Theory of constraints and Specification by example part II

· June 1, 2011

A lot of people read and appreciated the last post on Specification by example and Theory of constraints, so I thought I do a follow up.

I often find that your closest people are your best critics, so I asked a few of my colleagues for feedback. Håkan Forss (@hakanforss) is one of them that I respect very much in matters like these. He, like me, read and loved the Goal and I know that Håkan has great knowledge about theory of constraints and applying that kind of thinking in the system development process or other kind of knowledge work.

So it came as no surprise that Håkan had a few, rightful, objections to my reasoning. He’s points made me think another lap and I thought that it could be interesting to present it here.

To be able to follow this reasoning you should read the earlier post. Go on – we’ll be waiting, or maybe just start the next section before Winking smile

The five focusing steps

Håkans main point has to do with a concept in theory of constraints called the five focusing steps. They are a number of steps that aims to keep the ongoing improvement process going in an organization. From Wikipedia we learn:

Assuming the goal of the organization has been articulated (e.g., “Make money now and in the future”) the steps are:

  1. Identify the constraint
  2. Decide how to exploit the constraint
  3. Subordinate all other processes to above decision
  4. Elevate the constraint
  5. If, as a result of these steps, the constraint has moved, return to Step 1. Don’t let inertia become the constraint.

There’s a couple of things worth thinking about here in the context we’re talking about; system development.

Assuming the goal of the organization has been articulated

What is the goal of a system development project? Is it always the same? In the example of Wikipedia it’s “Make money now and in the future” for any organization.

The goal of a system development project I presume would be something like:

Develop a system that meets the need of the users

But this is entering dangerous “let’s write down a proper academic definition” land. That’s not the purpose of this post, and way out of my league…

The main thing here is that you should think about the goal of your project. It may differ from project to project and it affect how you reason and make tradeoffs in the five focusing steps.

Identify the constraint

How easy to write those words, but much harder to do. In the last post I used a factory as the example (Marcusoft Welded Steelplates) because it’s physical and concrete. In this world bottleneck would probably manifest itself as the steel plates being queued up before the machine that is the bottleneck, or people and machines with nothing to do (behind the bottleneck).

In our, invisible knowledge work, we need to help work to be visible. That actually a core principle of Kanban, as shown in this picture from the Kanban101 site.

Kanban Core Principles

So if you have a typical Kanban or Scrum-board a bottleneck will halt the progress of work, or you’ll see work that nobody is responsible for, items that have been blocked or empty “columns” (process steps) with no work to do.

BUT, in the last post I suggested that the bottleneck was the lack of knowledge. That’s actually the cause of notes being stuck in certain process steps. There’s (most commonly) not a step in your process called “gather all required knowledge” – unless your doing strict waterfall Smile.

So to identify the bottleneck here requires a reasoning and investigation of WHY the work has halted, came back to us, was abandoned etc. A great way to do this reasoning is to use root cause analysis (great paper on this here), which is another powerful Lean technique.

Exploit the constraint

Oh I love this step. It’s so counterintuitive and just beautiful. Instead of adding resources and elevating the bottleneck (step 4, two steps away) you first try to do the best from what you’ve got. Love it!

Ok, assuming that lack of knowledge is our constraint (more on Håkans criticism of this reasoning below), what should we do?

Acknowledge the fact that we don’t know!

And how do we exploit that constraint?

Align your process to the fact that knowledge is discovered as we go.

(To be quite honest I think that’s mixing in step 3 (Subordinate all other processes to above decision), but that is not important for the reasoning here.)

That’s the whole thinking behind most (all?) agile methods. That’s why we release often to get feedback, that’s why we test-drive our code out to get feedback early, that’s why we do small slices of the application at the time.

And that’s why Specification by example sits very good (for me at least). The whole idea is that you use the examples to talk about the system in terms of the user, so that all roles can ship in and have their say. Even before the code is created.

The examples then become a way for the developer to keep track of their progress through the implementation of the feature, by use of the outside-in development (ATDD) approach.

Finally the examples end up describing what the system does, as part of the living and ever changing documentation of the system.

Don’t let inertia become the constraint

Håkans criticism was that I always assumed that the lack of knowledge is the constraint. (There was also mentioning of me having a great hammer and seeing nails everywhere… Smile). This is actually even mentioned in the five focusing steps, as the inertia to resist change.

I think that it’s absolutely correct; you should enter this with an open mind and not assume that you KNOW where the bottleneck is. You might just end up optimizing parts of your process that not is the constraint, and theory of constraint teaches us that gets us nowhere. The constraint is still there, and it’s still slowing you down as much as before the optimization.

For me though, a break through in my thoughts was when I realized that the constraint might not be a individual step – such as testing or development. The constraint can be the underlying understanding that all the steps are using as the feature progress through our process – such as lack of knowledge of what we’re going to build.


So Håkans criticism made me think – that’s a Good thing (tm). Thank you Håkan for some excellent input.

Did that take my reasoning forward? Well, you should probably answer that…

But I think that many of the problems and constraints we face in system development projects actually has to do with us not knowing enough when we start. Miscommunication, bad specifications, code not matching the specification, hard time to verify the application against the specification are all example of this – manifestations of lack of knowledge.

But I wouldn’t go so far as to say that it’s ALWAYS the case. But often. It’s the “darn close” part of the quote at the end of the first post:

It (specification by example) is no silver bullet – but it’s darn close!

Twitter, Facebook