taleb), that this is not true.
    Both DeMarco and Taleb argue that the distribution curve has a long tail. This would make the method of calculating buffer sizes in CCPM somewhat inaccurate. On the other hand, I do not know any method that is better. (I know one or two that I consider equally good, though.)
    BTW, thank you for writing the article. TOC is a very useful tool, and it deserves more attention than it gets. I believe the TOC and Lean perspectives complement each other very well. Learning about one of them helps understanding the other.