This is a transcript of episode 139 of the Troubleshooting Agile podcast with Jeffrey Fredrick and Douglas Squirrel.
Frederick Taylor gets a bad rap - even from us, as we had a lot to say in Agile Conversations about how his methods have been misapplied in “software factories”. But in the right circumstances, his ideas about repeatable, simple processes have a lot of value, and can lead directly to valuable automation opportunities. We describe examples including how continuous integration developed from an arcane art to a routine process along Taylorist lines.
Squirrel: Welcome back to Troubleshooting Agile. Hi there, Jeffrey.
Jeffrey: Hi Squirrel.
Squirrel: Today we’re going to defend something that we usually bash.
Jeffrey: I was listening to Gene Kim’s IdeaCast podcast and in particular the latest episode, where he was talking to some of the co-authors of Team of Teams, a book that I’ve known about for many years. The lead author is General Stanley McChrystal, though there are some co-authors who were part of the Joint Task Force of Counterinsurgency against al-Qaida in Iraq. In reading through the book, I came across a section that shed a new light on something we discuss in our book: trying to escape the software factory–by which we meant Taylorism, which is a common bogeyman of Agile and Agile teams.
Squirrel: We better explain what that is, because some of our audience may never have operated in a software factory.
Jeffrey: Taylorism describes what became the dominant paradigm of industrial manufacturing and of organisational life in the 20th century. It was also known as scientific management. If you think of someone with a clipboard and a stopwatch doing time and motion studies to make sure that each work step as being as efficient as possible, then the first person who would have been holding said stopwatch would have been Taylor: the inventor of this approach to break down jobs so they are as simple as possible. This directly fed into the assembly line and workers doing a series of unskilled or at best semi-skilled tasks in sequence and doing the same activity over and over again.
Squirrel: Instead of a stopwatch, you could even imagine somebody sitting there with a burn up chart and measuring story points, which would describe quite a lot of my clients and people in our audience.
Jeffrey: When we talk about Taylorism we usually don’t mean the literal application, but instead the Taylorism mindset. This was the surprise to me reading through Team of Teams when they start talking about Taylorism, because they started talking about how great it was! The fantastic innovations of Taylorism and what a dramatic improvement it was over what came before. Which was shocking to me, because for me Taylor has been the villain for so many years,
Squirrel: We don’t like those very reductionist approaches where you take something that we think of as complex knowledge work and reduce it to this very simple activity.
Jeffrey: That’s right. I’m going to invoke something we have talked about before, which is the Cynefin framework, which divides problems into different domains, the simple, complicated, complex and chaotic. This Aha! Moment I had while reading Team of Teams was that the idea of identifying simple things that could be made standard was a real innovation and a genuine step forward, because the world that Taylor was immersed in was not what we know today. We live in a world that was created by Taylor and his descendants, by these management consultants of which Taylor was the first. He came into a world where factories, while they had a bunch of new machinery, the thinking in them hadn’t changed from a Craftsmen Guild artisanal mindset. Looking at it through lens of the Cynefin framework, every problem effectively was treated as the domain of experts. This idea that everything was complicated was a real barrier to what was happening. The idea that everyone was a craftsman meant there was no standardisation. Everyone had their own way of approaching the problems without a lot of justification for thinking why their way was the right way or the best way, it was just the way that they had. Then Taylor came through and said ‘first of all, let’s look for facts. Can we find a way that’s actually better than the others?’ You said if we were dealing with bricks, then there’d be a simple thing we could do to to measure productivity:
Squirrel: Count the bricks. I was describing this to a client earlier this week, they were asking, ‘how do we interpret this board pack that tells us that how many story points per half a fortnight…’ And I said, it’s not actually revealing anything because in our domain, in the Cynefin complex domain, you can’t do what you can in the simple or obvious domains. If you’re building a brick wall, you count the number of bricks per hour. That’s a meaningful measurement. In our world most of the time that’s not meaningful. But you discovered some cases where actually we can count the bricks.
Jeffrey: I was thinking that if we don’t appropriately apply the Taylor mindset, it’s easy to be stuck in a world where things are understood as obscure crafts. I realised that I had lived through a real change of this: when I started in the 90s it was common that there would be a build engineer. This was a separate job, a separate domain, a separate discipline. Someone whose job it was to build the software and deliver it. This was our version in the software world of saying, ‘there’s a guild of build engineers that do some sort of arcane, complex steps and there’s no way we could do it.’ But then we went through a Taylorist approach: maybe we could distil the steps of the build, rather then something that someone feels their way through, it could be something we can write down into a series of simple steps and put it into a run book. Once it’s in a run book–once we know what we’re doing–we can automate it. We can get even better. We can go and take these simple steps and rather than having a human, we could get the computer to do it for us.
Squirrel: Taylor was operating in the 1910s, so he didn’t have computers, but if you went into the factory and said, ‘Hey Mr. Taylor, I’ve got these nifty things called robots. Would you like to have them do some of the work?’ He would be very excited about that.
Jeffrey: Absolutely. That’s what his output was. He got people to act as robots, to do the same task again and again. This is where some of the problems of Taylorism comes in, as humans are not good robots.
Squirrel: But computers are really good robots!
Jeffrey: Exactly. So one of my Aha! Moments was that there’s a reason for us to keep the positive aspects of Taylorism in mind, which is to be thinking about the steps we’re doing and saying ‘is there something we’re treating as complicated when actually it could be made simple? Could we make this simple and systematised, where we’re always doing it the same way every single time?’ If we can do that, maybe we can actually automate it.
Squirrel: We don’t even have to always go as far as automation. I have a client which builds rooms to a standard, and they’re building a lot of these rooms. In their I.T. function they’re creating these run books. We haven’t invented actual robots that are electricians that can go into a room and install wires in the walls, unfortunately that doesn’t exist yet. But what they have done is taken the process which was artisanal, of setting up a room to a particular standard. They can create a standard room of their type according to a run book that they’ve written down. Although they still have an awful lot of challenges, and sometimes the electricians turn up and there’s nothing for them to do, but they are able much more repeatably to do the same processes again and again. That leads to greater standardisation and much greater efficiency, which means they can roll out a lot more of those rooms a lot more quickly, which is important in their industry.
Jeffrey: I think that if people give it some thought, they’ll probably find many opportunities to apply this kind of thinking. The value of thinking in a Tayloristic fashion is: can we standardise? Can we simplify, and can we get efficiency from that? This is a thought that is not outdated. It was an innovation and it’s an innovation that can still be applicable and fight our human tendency to be a bit lazy at times and just focus on getting the work done in the way that we currently know how, rather than asking ‘is there a better way to do this?’
Squirrel: If listeners are applying Taylorism, consciously or otherwise, we’d sure like to hear about it, and you can find us most easily at ConversationalTransformation.com. Thanks Jeffrey.
Jeffrey: Thanks Squirrel.