Power and politics in enterprise-architecture

Anyone who’s involved in any form of enterprise-architecture would know that it’s best described as ‘relentlessly political’: seems almost everything we deal with turns out to be some kind of tortuously-intransigent wicked-problem.

Which in turn seem so often to be rooted in some kind of power-play or other power-problem. There’s almost always some kind of power-problem behind each anti-client issue, for example.

But if so, what is ‘power’, anyway? And how does it cause those power-problems that plague us in enterprise-architecture?

Over the years I must’ve looked at dozens of definitions of power, particularly in relation to politics and the like. Yet to me, almost none of them have made much sense, especially over the whole of that huge scope that we cover in EA: they’ve depended too often on too many special-cases, special-pleadings, or down-and-dirty term-hijacks and the like. So the definition I keep coming back to is one that’s drawn from formal physics:

  • Power is the ability to do work

And ‘work’, in turn, is ‘the rate at which energy is expended’.

That’s it. Completely flat, symmetric, no special-cases, no special-pleadings. Just power as ‘the ability to do work’.

In essence, that’s a functional definition of ‘power’. It’s what we see in mechanical systems, for example, in relation to mechanical work.

It’s what we also see in human systems, in relation to human work: the only difference is that humans expend energy across a much broader range of types of work – physical mental, relational, aspirational and so on.

And once we link it to some notion of purpose or intent, we can derive metrics of efficiency and effectiveness: how much energy is expended on something other than the intent, for example, or how much it’s kept on-track to purpose. Again, that’s essentially the same for mechanical systems and human ones too – in principle, anyway.

Yet how come it so often goes so badly wrong in the human context? How is it that human interactions and systems are so often so spectacularly inefficient or ineffective – in terms of almost anyone‘s intent or purpose?

The short answer is that that happens whenever anyone uses not the functional definition of ‘power’, as above, but the dysfunctional one:

  • Power is the ability to avoid work


For enterprise-architects, the implications and conflict between those two definitions set out almost everything you need to know about power-problems in the enterprise… It applies sometimes to relationships between mechanical systems; it often applies to relationships between IT-systems; but applies especially to relationships in human-systems. A lot. Far, far more often than would like; and probably far more often than most of us might expect.

Which is why getting some clarity around these issues will help a heck of lot in getting to systems that actually work.

Kind of important, then…

Let’s step back a bit. Somewhen around a decade ago, I developed a power-diagnostic called SEMPER, based on the conflict between those two definitions. This gave us a five-point scale to assess ‘available ability-to-do-work’, which in practice was a scale of maximum effectiveness within a given context:

  1. actively-dysfunctional – destructive competition and suchlike
  2. passively-dysfunctional – ignorance of evasion of responsibility
  3. neutral – best that can be achieved with command-and-control
  4. local-effective – control is dropped, enabling local self-adaptation
  5. system-effective – command is also dropped, enabling autonomous systemic-adaptation

The scale is applies in much the same way to all types of systems – machine, human and/or IT – but for the moment let’s focus just on human systems.

In most business contexts we’ll find that level-5 is fairly unusual – in part because of the motivation-dilemma – and even when we find it, tends to be somewhat unstable, decaying back to a level-4 unless rigorous attention is paid to keep it going. Level-4 is achievable and maintainable just from system-design, though: and given that most organisations seem to get by on a muddled mixture of level-2 and level-3, a stable level-4 across large areas of an organisation represents a huge difference in enterprise-effectiveness – reflected in much higher productivity and profitability overall.

So how does this work in practice? And how do we use such concepts of power and power-metrics to guide the design for enhanced effectiveness of human-systems?

A couple of assertions first. (There’s a lot of psychology and suchlike to back up these assertions, but I’ll skip over that depth of detail for now.) These are:

  • the only source of human power is from within the self (‘power-from-within’)
  • individual people can assist each other in finding that power-from-within (‘power-with’)

(There’s also the obvious point that conversion of food to energy, and suchlike, will play a significant part in individual ability-to-do-work, but I want here to concentrate on the power-issues that arise from how people interact with each other.)

That’s the functional side of power. And one of the most important points here is that power in this sense is a personal responsibility (literally, ‘response-ability’): it actually can’t be ‘taken’ from anyone else. But therein lie the roots of a huge variety of interpersonal confusions, because there’s very strong delusion and desire that power can be ‘taken’ from others which is exactly what underpins the dysfunctional definition of power as ‘the ability to avoid work’. Instead of power with others, it becomes power used against others, in an inherently-futile attempt to ‘take’ those others’ power. In practice this comes out in two distinct forms:

  • offload responsibility onto the Other without their engagement or consent (‘power-under’ – colloquially known as abuse)
  • prop Self up by putting Other down (‘power-over’ – colloquially known as violence)

Those are the ‘win/lose’ versions. Although somewhat less common, there are also ‘lose/win’ versions – respectively, ‘take on responsibility from the Other without engagement or consent’, and ‘put Self down to prop Other up’ – which are, however, arguably no less dysfunctional than the ‘win/lose’ forms.

In practice, whenever power-under becomes predominant in a context – “it’s not in my job-description, it’s not my responsibility” – we see SEMPER level-2 ineffectiveness; and whenever power-over becomes predominant in a context – “it’s dog-eat-dog out here” – we see SEMPER level-1 ineffectiveness.

Worse, it tends automatically to fall into a downward vicious-spiral: the minor dysfunctionalities and irritations of level-3 ‘neutral’ tend to drive level-2 silo-based responsibility-avoidance, which then spirals downward into full-on level-1 blame-fests and the like. Not fun, for anyone…

The trap here is that the attempt to ‘take’ power from others doesn’t actually work: the needed work either isn’t done at all, or is at best done with low overall effectiveness. And because it looks like we’ve been able to force the Other to do our work for us, yet it doesn’t actually work, it’s highly addictive: exactly the same drivers, in fact, as for any other type of addiction.

The blunt reality is that there is no such thing as ‘win/lose’ or ‘lose/win’: they’re both illusory forms of ‘lose/lose’. In every interaction, the only functional choice we have is to ensure that everyone wins – otherwise everyone loses. And that’s a requirement that applies everywhere, to everything – not just to human relations and human-systems, but to machine- and IT-systems too.

In my understanding, that needs to become a fundamental design-principle for any form of enterprise-architecture: that we must always assess a context for potential power-issues, and we must always design for an identifiable ‘win/win’.

In future posts I’ll explore in more depth what this looks like in real-world practice, and how we can incorporate that design-principle into real-world architecture contexts. And if this is of immediate practical concern for you, perhaps take a look at the earlier post ‘Power, people and enterprise-architecture‘, or the ‘manifesto’ reference-sheet on power and responsibility, for a summary on how this plays out in more general terms within the workplace.

For now, though, I hope this is enough for you to get started with? Over to you for comments, anyway.

Leave a Reply

Your email address will not be published. Required fields are marked *