Policies are Constraints After All

Introduction

This is a short technical discussion of why I think policies are constraints after all. Social media may not be the best place for such a technical discussion, but I'll try.

I should state upfront that personally I have no real stake in whether polices are accepted as constraints or not. However, the more that I thought about it, the more it seemed wrong to me that they should not be considered as constraints. See if you agree.

For this discussion I want to once again use the causal loop diagram from Senge. I previously used this diagram in an article on Theory of Constraints and Systems Thinking. This is the causal loop diagram, redrawn from page 97 of The Fifth Discipline.

Let's run through it once again. On the left we have a amplifying loop for something positive and this is expressed as a growing action. On the right we have a damping loop for something limiting and this is expressed as a slowing action. On the far right we find the input of this limiting condition. The consequence of these two opposing actions is the condition in the middle. The example I used in the previous article was the Malthusian argument for population growing exponentially and food growing linearly. Population on the left, the growing action, food supply on the right, the slowing action.

Now, with that in mind, let's turn our attention to the vexed issue of policies and whether they are constraints or not.

Why Are Policies Not Constraints?

It is my inference that Goldratt's decision that policies cannot be constraints is technical in nature, and if there is a flaw in that technical argument we should, as a duty, try to correct it. Certainly "in the beginning" polices were indeed considered to be constraints. And for a long time after, both policy and physical constraints enjoyed, if you like, the same space. So what happened?

I believe that when Goldratt applied the five-step focusing process to policy constraints, it failed to yield a satisfactory answer as it had for, say, internal physical constraints. The five-step focusing process as you may recall goes like this:

1. Identify the constraint.
2. Decided how to exploit the constraint.
3. Subordinate everything else to that decision.
4. Elevate the constraint.
5. If the constraint has been broken, go back to step 1.

It seems to me that Goldratt's objection was based upon the fact that you can't exploit a policy constraint. You can't have more of a bad thing! In fact, you wouldn't exploit or even elevate such a constraint, you would eliminate it instead.

While I appreciate this argument and have subscribed to it for many years, if the exploitation issue is the crux of the matter, then I think there is another way of looking at it. Time to bring that causal loop diagram back into the picture.

Why Then Should Policies Be Constraints?

Lets modify, or rather supplement, Senge's diagram. It looks like this.

We still have a limiting condition on the right leading to the slowing action or dampening loop. But equally there must be an enabling condition on the left leading to the growing action or positive reinforcing loop. And what is in the middle? Well our constraint is in the middle. Let's work through this idea for the case of a physical constraint first.

We could write out the causal loops for a physical constraint like this.

If we identify a physical constraint, then we have identified something which has a long-range capacity that is less than any other part of the process. When we decide how to exploit such a physical constraint, in essence we find new ways to increase its utilization. The obvious ones are increasing up-time and reducing wait-time, to name two. Less obvious ones are not making things that are not needed at this time, or not making things that yield a low Throughput$ when there is an unmet demand for products with a higher Throughput$. These are all growing actions. Conversely, continuing to do the opposite, creating mis-utilization, are the slowing actions that give rise to the constraint in the first place.

But of course there is also something more. It is our prior assumption that all parts are equal, or that all parts have near balanced capacity. Or delving a little deeper, that dependency and variation are decoupled by the presence of considerable work-in-process between each of the stages in the process. These are all important considerations.

In terms of our causal loop, the constraint remains the constraint for the period that it is most limiting. But early on the limiting condition, the mis-utilization, causes it to limit the whole process. Later, that very same constraint, empowered if you like by the enabling condition, the utilization, enables the whole process. The very same constraint can be limiting at first and enabling later on.

Let's now look at the case for a policy constraint.

We have now have thoughtfulness as the enabling condition leading to the growing action and instinctfulness as the limiting condition leading to the slowing action. Let's take a moment to define these.

• Instinctfulness = a large amount of our reptilian or instinctive brain, a large amount of our limbic or mammalian brain, together neatly wrapped up in a rational and logical story produced by our primate neo-cortex. If you like, its mostly our genes.
• Thoughtfulness = a tiny bit of our reptilian or instinctive brain, a small amount of our limbic or mammalian brain, and a very large amount of our rational and logical primate neo-cortex. If you like, its mostly our memes.

We need to understand this because most policy constraints are instinctful. So instinctful that they are automatic or implicit, or if you are unfortunate enough to work in a bureaucracy (or a hospital) then these implicit rules for about just about anything and everything can also become explicit as well. Ackoff provides a useful definition of policy: they are rules for making rules.

Now, have you ever seen someone instinctively make a small batch in a work setting, or for that matter holding a customer's order back for later release? Those are the sorts of actions, the sorts of policies, that we are talking about. They are manifestly instinctful, everybody does them.

So back to Eli Goldratt. Let's assume we have large ("economic") batch sizes. In drum-buffer-rope these will become constraining at some point. Usually they are addressed ahead of time and so they are not so apparent. But what would happen if we didn't address this up-front. Well, firstly, we would be making "stuff" now (part of a big batch) that would not be needed now, may in fact never be needed at all, but that is another story. Secondly, depending on the transfer batch (um) policy, each large bolus of work slows down the whole process to the rate of progress of that large bolus. If nothing else, this will cause starvation of the constraint at sometime (even when their is work everywhere). So, let's assume for a moment, specifics aside, that a large batch as a policy will damage a physical constraint at some stage, in one way or another. It will itself become the new binding condition, and it will be instinctful. No one runs around yelling: "cut the largest batches in half!"

If we go through our five-step focusing process we have done step 1, we have identified a constraint to the system's performance. Step 2 would be to exploit that constraint. Now Goldratt quite rightly pointed out you can't exploit a bad decision (although it has to be said many politicians try). But go back to the causal loop diagram for a second. It is not the presence of the instinctfulness that we need to address, quite correctly we don't want more instinctfulness. It is the absence of thoughtfulness that we need to address, we do need more of that, we do need more thoughtfulness. We can exploit the constraint by increasing the thoughtfulness. And in exploiting thoughtfulness we also automatically subordinate instinctfulness - isn't that a neat trick!

The batch policy didn't go away, the batching policy wasn't eliminated, we still have one, indeed must have one, it was simply modified. Maybe the new policy reads something like the 20% of the batches that contain 80% of the work will be halved, and the frequency doubled (a remarkable small increase in setup when you do the maths). And if that works, well, maybe we will go for a quarter after that. Of course what happens is things begin to flow and, believe me, additional, new, and unexpected capacity at the physical constraint will begin to present itself that you couldn't even imagine.

A policy can be exploited. A policy can be a constraint.

And Here's the Thing ...

If you think about it, all else aside, a policy must be a constraint, because it is a policy, or rather it is a change in policy, that makes physical constraints what they are. We go from an original policy of there is no, one, single, rate-limiting step to "oh yes there is!" Remember, we said that our prior assumption was that all parts in a process are equal, that we assume that all parts have near-balanced capacity. In fact we codify that, that is what line-balancing is. And what is that codification? Well isn't that a policy if ever there was one!

All else aside, the logic says policies must be a constraint because physical machines don't select their own capacities for themselves and put them where we find them, we do. They are the consequence of policy. Likewise physical goods don't arrange themselves in the market and organize for their own distribution, we do. Lastly, customers don't inhibit themselves, or forced themselves to look for alternative suppliers, we do! Our internal and external constraints are always a consequence of our own policies - implicit or explicit - and those policies in-turn are always a consequence of our instinctfulness rather than our thoughtfulness. And our thoughtfulness is eminently exploitable.

Summary

I hope that this small discussion will go a long way to clarifying how we understand policy constraints. Policy constraints like physical constraints can be limiting to the system as a whole, and they can also be enabling for the system too. A policy constraint like a physical constraint can be exploited and to do so something else must be subordinated. Moreover, physical constraints are simply the concrete manifestation of our policies in any case. Lastly our policies are simply a causal loop, a tug-of-war, between our instinctfulness and our thoughtfulness, one a slowing action, one a growing action, one a limitation, another an enablement.

Find many other articles here, and my website Introduction to Theory of constraints here.