A recent discussion on what is the appropriate TOC definition for ‘constraint’ leads me to state some historical facts that highlight the development of Goldatt’s approach to constraints.
Prior to the Theory of Constraints (TOC) the breakthrough idea was to distinguish between bottlenecks and non-bottlenecks. The definition of a bottleneck was simple: “The load placed on the resource is more than what the resource is able to do.” Thus, a bottleneck is always a resource.
The term ‘constraint’ was defined “Anything that limits the system versus its goal”. It was conceived to answer three significant limitations of the term ‘bottleneck’.
1. When all resources have enough capacity to process all the demand then there is no bottleneck. However the system is able to do more. Thus, looking on the market demand as a ‘constraint’ is quite valuable. It allowed managers to understand that there is no excuse not to ship everything on time.
2. Being a bottleneck does not ensure being a constraint. There might be another bottleneck with even more load.
3. We might have a true resource-capacity-constraint (CCR) that is not a bottleneck. While on average there are idle times, in other times the queue behind the CCR is so long that some potential demand is lost.
It was realized from the start that the constraint limits the throughput (T) of the organization. Goldratt even played with the idea of introducing the term of ‘inventory constraints’ referring to trouble-makers that force the organization to maintain more WIP. He backed off this term to keep the simplicity.
The real power of the term ‘constraint’ came through the paradigm that an organization cannot have many constraints. Dependencies coupled with statistical fluctuations do not allow interactive constraints in the chain. This realization led to the conclusion that the shop-floor can handle only one constraint without creating chaos. In 1989 Goldratt wrote The Haystack Syndrome and presented a rather complicated algorithm to handle multiple constraints. The whole development of the ideas was set around capacity constraints. The chain analogy, where there has to be one, and only one, weakest link in the chain was widely used. Thus, the default for a constraint was lack of enough capacity of a resource.
Limited capabilities, like being unable to produce top quality products, were not considered constraints. Limited capabilities are less exposed to statistical fluctuations.
The wide definition of the term constraint did cause problems. People used to say that the constraint lies between the eyes of the CEO. Flawed policies, especially policies concerning efficiency, were called ‘policy constraints’. So, the idea was that the system is limited by a capacity constraint, and failing to exploit it is due to policy constraints.
The full set of TP (thinking processes) was developed in 1990. Effect-cause-effect trees and the cloud existed before (even before the 5fs) but not the other tools we know today. The definition of the CRT raised the notion of the core problem – the conflict (cloud) that causes all the undesired-effects. Resolving the conflict by challenging a basic assumption behind the conflict would push the organization to a new level of performance.
So, is the core-problem the real constraint?
It remained an open question for a while. Core problems touched upon local versus holistic thinking, but also on behavioral patterns and opened the door for re-evaluating the value the organization brings to the market. The core-problem could also challenge the paradigm why do we exploit a CCR rather than immediately elevate it.
Fact is: we did not ask ourselves these questions in the 80s.
Goldratt publicly regretted calling flawed policies “policy constraints” sometime in the 90s, explaining that policies should be eliminated and not exploited and subordinated to.
A major development in the TOC thinking came around 2003 with the idea of the Viable Vision. Suddenly the way to improve an organization did not come through elevation of a capacity constraint and even not through challenging the conflict behind a policy-constraint. With the term “decisive-competitive-edge” the TOC thinking has realized the need to challenge the value the organization offers to its customers. The core idea was to answer a need of the customer in a way no other competitor can.
Explaining how come the VV did not care what is the constraint, Goldratt spoke about two different changes. One is minus-minus – you identify something that is not right (minus) and you change it (minus of the minus) and a plus-plus change where you take a big step towards the “pot-of-gold”. When such a step is taken one needs to carefully re-think all the conditions that would be sufficient to bring the organization to growth along the “red curve”. Lack of capacity of a specific resource becomes a triviality that needs to be eliminated. Many other potential constraints would be elevated long before they become constraints.
Food for thought?
7 thoughts on “A concise history of constraints”
This is a very useful blog entry. Thank you for taking the time to show the progression in the TOC body of knowledge. Your explanation is very clear.
I would say something a little different about VV, I don’t know if it is disagreement or not.
It seems to me that Eli Goldratt recognized a pattern early this century that the first steps of applying TOC to an internal constraint typically broke that constraint and any other internal constraints, quickly moving the company’s active constraint into the market. Yet many TOC experts, who were helping companies improve, persisted in perfecting operations using DBR, later sDBR, CCPM or the TOC Replenishment Solution (presumably they were restricting their focus to their areas of expertise or simply just wanted to properly finish what they started) rather than switching to address the companies’ new market constraints, despite the four exclamation points in Focusing Step #5. Goldratt called for a stop to these “functional implementations” in favor of “holistic implementations” like VVs.
Goldratt saw clearly that, unless the market constraint is exploited, the company would produce fewer goal units. So, the concepts of decisive-competitive-edges, minus-minus and plus selling were decisions of how to exploit the company’s market constraint (FS #2). The actual application of these are subordinations to the exploitation decisions, as often detailed in a Strategy and Tactic trees, (FS #3).
A company practices FS #4 by addressing an ever larger market, hence, Goldratt’s conclusion to Isn’t It Obvious with the words “Even the sky’s not the limit.” This insight lead to his last address delivered by Lisa Scheinkopf at the TOCICO in New York where he claimed a fourth pillar upholding TOC: supported by “Never Say I Know” and your corollary “Never Say I Don’t Know” – two sides to the idea that the BIGGER and STRONGER the BASE, the HIGHER the NEXT LEAP.
Henry, I agree to your description of Goldratt’s thoughts when he went into the VV direction. We don’t need to perfect Operations (SDBR, CCPM and Replenishment) as long as we don’t establish big enough market demand with the high appreciation and awareness to our unique value.
However, once we have a large market the ball bounces back to Operations with a critical question:
Can you sustain the growth?
So, it is possible that you might need to further perfect Operations. But, a very important lesson emerges:
Do not exhaust any or your resources!
The meaning is: do not allow any internal capacity constraint in your value-flow!
Is the VV an exploitation and subordination scheme of the market constraint? It looks to me closer to the wider meaning of “elevate the market constraint.” Exploit and subordinate are focused on the current situation: given demand and given capacity (and capability) profile. The first three steps guide to draw the most of the current. Elevation speaks about the future, but then the 5fs are too vague to practically guide the organization on “how to elevate the market?”
I plan to write more about the issue of elevating the market constraint.
I have a question. A capacity constraint I understand — a resource that has less capacity relative to demand than other resources. For a given system and scope, such as a production system, or a project system, I can see for any resource what their current demand is, what their current performance is, and what their maximum performance is. Even the last is observable and measurable (if we can experiment a little). So I can find the constraint(s) or near constraints, and when I change them, I can tell if their capacity is improved, or not.
But for a “market constraint”, a “policy constraint”, or a “management constraint” — how can I determine their “demand”? Their performance? Their capacity? If I change something, how do I tell if it “worked”?
In what sense are these really ‘constraints’?
Richard, I have delivered a webinar, which TOCICO members can watch for free, on the topic of “Re-evaluating the five focusing steps”, and I have tries to re-define what is a constraint. This caused a lot of discussion.
The old definition of a constraint was: “Anything that limits the system versus its goal”. It is based on the goal, not necessarily on demand and how we define demand, like the clear difference between current demand and potential demand.
This definition does not deal with the ability, or how easy it is, to identify a constraint and measure it. Machines and equipment are relatively easy to measure their capacity versus demand. The capacity of humans is much more difficult to measure. But, difficulty is measuring does not mean we cannot safely assume something is the constraint and does not mean we cannot improve the exploitation and subordination of it.
Richard, if you like I can send you the slides of my controversial webinar – just send me you email to: firstname.lastname@example.org
Thank you so much for pushing TOC forward. The term constraint really puzzles my mind in the sense that Goldratt (I believe) was very cautious with words and definitions and nevertheless he decided to leave constraint as a vague term. Am I right? Why was that?
As you say “The term ‘constraint’ was defined “Anything that limits the system versus its goal” and also “It was realized from the start that the constraint limits the throughput (T) of the organization.”
To a certain extent I think I understand that the constraint must be linked to T because that is the way to attain more goal units. On the other hand, this automatically excluses any kind of OE reduction initiative. Yes, I am aware and fully agree that the focus should be in increasing T (sooner or later) but what if OE is really killing the company (in the short term)? I think we cant prove that this is never the case.
As a last thought sometimes I get the (erroneous?) impression that the constraint could be the thing that, if explored, could contribute to the highest ROI (eg scrap is 1M€ and reducing cycle time on the constraint would lead to 500k€ more). Targeting for the highest ROI is alligned to the goal for profit organizations, right? But if thats the case we would be talking about traditional Pareto and this is not what TOC claims what we should do.
Id be very happy if i could receive your slides of the controversial webinar.
Please keep up with your blog as I find TOC fascinating.
Regards from Lisbon,