Yet more on ‘No jobs for generalists’ [3d]

Why is it so hard for enterprise-architects and other generalists to get employment as generalists – despite the evident very real need for such skills in the workplace? And what part do current business-paradigms play in this problem?

[[This post has been split into six parts, in line with the chapter-headings:

• Part 3a: ‘An incomplete science’ – about Taylorist notions of ‘scientific management’
• Part 3b: ‘Management as a service’ – a service-oriented view of the role of management
• Part 3c: The impact of the ‘owners’ ‘ – about how and where financial-investors come into the picture
• Part 3d: ‘A question of fitness’ – exploring the use of ‘fitness landscapes’ to guide selection of appropriate architectures
• Part 3e: ‘A question of value’ – how we could describe the business-value of generalists
• Part 3f: ‘No jobs for generalists?’ – a brief(ish) summary of the whole series

Previous: Part 3c: ‘The impact of the ‘owners’‘]]

A question of fitness

One of the hardest challenges in all of this, for enterprise-architects, is keeping clear of the politics: so much of this runs so counter to deeply-held beliefs and deeply-entrenched vested-interests that this whole area of enquiry will often seem like a career-minefield…

The practical problem we face is that, in relation to current business-realities, those deeply-held beliefs simply do not work ‘as advertised’; and many of those deeply-entrenched vested-interests are actively damaging the viability not just of individual organisations but of the entire business-ecosystem in which all of our organisations operate. Some aspects of that, I’d hope, should be clear from what you’ve read (and, I hope, tested in relation to your own experience?) in this series of articles and elsewhere in this weblog. And if something in the enterprise doesn’t work, it is our professional responsibility as enterprise-architects to reduce and resolve the respective risks to our organisations. Yet even to talk about it – let alone publicly do anything about it, in terms of architecture-designs – will often seem so dangerous politically as to be tantamount to career-suicide, or worse. So what can we do here that would sidestep those dangers? – and actually achieve viable, verifiable, sustainable results for our client-organisations?

The first and most essential requirement is to keep emotion out of the picture as much as practicable. A lot of us will feel decidedly passionate about this – I certainly do, anyway 🙂 – but emotion doesn’t help: all it does is make us into easier targets to attack. There’s also, of course, the very real possibility that we’re the ones who are wrong about some or all of this… So we do need some mechanism or ‘safe-space’ through which to explore all of the arguments whilst keeping the emotions at bay.

And one of the best options for this would be some variant of what evolutionary-biologists describe as a fitness-landscape.

In business, we talk about ‘fitness for purpose’: but how do we describe what ‘fitness’ is? If possible, how would we measure it? How could we compare different ideas, different models, different designs or implementations, in terms of their ‘fitness’? That’s where a fitness-landscape can help.

[Note that what follows is somewhat different to the strict formal use of fitness-landscapes in evolutionary-biology, where ‘fitness’ is described in terms of the viability and adaptability not of an individual but of a whole species to its ecological context. (Hence ‘survival of the fittest‘, as an exact technical term rather than as a half-baked sociopolitical slogan…) Just note that the usage here isn’t quite the same, in fact in places we use it in a kind of back-to-front way: if that worries you, think of ‘fitness-landscape’ merely as a useful metaphor – nothing more than that. Okay?

(By the way, for another example of business-oriented usage of fitness-landscapes, see Ralph Ohr’s post ‘Evolutionary and Revolutionary Innovation‘ – recommended.)]

A key challenge for any form of fitness-modelling is the choice of parameters for the landscape – because that can be viewed as ‘political’. Perhaps the safest approach here is to view it as a form of context-space mapping: choose a politically-acceptable base-map, and then build overlay cross-maps onto that base. For example, we could start from the sensemaking version of the SCAN frame:

Remember that in SCAN the vertical axis is ‘time-available until decision/action’, and the horizontal axis is a spectrum of modality from fully-predictable to inherently-uncertain. The key boundary on that horizontal axis – the position of the red dividing-line – is the Inverse-Einstein Test: where doing the same thing either always leads to the same results (left/blue), or may lead to different results (right/red).

The fundamental concept that underpins the linear-paradigm is that everything is ultimately subject to explicit, certain, predictable control. By definition, anything based directly upon that paradigm – such as Taylorism and its concept of ‘scientific management’ – must follow the same assumptions. If we use the SCAN frame as a fitness-landscape, what that tells us is that Taylorism will only succeed – have strong ‘fitness’ – where the context is always predictable: it won’t – in fact can’t – have high fitness for anything inherently-ambiguous or inherently-uncertain, over on the far side of the Inverse-Einstein Test. Or, to put it the other way round, wherever we hit anything in business that is not inherently predictable and controllable – and a lot of things in business are not – then Taylorism is not going to be a good fit for that business need. Which means that if everything in that business-context is built around Taylorism – and it still so often is – and/or if Taylorism is the only choice we’re allowed, then by definition we will have business-problems there that we will not be able to solve…

And yes, it gets worse. Remember that Taylorism insists on a strict separation between ‘brain’ and ‘brawn’: managers think Complicated thoughts, whilst workers do only the Simple things. Yet work takes place at the point of action, whilst analysis takes place at some distance in time and, usually, space from the point of action – which leads us straight into the Taylorist trap. Whenever the workers hit anything that doesn’t match up with their Simple work-instructions – in other words, wherever reality shifts over to the far side of the SCAN frame – the decision must be ‘escalated’ to a ‘higher’ level: but that ‘higher’ level usually lacks the contextual knowledge needed to solve the problem. And even if the knowledge can be found from other than at the point of action, it all takes time – hence a response that may be too slow to keep pace with the real-time action, and perhaps, if we’re not careful, a total analysis-paralysis.

The standard Taylorist solution to this is to try to speed up the analysis-process, to bring the Complicated decision-making closer to real-time: hence neo-Taylorists’ obsession with automation. Yet the catch with automation is that, unlike humans, it has no means to cope with anything on the far side of the Inverse-Einstein Test – which means that as soon as does hit against something that it doesn’t know how to control, it’s stuck: it can’t make any decision at all. So we end up with a kind of ‘arms-race’, where system-designers try to add more and more and more ‘business-rules’ to the system, to try to plan for every possible eventuality: yet because Murphy’s Law is the only true law in town, there’ll always be yet another ‘something-that-doesn’t-fit’. In short, any system that attempts to rely exclusively on linear-paradigm automation is guaranteed to fail at some point in any real-world context.

Let’s turn this the other way round. It should be clear by now that Taylorism cannot succeed as ‘the one and only model’ for all business needs. Yet for all its evident flaws and limitations, Taylorism and its ilk are a good fit for some aspects of business-needs. So, for example, if we map out all the different types of business-needs in terms of the SCAN frame – predictable versus ambiguous, real-time versus time-available-for-experiment, and so on – then there will certainly be areas in that cross-map that do fit well with Taylorism’s assumptions. If we then add another dimension to the frame to represent Taylorism’s degree of fitness at that point, we have a fitness-landscape that tells us where and where not to use Taylorist-type approaches, and the extent of potential reliability or risk if we do use those approaches for that business-need. And we can do the same with all the other techniques and management-models that might be supported within our business – leading to a context-sensitive management-architecture which selects appropriate techniques according to their respective fitness to the context.

Conversely, a single model is a point-solution within the overall fitness-landscape. So if we purport that Taylorism – or Agile, or matrix, or hierarchy/flat, or whatever – will always be the best model for every context, what we’re actually saying is that the fitness-landscape consists only of the parts where our model fits well, and that somehow the rest of the business-world does not exist. Which is definitely a back-to-front approach to enterprise-architecture… if we’re working with a true whole-enterprise scope, we need to work with the whole of the fitness-landscape, not just the parts that happen to fit well with our preferred ‘solution’!

There’s one more challenge that we need to note here – and it is an important one. That mapping above assumed a static relationship between business-needs, the SCAN frame, and the management-models and suchlike whose fitness we want to assess on that combined cross-map. The catch is that the business-world is dynamic, not static: every context is subject to what I describe as its variety-weather, the variety of the variety in that context. Things change; parameters change; the nature and the pace of change itself can change, in complicated, complex and even chaotic ways, itself changing and changing over time. In short, it ain’t simple. 🙂

The fitness-landscape is in part defined by the business-needs: so if those business-needs are dynamic, then effective-fitness will also be dynamic too. In effect, fitness within a context changes along with the variety-weather in that context – and that can mean that at times we might have a ‘fitness-landscape’ that’s as turbulent as a stormy ocean. Taylorism does demand predictability, but can actually have high fitness for almost any level of complication in hard-system terms, as long as everything remains much the same: it only gets into trouble if there are too many exceptions to the expectations of its algorithms, or if the pace of change in the number or content of its parameters is faster than the overall system’s ability to adapt. So it does work well for the kind of near-stable, slow-changing business-contexts that applied for much of the past century – which is why people came to rely on it so much. Yet that’s no longer the kind of contexts that most businesses deal with these days: the variety-weather is changing so fast, and from so many different directions, that it can seem more like the turbulence of a hurricane – and all the signs suggest that there’s still a long way to go before there’s any likelihood of it easing off.

So Taylorism isn’t ‘wrong’ as such, and neither is classic hierarchical top-down management inherently ‘wrong’ as such: I hope that should be clear by now? Yet what the fitness-landscapes show us is that there are fewer and fewer contexts at present where they would fit well – and even those contexts where they do fit might cease to be so at any moment. Hence, clearly we do need alternatives that can cope well with this type of turbulence: alternatives that are more resilient, more adaptable, more self-adaptable too. And those alternatives all depend on a much more explicit role for generalists who can help to take up the strain, and who can make and remake the changing connections that keep the enterprise on track to its aims even in the stormiest of variety-weather.

In other words, generalists such as whole-scope enterprise-architects. Who need employment as generalists. Which is where we started on all of this exploration.

Context-space maps, fitness-landscapes and similar techniques may seem somewhat technical, even tedious at times. Yet the real reason we use them in ‘politically-dangerous’ domains such as management-architectures is that they help to keep the focus on reason rather than emotion. We can map different contexts according to explicit, clearly-defined parameters; we can measure fitness to cope with ambiguity in an unambiguous way. It’s quite hard for others to attack us if our arguments are demonstrably based on the types of parameters that they themselves would choose.

In short, tactics such as fitness-landscapes really do help here: use them to take the heat of the otherwise-inevitable arguments about Taylorism, management-hierarchies and the ‘invisibility’ of the generalists’ role in business.

[[Next: Part 3e: ‘A question of value’]]