Power, people and enterprise-architecture

(This 'from the archives' article provides supplementary information for the YouTube video 'Power and people in enterprise architecture - Episode 68, Tetradian on Architectures', released on 16 May 2019. The post was originally published on the Tetradian weblog on 01 February 2011 as 'Power, people and enterprise-architecture' .)

---

We really can’t explore the theme of people in enterprise-architecture without addressing the theme – and problem – of power.

In principle, power should be straightforward. The physics definition – roughly speaking – is that power is the ability to do work. Wherever there’s work to be done – in whatever form that that ‘work’ might take – there’s a need for the power to do that work. Should be simple enough to identify and model that within an enterprise-architecture, surely?

Unfortunately, no, it’s not that simple – because most social definitions of ‘power’ tend to be closer to ‘the ability to avoid work‘. Therein lie lots of, uh, interesting problems for enterprise-architecture…

Hence power is something that we really do need to address in enterprise-architecture – even in an IT-centric architecture, let alone one which covers a true whole-of-enterprise scope. Read on?

To say that questions of power are hugely political is perhaps the understatement of the century… so I’ll take some care here to use definitions and descriptions that are strictly generic. (So generic, in fact, that you should also be able to use them to assess relationships between machines, between IT-systems, and between ‘systems’ in any abstract or concrete sense. More on that later.)

For the architecture, the core requirement is to distinguish between power-relations that are functional (aligning to the physics definition of ‘ability to do work’) versus those that are usually dysfunctional (aligning to that social definition of ‘ability to avoid work’), and then develop an understanding about what – if anything – to do about the latter.

Before we do that, we need some clarity on what we mean by ‘work’. Again following a physics definition, we could say that work is the rate at which energy is expended – and in the human context, usually energy expended towards some desired end. But again in a human context, there are many forms of energy, and hence many kinds of work. Digging a ditch is work; so is solving a technical problem; so is building a working-relationship with others; and so is reclaiming hope from despair, finding the energy to carry on after apparent failure. Different forms of energy are required to do those different types of work – physical, mental, relational or emotional, aspirational or ‘spiritual’ (the latter being defined as related to “a sense of meaning and purpose, a sense of self and of relationship with that which is greater than self” – which applies in the business-context just as much as in anywhere else). Machines are great at doing physical work; IT-systems can be configured to do some kinds of mental work; but only living-creatures (usually humans, in this case ?? ) seem to have the ability to do the other forms as well.

People are great at transforming between energies – converting excitement and motivation into physical work, for example, or gaining aspirational satisfaction from resolving a technical challenge. But they’re also great at avoiding the work – and that’s where the problem really lies.

The SEMPER metric for ‘ability to do work’ uses a simple five-step scale:

• 5: Wholeness-responsibility: ‘command’ relinquished – individual actively committed to organisation and enterprise
• 4: Adaptation: ‘control’ relinquished – organisation enables and support individual difference
• 3: ‘Best practice’: best that can be achieved with a ‘command and control’ organisational model
• 2: Passive dysfunction: silo mentality – locally efficient but globally ineffective
• 1: Active dysfunction: destructive, ultimately often fatal to the organisation

In general, machines will only be able to reach a level-3 on this scale, whilst some ‘autonomous’ IT-systems will reach to a level-4. Only real-people will be able to achieve a level-5, because it requires personal commitment and personal responsibility; but likewise only real-people will fall to a level-2 or level-1, which are the outcomes of an explicit absence or rejection of personal responsibility…

Level-1, ‘Active Dysfunction’, or ‘power-over‘, can be summarised as any attempt, in any form, to prop Self up by putting Other down. (That’s the ‘win/lose’ version: there’s also a less-common ‘lose/win’ version, ‘propping Other up by putting Self down’, which some people might think of as praiseworthy, but in the long run is equally dysfunctional.) We see this often in businesses that believe that they depend on destructive-competition: the sales-people, for example, soon learn that if bonuses are based on relative performance, they can seemingly ‘make more money’ by sabotaging each others’ work, sending the company into a downward spiral from which it is unlikely to recover. Much the same occurs with machine-systems that are (usually unintentionally) set to ‘compete’ with each other, or business-systems with fixed budgets that force business-units to fight against each other for ‘their’ slice of the budget cake. We also see it in sometimes within computing systems, such as mutual-deadlock or ‘deadly embrace‘. Note too that ‘the Other’ may in some cases actually be the self – for example, putting oneself down in the past to provide the illusion of propping oneself up in the present.

Level-2, ‘Passive Dysfunction’, or ‘power-under‘, can be summarised as any attempt, in any form, to offload responsibility onto an Other without their engagement and consent. (As above, there’s also a ‘lose/win’ version, ‘taking on responsibility from an Other without their engagement and consent’, which again might seem praiseworthy but is again equally dysfunctional.) In business, the archetypal form of this is organisational silos, or the “not my responsibility, mate” buck-passing. It can also be seen in Dilbert-like phrases such as “the only reward for responsibility is more responsibility”, and “no good deed goes unpunished” – the latter being closely related to the dangerous tendency to equate responsibility with blame, acting as a huge disincentive against functional behaviour.

Collectively, power-over and power-under could be categorised as ‘power-against‘, since that’s one of its main characteristics: being against something, being ‘against’ others. It’s likewise typified by competition-against – where the aim is not so much to ‘win’ as to make all others seem to ‘lose’ – and (where collaboration occurs at all) by collaboration-against – collaborating with ‘same-as-Self’ against an often arbitrarily-selected not-Self, or Other. By contrast, the other levels are characterised by ‘power-with‘, where competition, for example, usually takes forms in which competition-with is used to push each of the players to greater achievement, and hence, in a sense, everyone ‘wins’.

The crucial point is that power-against – power-over and/or power-under – is inherently ineffective, especially in the longer term, because much of the available effort is placed into avoiding the work, or into finding means to offload it onto others, rather than in actually getting the work done. It’s also highly addictive, because although it provides a short-term illusion that work has been done, in reality the work still remains to be done – leading to a tendency to spend even greater effort in avoiding the respective work. This is especially true where the work to be done is inherently personal, such as relational or aspirational/’spiritual’ work – which by definition cannot be done by anyone else, no matter how much we might attempt to offload it onto others.

This matters to enterprise-architecture because, at a systems-level, much of the inefficiency or ineffectiveness of a system can be traced to various forms of power-against, whether between real-people, machines or other systems. It also represents an increasing order of organisational risk, since level-2 (e.g. silo-mentality) may easily decay to level-1 (e.g. destructive competition); and most level-1 problems can easily become ‘undiscussables’, with so much (self-)dishonesty tied up in them that they can only be addressed indirectly. (The SEMPER model describes various methods for how to tackle this type of ‘wicked problem‘, by the way.)

The colloquial term for level-1 or power-over is ‘violence’, and for level-2 the usual term is ‘abuse’. We’re perhaps at some risk here of wandering off into very dangerous political-territory, but there’s a lot we can learn from looking at the use of those terms in the more usual social contexts. One such context that I’ve worked in professionally quite a bit over the years is domestic-violence (DV). For safety’s sake, I’ll elide over the political minefield associated with DV, other than note that the usual approaches show quite a strong similarity with IT-centric ‘enterprise’-architecture: a small subset of the actual scope, but purporting to be the whole of the scope, and hence acting as a block to actual progress, much as described for EA in the post about ‘Crossing the chasm‘. As with IT-centric ‘EA’, only a small amount of rework is sufficient to capture the actual underlying generic themes, at least to the level where it’s possible to work past the conceptual block. For example, a rework of the ‘standard’ Duluth model on DV yields a set of categories of ‘power-against’ that are directly usable in any business context, entirely separate from any preconceived notions of gender or the like:

• Coercion and threats
• Intimidation
• Economic abuse (price/value mismatches, etc)
• Emotional abuse
• Using privilege (such as hierarchical ‘authority’)
• Using isolation (dysfunctional misuse of ‘need to know’ etc)
• Using children (e.g. issues around disadvantaging parents etc)
• Using others (third-party abuse)
• Minimising, denying and blaming

(A tenth category, ‘Sexuality’, is less actively present in business contexts, but does occur in issues such as sexual-harassment and ‘glass-ceiling’ / ‘glass-floor’ issues – which, by the way, do affect both sexes, not solely women alone.)

There’s a lot of detail that needs to be worked-through there, as can be seen in this reference-sheet. (That’s the ‘neutral’ version of the revised-model which, with a certain amount of thought, can be adapted to assess inter-system and even inter-machine relationships. If you specifically want to assess interpersonal or intrapersonal issues, take a look at the ‘both-gender’ version, with matched gender-pronouns used throughout: be warned, though, that many people may find it personally challenging, even though it’s the only approach that actually works.)

Putting it into practice

So, how do we put all of this into practice in enterprise-architecture? And why should we bother anyway?

It matters, because the core of EA really comes down to a single phrase: “things work better when they work together”. What we’re after in practice, in any EA design, is overall effectiveness – ‘efficient on purpose’, if you like, though there’s actually a bit more to it than that. And functional power is what will be needed to achieve that aim; whilst dysfunctional ‘power’ – power-against, the believed ‘ability to avoid work’ is what will always act against it.

So, in short, we need to maximise functional power, in any and every form, within our EA analyses and our EA models. Conversely, we need to find ways to minimise dysfunctional ‘power’, in every form in which it occurs – between people, between machines, between processes, services, whatever.

To apply all of this:

Memorise those definitions of functional-power – the ability to do work – and dysfunctional ‘power’ – the purported ‘right’ or whatever to avoid work. Watch for how each of them occur and act out in practice in any EA context.

Memorise those definitions of power-over – ‘any attempt to prop Self up by putting Other down’ – and power-under – ‘any attempt to offload responsibility onto the other without their engagement and consent’. Watch for how each of these occur and act out in practice in any EA context – and remember that any occurrence of either of these will reduce overall effectiveness and increase organisational and/or enterprise risk.

Search always for ways to convert ‘ability to avoid work’ to ‘ability to do work’. (This is closely parallel to Deming’s exhortation to ‘Drive out fear!‘, by the way – and is needed for much the same reasons.) Seek it out in performance-metrics, for example, that currently reward behaviours that are damaging to overall enterprise effectiveness. Seek it out in systems that assert priority over other systems (‘Using privilege’) without any defensible reason for needing to do so. Seek it out in systems that focus only on financial metrics (‘Economic abuse’), without reference to other forms of value that are relevant or important to the overall enterprise. Seek it out, and find ways to restructure the systems towards a functional ‘ability to do work’ relative to the vision and values of the shared extended-enterprise.

Seek always to express our own responsibilities in this: for example, as Nick Gall put it, “Motivating people is more important than modeling them“. And in a practical expression of ‘power-with’, seek always to find ways to collaborate with others – organisational change, knowledge-management, security, safety, quality, strategy and just about everything else – especially where our authority or responsibility overlaps with theirs.

Expect this to be challenging, and highly political, every step of the way.

Have fun? ??

Update: I forgot to mention that there’s a detailed ‘manifesto‘ on all of this in the business context that I wrote some years back, as part of what became my book Power and Response-ability: the human side of systems. Perhaps take a look? – free download of three-page PDF here, anyway.