Why broken?

If the mainstream approach to enterprise-architectures is broken – and so badly broken, at that – then why is it broken? In what way is it broken? What are the core reasons behind why it’s so broken?

And why ask this? Well, short answer is that we’ll need a better understanding of what’s going on before we can do much about sorting out the mess…

To illustrate this, I’ll revisit some of the content from those mixed-up messages referenced in the previous post on ‘Broken’.

[Do remember, though, that those quotes represent merely one example amongst many: it’s not one person or organisation that’s screwing things up so badly, but an entire mindset that’s not just about enterprise-architecture, it’s the same mess turning up again and again in varying forms across almost every industry. So as before, I won’t provide a link to the original, and I won’t say who it’s from: it’s only an example. Fair enough?]

As we’ll see later, we can trace all of the problems seen here back to the same root cause: attempts to apply the linear-paradigm to contexts where by definition it cannot succeed. In effect, it’s exactly the same root-cause as in all the problems that we saw with ‘No jobs for generalists‘: a set of ideas and assumptions that are not inherently wrong as such, but definitely are wrong when applied to the wrong kind of contexts. Which, these days, mostly are the wrong kind of contexts.

Anyway, let’s get started. Here’s the first quote we’ll explore:

…it only influences toward a notional future that is still not totally proven

“A notional future that is still not totally proven”: right there, in just nine words, is the epitome of the linear-paradigm, and why it doesn’t work in business. There’s an explicit assumption there that the future is predictable, that we can have – and are entitled to have – proof that would provide certainty about future outcomes. Nice idea, and yeah, in many ways that would be nice to have, especially in business – but in reality, unfortunately, it is literally unrealistic.

Prediction only works when the real world is kind enough to line up to our assumptions. When it doesn’t line up… well, that’s when things are literally unpredictable. But we must never forget that the real world is the real world: if we don’t get the results that we expect, it’s our expectations that are at fault – not the real world.

A real non-business example: want to know why Mars has been such a graveyard for space-probes? The real reason is that, no matter how good the engineering may be, there are just too many factors there against making a safe landing a certainty – especially for an autonomous probe that has to make all of its decisions on its own. Landing on an airless moon is relatively easy: it’s just straightforward celestial-mechanics and thrusters plus a small amount of course-correction. Landing through a high-drag atmosphere is a lot more challenging – especially in terms of materials-engineering – but it’s still relatively predictable. Yet Mars manages to combine the worst of everything: just enough atmosphere – and unpredictable storms, too – to be a real problem, but not enough atmosphere to slow things down to a stop. Just take a look at what the NASA Curiosity team had to do to get their lander safely on the surface of Mars – and hear the elation in their voices when they did it…

But now turn the focus back to Earth, back to everyday business – and realise that most businesses are dealing with wicked-problems that are often many orders of magnitude worse than that of the Mars lander. And you expect that to be predictable, a ‘proven future’? – c’mon, let’s be a bit more realistic, shall we?

There is only one time when a future can be considered ‘proven’ – and that’s when it’s no longer a possible future, but the unchangeable past. Imaginary worlds are predictable; selected subsets of the real-world can be in some cases be almost-predictable; but in the real-world as a whole, there is always some inherent-uncertainty. Hence in the real world, there is no such thing as a ‘provable’ future – and the idea that there could be such a thing as a ‘provable future’ is a delusion, a fantasy, nothing more than that.

Yet it’s all too evident that it’s a very popular fantasy, a very desirable fantasy. Perhaps especially so in business: predictability and certainty and a sense of being ‘in control’ are the great promise of the linear-paradigm. As Iona Heath put it in the current issue of the BMJ:

So much of policy direction seems to be driven by a Faustian ambition to control the threat of the unknowable future.

Hence it’s a very easy fantasy to sell, and especially so when it’s sold by people who themselves already believe in that fantasy – otherwise known as ‘analysts’, because analysis is the core tool of the linear-paradigm. And the hard part is that there’s usually just enough evidence to make it seem credible – because when the variety-weather of a context stays calm, in effect it is predictable. Mostly. For a while. Until the weather changes…

The simplest summary is that the linear-paradigm does work well when the context is stable. It can have almost any amount of variety – be as complicated as it likes – as long as the variety itself of that variety remains much the same. And that is the kind of contexts that the business-world experienced for much of the past century – particularly from the early 1950s to perhaps the early to mid 1980s. Hence, not surprisingly, an increasingly-entrenched assumption that the linear-paradigm was the way to do things in business.

But the reality is that the linear-paradigm is not ‘the way’, but merely ‘a way’ – and it’s a way that only works well when the variety-weather is stable. Which, these days, is increasingly no longer the case in business: it’s already decidedly stormy, and every trend and indicator suggests that it’s going to get much, much stormier yet. Hence, in current business practice, and other than in certain very narrowly-constrained business-contexts, trying to rely on the linear-paradigm alone is guaranteed to fail.

Which is a problem when we come across an assertion such as this:

You don’t want your strategies following spaghetti roads — you want them moving through your company on logical, straight highways…

Because these days, for current business-contexts, that’s exactly what not to tell a business-client. If we did say that, all we’d be doing is propping up a dangerously-delusional fantasy for the context – a linear-paradigm fantasy that’s guaranteed to fail. Not good; not wise; and, if applied knowingly, not exactly ethical either. Oops…

Anyway, let’s move on a bit, to this:

Techies invariably screw up the business; business guys screw up the tech. For decades we’ve looked for someone to span both — and that’s what Enterprise Architects do.

Here’s the blunt reality:

  • in general, techies do not screw up the business
  • in general, business guys do not screw up the tech
  • both techies and business guys do screw up each others’ expectations and predictions about the business, the tech and the relationships between them

In short, it’s the darn linear-paradigm again, but with an additional twist: our view is right, therefore everyone else’s is wrong.

Perhaps the best way to summarise this is one of Edward de Bono’s ‘Laws of Thinking’: everyone is always right, but no-one is ever right. Everyone is right from their own perspective, yet no-one does or can hold a fully-predictable version of the whole picture. So when there’s a clash of perspectives, all and none of the perspectives are ‘true’: they each hold a partial truth, not ‘the whole truth’. Which means that the moment we have a clash of perspectives, ‘the truth’ is going to be inherently ambiguous, inherently uncertain – and inherently arguable, too.

But here’s the catch: the linear-paradigm can’t handle ambiguity. That’s actually its great strength, because its refusal to accept ambiguity is what makes predictability possible. Yet that same strength will become a serious problem whenever the context is inherently-ambiguous: in a sometimes all-too-literal sense, the linear-paradigm collapses into a kind of insanity there.

Which means that, by definition, anyone who bridges across more than one domain – such as enterprise-architects – must be able to go ‘a little bit insane’, in order to reach above the self-contained ‘sanities’ of each of the different domains. Also, since everyone is always right but no-one’s ever right, there is no single domain that holds ‘the truth’ about the business: so in a viable ‘architecture of the enterprise’, we must assert that everywhere and nowhere is ‘the centre’, all at the same time.

And since business-strategies must, almost by definition, straddle across every domain of the enterprise, there is no way that we could move those strategies “through your company on logical, straight highways” – because the ‘logic’ itself will change as we go through each of the domains.

Just to make life even more fun, there’s another twist here, which we can perhaps illustrate best with the SCAN frame:

SCAN core-graphic (revd 10Nov11)

The horizontal axis here is a measure of modality in the context – in effect, the extent of its ambiguity and uncertainty. The red dividing-line represents what I call the Inverse-Einstein Test: to the left, doing the same thing gives the same results, whilst to the right doing the same thing may give different results, or we may need to do different things to get to the same results. To use Cynthia Kurtz’s terms, the left represents a world of order, whereas the right is a world of unorder.

In effect, the real world bounces us around all of that frame, all of the time. But if we can constrain it for a while, we can usually get to something that sort-of resembles predictability or control. For a while, anyway. That’s how we get things done.

Most of business lives within contexts that always contain some degree of ambiguity and uncertainty – a world of at least some inherent unorder. Most business-folk don’t like this, will spend a lot of effort trying to get away from this – but they’ll know all too well that it is an inescapable reality in their world.

By contrast, most machines and, especially, most digital technology will and must reside almost exclusively in a world of order. (If whatever-it-is isn’t in that ordered world, we’d usually say that it’s not working, or broken.) In effect, it’s a world that really does depend on the linear-paradigm remaining valid within its contexts.

So whilst business guys and techies will both desire an ordered world, most techies will assume that they’re already in it: not just the linear-paradigm, but the linear-paradigm squared. Which means that the mindset most needed to work well with the often enormous complications of the IT-world is also likely to be the least suited to deal with the inherent ‘insanity’ of bridging between different domains.

Or, to put it another way, IT-folks are almost the least likely people to be good at doing any kind of true cross-function enterprise-architecture.

Which is why an assertion like this is kinda troubling:

93% of our clients … have EA [enterprise-architecture] practices whose epicenter is IT

In a viable enterprise-architecture, everywhere and nowhere is ‘the centre’, all at the same time: there is no single epicentre. Asserting that a single point or domain is ‘the epicentre’ of an enterprise-architecture is, by definition, falling into the linear-paradigm delusion that there is a single ‘the truth’ to which all others should align (hence the dread phrase ‘business/IT-alignment’ and suchlike). Centring everything around a single epicentre almost invariably leads to an architecture whose ‘roads’ look straight and logical from that single viewpoint, but are completely broken and unusable when seen from anywhere else. Hence the moment someone asserts that there is a single epicentre to their enterprise-architecture, they’ve already blown the architecture – it really is as simple as that.

We can have a sort-of epicentre to a domain-architecture, because the whole point about a domain-architecture is that is all about that one domain. (TOGAF is a good example of a domain-architecture with its epicentre placed somewhere within the overall IT domain.) And we can use somewhere – anywhere – as a kind of ‘temporary epicentre’, a place where we focus most of our current attention: but that’s very different from asserting that it is ‘the centre’ – a perhaps-subtle distinction, but one that is crucially important in practice.

So what we have here, in that assertion that “93% of our clients have EA practices whose epicenter is IT”, is either one or both of the following:

  • it’s IT-architecture, across one or more sub-domains of the overall IT domain
  • it’s trying to apply IT-oriented linear-paradigm assumptions to the broader enterprise space

What it’s not is enterprise-architecture – a literal ‘architecture of the enterprise’ – because that requires:

  • ability to bridge across all domains, without assigning arbitrary priority to any purported ‘epicentre’
  • ability to resolve conflicts between different instances of the linear-paradigm in different domains
  • ability to work ‘above’ the linear-paradigm, in order to cope with inherent-uncertainty and variety-weather

In short, if it’s not a literal ‘architecture of the enterprise’, dont call it ‘enterprise-architecture’. Please?


Yeah, you’re right, I’ve complained enough already… So in the next post in this brief series, I’ll stop complaining about what doesn’t work, and focus more on what does work: some real, practical, proven tactics for working with a true cross-domain space.

In the meantime, comments, anyone?

Posted in Business, Complexity / Structure, Enterprise architecture, Futures Tagged with: , , , , , ,
10 comments on “Why broken?
  1. Gene Hughson says:

    As a techie, I feel comfortable saying that techies who assume an ordered world are setting themselves up for failure. Context matters; thought/evaluation is required; there is no one true way (for anything). The machines may be binary, but the people and circumstances are not.

    That being said, I agree 100% that IT-centric EA is myopic. It’s attempting to define the whole by way of a component.

    • Tom G says:

      Gene – “Context matters; thought/evaluation is required; there is no one true way (for anything).”

      Yes, exactly. And that applies as much within IT-architectures (and design) as it does elsewhere: the problems arise whenever people forget this…

      We get the same ‘linear-paradigm squared’ problems in the sciences, too: ‘scientism’ elevates the linear-paradigm to something close to a religion, confusing the linear-paradigm presentation of end-results with the definitely-not-linear-paradigm methods via which those results were actually obtained. For the latter, perhaps see one of my favourite books, WIB Beveridge’s The Art of Scientific Investigation – the full content is available on Archive.org for viewing and free download at http://archive.org/details/artofscientifici00beve

      Thanks again, anyway!

  2. Carl Chilley says:

    For many years I have been trying to get my colleagues to understand that deterministic linearity is not only a myth but probably flies in the face of nature. Regardless of which discipline they are from (although there is one exception I know of) the majority of the senior management folk I work with believe that by defining an end point you know where you are going and the likely rewards this will bring. Putting such a belief on a gantt chart only confirms this baseless faith: if you can plan it you can deliver it is possibly one of the more exasperating concepts I have come across but one that often feels right.

    In a way I cannot help but feel we have two issues that need to be addressed: Tom mentioned and wrote about the generalist issue before but the other one which has to sit in parallel with the generalist concern is this deterministic belief: if you know your stuff (sic) then you know how to deliver results! In a society that seems to reward mono centric thinking patterns and the can do belief in the final outcome the notion of multi disciplined folk who know that whatever journey you set out on will change along the way is something of anathema. Trying to explain to folk that autopoiesis has relevance in IT, management and life does get you some funny looks for sure.

    But if you are a generalistic and someone who buys into the chaotic view of reality – and I put myself in that camp – you simply have to get on with it and deal with the linear deterministics the best way you can. Calling what you do ‘disruptive innovation’ helps a little but folk still don’t get it.

    Another thought provoking article, Tom. Thanks.

    The one discipline

    • Carl Chilley says:

      Apologies for the abrupt ending to the post but the one discipline that seems to understand non linear non determinism is psychology. This may seem a strange statement to make but remembering the old quote that “there is not so queer as folk” and noting that folk with mental health issues are often perceived to be in some chaotic state then I find more nor an acceptance for the chaotic than in most others disciplines.

      Again, apologies for the poorly terminated previous message.

      • Tom G says:

        Many thanks for that, Carl – a great summary of so many of the sources of frustration for we of the ‘chaotic’ camp… all too accurate a summary, in many ways… :sigh:

        (Re ‘abrupt ending to the post’, don’t worry – if you check carefully through my posts you’ll find all too many errors – repeated-words, missing-words, oddly-scrambled syntax – and that’s after I’ve checked it several times. Must be gettin ‘ old… 🙁 )

  3. Stuart Boardman says:

    I’m totally in agreement. Great. Where does that leave me?
    The reason that people (and in particular managers) are attracted to the linear paradigm is that it offers a “solution”. They don’t care whether you should really have used non-linear, simultaneous, differential equations as long as it delivers an answer. One answer. An answer that remains true for an effectively unlimited period (managers have close horizons).
    The hardest thing, something that I run up against time and again, is providing a convincing explanation of something complicated to someone who wants a straightforward answer. It’s taken me years to learn to avoid the “too much information” problem – an error I continually made (and still occasionally make), which just results in making them (managers) suspicious. Let’s face it, architects are (in)famous for announcing “well, it’s not that simple”.
    Let’s be clear. I’m not attacking managers – not in general anyway. I’ve worked for some great managers, who have made it possible for me and my colleagues to do our jobs. I’ve even had to be a manager a couple of times, which is an important learning experience. A good manager knows s/he can’t do everyone else’s job but also knows s/he has to make decisions and needs clarity from the experts.
    So somehow we have to learn how to present the non-deterministic non-linearity in such a way that managers (at whatever level) have the confidence to make decisions in the face of uncertainty. It behoves us to do more than tell them how complicated everything is. We have to provide them some level of objective basis to help them make the right decision at this time in the knowledge that they may have to change direction later. Every now and then I do that well. When I do, it’s because I’ve made myself think what decision I would make, if it were my responsibility. That forces me to make sure that I am providing enough information to support such a judgement – and no more.

    What does anyone else think?

    • Gene Hughson says:

      Over the years I’ve been in the position of having someone want “the” answer. I’ve never given it to them because it does not exist. I’ve always held firm that I can provide the technological pros and cons, but that I cannot tell them what they should value more. Pretending that there is one right answer puts you in a difficult position ethically and sets you up to be the scapegoat for all dissatisfaction that results from it.

      • Tom G says:


        “I’ve never given [‘the answer’] to them because it does not exist.” – yes, exactly.

        “I’ve always held firm that I can provide the technological pros and cons, but that I cannot tell them what they should value more.” – yes, exactly. If it’s their choice, it’s their choice – we can’t make it for them, because if we did, it wouldn’t be ‘their choice’…

        “Pretending that there is one right answer puts you in a difficult position ethically” – yes, exactly. A scary number of people seem to want to ignore the ethical dimensions in all of this… 🙁

        “and sets you up to be the scapegoat for all dissatisfaction that results from it.” – yep. 🙁 Which, in far too many cases, is the intention: it’s a Berne-style ‘game’: “if it works, I get the credit, because I was the decision-maker; if it doesn’t work, you get the blame, because you made the decision for me”. Not a good place for any of us to be in – or a wise position for us to place ourselves in, either.

    • Tom G says:

      Stuart: “The reason that people (and in particular managers) are attracted to the linear paradigm is that it offers a ‘solution’.”

      Yes, exactly. When ‘control’-oriented folks feel they’re not ‘in control’, they’ll leap for a ‘solution’ – any solution, as long as it looks and feels like ‘decisiveness’. What they won’t do is sit with the uncertainty, and wait until the appropriate option does come along. This refusal to ‘allow oneself to be comfortable with being uncomfortable’ is probably the core driver for wicked-problems – or at least, why they become so relentlessly ‘wicked’.

      This brings me back to another core theme in that dread struggle with Snowden. In the Cynefin framework, his core tactic for the ‘Chaotic domain’ was exactly that of the linear-paradigm: ‘act > sense > respond’ – that the only thing we should do in the Chaotic is to not be there, that the Chaotic itself is ‘wrong’, ‘unnatural’ or whatever. Contrast that with the much more useful advice “printed in big friendly letters” on the back of the Hitchhiker’s Guide To The Galaxy: “Don’t Panic!”. 🙂 Which brings us to:

      “So somehow we have to learn how to present the non-deterministic non-linearity in such a way that managers (at whatever level) have the confidence to make decisions in the face of uncertainty.”

      Give them a copy of Hitchhiker’s Guide, perhaps? 🙂 Or, for that matter, tell the tale of the original Marathon: the conversation between the runner Pheidippides and the god Pan about ‘staying in one’s centre’ so as to pull back from panic.

      “We have to provide them some level of objective basis to help them make the right decision at this time in the knowledge that they may have to change direction later.”

      As I’ve explored in the work on SCAN and decision-making, I’m far from sure that any actual human decision is rational or ‘objective’. (Edward de Bono makes much the same point in his ‘Six Thinking Hats’, in his assertion that every decision is ultimately ‘Red Hat’, in other words emotion-based.) It seems to me that the role for ‘objective basis’ is to set the ground for an emotion-based decision such that that decision does go toward an appropriate choice. The delusion of the linear-paradigm is that its choices are ‘rational’, when in reality the so-called ‘rationality’ is often a cover-up for the reality of underlying panic, where “any decision will do as long as it’s a decision!”.

  4. Dave Snowden says:

    Tom, I don’t want to get into the argument again, but you continue to misinterpret what Cynefin says about chaos. . The situation you describe above is not chaotic, its complex (in Cynefin terms). In those situations you Probe-Sense-Respond. You absolutely do not “get the hell out of there” you absorb the uncertainty, you test possibilities.

    Chaos in Cynefin is a transitionary state, entered purposefully for innovation. But it also is the end point of a collapse from complacency. Even then Act-sense-respond is not necessarily linear. If you look at the crisis management dynamics one is (the shift back to simple), but that is depreciated. Best is to create enough constraint that the system becomes complex and you can probe/experiment again.

Leave a Reply

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