13 years, 3 months ago

Round in circles on enterprise-architecture

Link: http://weblog.tomgraves.org/index.php/2011/04/06/round-in-circles/

One of the real pleasures of enterprise-architecture is that it covers the entire panoramic panoply of the enterprise, the many ways in which everyone and everything can work together towards a shared goal, creating a common bridge from Why to How to What and When and Where and Who.

One of its huge frustrations, though, is the way in which people in the IT space insist on calling what they do ‘enterprise architecture’.

Which it isn’t. At all.

IT-architecture is about machines – and a specific class of machines at that. It has almost nothing to do with the ‘people-side’ of the enterprise – which is actually what defines the enterprise.

It’s about just one aspect of the How of the enterprise. It barely touches What, Where or When. It’s rare that it takes any notice of Who. And often it doesn’t even glance at Why, at purpose: in an all too literal sense, it often has no purpose other than… well, itself, for its own sake, really.

Yet the IT-folks still persist, loudly, in purporting that this small subset of the whole is ‘the whole’, ‘the architecture of the enterprise’ – which it patently isn’t. But because they insist on calling it ‘enterprise-architecture’, and because they’re so insistent about doing so, it becomes next to impossible to describe to anyone else what enterprise-architecture actually is, or the value that a real enterprise-architecture brings to the business context.

In effect, the IT-folks are actively sabotaging our work – and then seem surprised when we get upset about it…

Yes, it’s true that slowly – very, very slowly – we do seem to be getting somewhere in this. After something like half a decade of constant hammering, the Open Group are finally starting to grasp that ‘business-architecture’ is a bit more than just ‘anything not-IT that might affect IT’. A lot more people are politely challenging John Zachman’s beloved metaphor of ‘engineering the enterprise’ – because ‘engineering’ is not a good a metaphor to use in any human context. And a lot more folks are coming round to the recognition that architecture is less about ‘functional’ requirements, the How, and more about the ‘non-functional’ qualitative requirements that connect more to Why; less about the ‘boxes’, and more about the ‘lines’ that connect those boxes, the ways in which everything links together.

Yet it often feels like Sisyphus: inching a huge boulder up a steep hill, only to have all that work knocked down again by yet another IT-obsessed [impolite-pejorative-deleted] who clearly doesn’t have much of a clue about what it is we’re actually aiming to do…

Yesterday was one of those days.

(I hasten to add that what follows is just one example amongst way too many: it’s not about any one person, but many if not most of the players in an entire sub-industry. We get this kind of obtuse circularity from all of them – and that’s why it’s so darn frustrating…)

In this case the starting-point was a Tweet from Roger Sessions:

@RSessions: Does EA have a mathematical basis? The Mathematics of IT Simplification, my newest at http://bit.ly/f4pAOk

A bit of context here. Roger is unquestionably one of the grand-masters of IT-architectures; he’s highly respected in his own part of that field, and with good reason. His specialty is the use of set-theory to reduce the ‘complicatedness’ of large IT-systems – what he used to call ‘simple iterative partitions‘ and, from his new paper above, now calls ’synergistic partitions’. I don’t do that kind of maths or that kind of IT, so I’ll say straight away that I’m not competent to judge it; but people whose opinion I trust in that field tell me that it does work well for that specific purpose.

The problem comes when it’s purported to be a ’solution’ for use outside of that one purpose. For example, Roger describes it as reducing ‘complexity’. But in terms of the Kurtz/Cynefin categorisation it’s all in the realm of Order, not Unorder; it’s highly complicated, but only Complicated – it doesn’t touch the true Complex domain at all. More worryingly, using this kind of approach has a high risk of creating true complexity rather than reducing it. By splitting a tangled yet fully interdependent whole-system into independent sub-systems, partitioning can easily push the overall context into the Unordered realm, creating a system-of-systems with truly complex interactions that are, by definition, beyond the control of any Order-based IT-system – and making an already difficult problem much harder to resolve. Not helpful.

And there’s no shortage whatsoever of evidence – such as the sad history of Business Process Re-engineering, to take one infamous example – that attempting to apply ‘mathematical’ approaches to the human contexts of whole-enterprise architectures really doesn’t work. But the problem is that that kind of model appeals so strongly to the so-desirable delusion of ‘control’: so people still keep on pushing it as ‘the solution’ to everything – even though we know it doesn’t work in that context. In fact it’s guaranteed to fail, expensively – in every sense of ‘expensive’. And when it fails, it falls to the real whole-enterprise-architect to try to sort out the resultant mess – with the IT-obsessed still providing their ‘helpful’ hindrance every inch of the way.

Which is why, when I saw that Tweet from Roger Sessions, the red-mist came down… What he’s describing is not enterprise-architecture – in fact in many ways it’s the exact antithesis of enterprise-architecture. And all of us need to get very clear on that fact, and push back, hard, against anyone who misleads others by describing IT-architecture as ‘EA’.

Anyway, there was quite a sizeable Twitter-chat that followed from that. Minus a couple of Tweets where I finally gave up in sheer frustration, it’s all there after the break (with my occasional add-on comments in italics as usual, and also some of the duplicate ‘@’-handles removed). See for yourself, perhaps?


  • JohnPolgreen: RT @RSessions: Does EA have a mathematical basis? The Mathematics of IT Simplification, my newest at http://bit.ly/f4pAOk >useful for IT, but is not EA… a real sense of “oh no not again…”, this is detail-layer #itarch (‘EITA’), with not a single hint of #entarch (‘EA’) anywhere in sight… wrong model, wrong metaphor, darn near wrong-everything when applied to real-EA…
  • RSessions: @tetradian Depends on your definition of EA.
  • tetradian: @RSessions you say ‘enterprise IT-architecture’ is identical to ‘the architecture of the enterprise’? – quote “this does not compute”… // …hence _please_ stop indulging in your misleading ‘term-hijack’ of EA, and use the proper term ‘enterprise IT-architecture’?
  • RSessions: @tetradian No, I say complexity control is the task of enterprise architecture. // The projection of business architecture onto IT architecture is enterprise architecture.
  • krismeukens: MT @RSessions: projection of business architecture onto IT architecture is #entarch < disagree, way too narrow
  • aojensen: @RSessions But Enterprise Architecture does not necessarily comprise IT Architecture.
  • krismeukens: MT @aojensen: @RSessions #entarch does not necessarily comprise IT Architecture < true though very unlikely in present times
  • RSessions: @aojensen I see IT arch as quite distinct from #entarch.
  • tetradian: @RSessions yes, ‘complexity control’ is _a_ task of #entarch, but why do you insist that it’s IT-only??? – sorry, but that’s daft… // “projection of #bizarch onto IT is #entarch”: agreed, that’s _one_ aspect of EA, but _not_ the whole – drop the ‘term-hijack’!!
  • RSessions: @tetradian We shouldn’t take on more than we can handle.
  • tetradian: @RSessions “we shouldn’t take on more than we can handle”: yes, _you_ handle IT-complexity well – and others _do_ work on other parts of EA // “I see IT arch is quite distinct from #entarch”: maybe, but still constraining ‘EA’ to IT only – enterprises are more than IT!! // simple ‘hard-systems theory’ math may work for IT-complexity, but it does _not_ work in human-complexity aspects of #entarch
  • RSessions: @tetradian Actually in the paper I don’t use the term #entarch because it is so poorly defined.
  • tetradian: @RSessions “Actually in the paper I don’t use the term #entarch”: so why use it in your Tweets??? – it’s not EA, so _don’t mislead_!!! // that’s all we’re asking from you: don’t mislead!! – if you yourself say it’s not EA, _don’t use that term_ for the work you do!!
  • RSessions: @tetradian I haven’t seen others offer a compelling value proposition for #entarch .
  • tetradian: @RSessions given that you’re one of the people most actively sabotaging real-EA by misuse of the EA term, that’s hardly surprising, is it?
  • RSessions: @tetradian In the paper I discuss the limits of the mathematics. // I didn’t say isn’t EA. I said EA is poorly defined. // Thank you, and I don’t think I am sabotaging. I am just defining what I think is a critical problem that only #entarch can solve.
  • [I gave up at this point…]
  • itraor: Intrigued by tweets between @RSessions and @tetradian, I just read the paper. Good. I wish it said ‘deterministic’ => ‘converges’ though
  • adrianrcampbell: Real #entarch is NOT the same as enterprise IT Architecture – they should not be confused
  • atmanes: Zachman claims EA = architecture of the enterprise. Are you requesting we use ‘enterprise IT-architecture’ instead? @tetradian @RSessions // EA looks at IT from “the good of the whole” perspective rather than “the good of the few” // All IT projects should reflect business architecture/strategy, with or w/out EA  // But without EA, optimizing one system may deoptimize other aspects of the enterprise
  • RSessions: @atmanes I see Enterprise IT as the overall IT architecture. // Isn’t that true for all parts of the business? // #entarch should be solving those problems that lie in the interstitial space between IT and Business. // That’s why we have CxOs.
  • atmanes: CxOs rarely get involved in individual project design. They rely on EAs to ensure “good of the many” influence. @RSessions @tetradian
  • ArchiTool: RT @adrianrcampbell Real #entarch is NOT the same as enterprise IT Architecture – they should not be confused <– Indeedy!
  • javier_castanon: @atmanes joins @tetradian and @RSessions discussing E[IT]A. Food for thought #entarch #bizarch #entITarch (last tag to be used instead?)
  • RSessions: @atmanes reducing complexity is good for the many. Not saying #entarch can’t do more, but that alone would be real value.
  • RSessions: @javier_castanon The problem us that none of these terms have well agreed upon definitions.
  • tetradian: @RSessions “none of these terms have well agreed upon defns” – please see Len Fehskens on Open Group website: http://bit.ly/fMD31C #entarch
  • itraor: @RSessions @javier_castanon An EA approach solely expressed in financial terms might get the CFO’s attention. // Likewise, EA approach devoid of terms unfamiliar to biz should often get attention // In my opinion, many EA practices make CxO’s feel like passengers, part of the problem.
  • RSessions: @tetradian We don’t lack opinions. We just lack agreements.
  • javier_castanon: @RSessions Agreed, it took years to IEEE 1471 to agree *just* on their current architecture definition
  • RSessions: @itraor @javier_castanon I agree. And I think complexity reduction can be expressed in financial terms. // I think trying to agree on EA is unlikely and probably not a good use of energy.
  • javier_castanon: @itraor @RSessions Couple of ideas: Are CxOs the true enterprise architects? Do they care about fancy IT metaphors?
  • RSessions: @javier_castanon @itraor If we go with @atmanes definition of EA, then I would say yes.
  • itraor: @RSessions Your paper gives me hope that we can move beyond semantics and really get to results
  • RSessions: @itraor I agree. The process has become the deliverable. Not good.
  • itraor: @RSessions Variants of your approach, each time with new/relevant complexity dimension would be useful
  • RSessions: @itraor Thank you! I really appreciate your affirmation!
  • itraor: @RSessions Carefully selected subsets of context variables define complexity, in a systemic way
  • RSessions: @itraor Interesting thought.  I have found a journey, not a destination. Others are welcome to join.
  • itraor: @RSessions Modelling EA along such subsets makes it possible to engage each stakeholder adequately // Absolutely! You’ve found a journey, I’d say too.
  • RSessions: @itraor I hope you will expand on these interesting ideas. Perhaps a blog?
  • itraor: @RSessions Indeed, it’s probably bigger than one blog post, e.g. to avoid the ’silver bullet’ trap
  • RSessions: @itraor This territory is rich for exploration. I’m delighted to have fellow explorers!
  • itraor: @RSessions I feel honoured. Someone with your clout is a gift to this discipline.
  • RSessions: @itraor I have strong opinions. Clout is still a work in progress. But thanks!
  • chrisonea: RT @RSessions: reducing complexity is good for the many. Not saying #entarch can’t do more, but that alone would be real value. < And the good of the many outweighs the good of the few (or the one). #spock
  • RSessions: @chrisonea The good of the many IS the good of the few. They just don’t know it.
  • itraor: @RSessions Your strong opinions help. I thought sets were of good use in EA, you’ve articulated it well // @javier_castanon It shouldn’t matter who the true entreprise architects are, it matters who participates
  • chrisonea: RT @RSessions: The good of the many IS the good of the few. They just don’t know it. < That’s a bit too universalist or one-size-fits-all for me. But I get your point.
  • pelujan: RT @RSessions I didn’t say isn’t EA. I said EA is poorly defined.¦ So are a lot of industry terms
  • RSessions: @pelujan Sad, but true.
  • taotwit: @RSessions @atmanes Given CxOs aren’t known for critical thinking WRT systems interactions if EA doesn’t help here who does?
  • markjappleby: good @tetradian tweets today challenging misuse of term Enterprise Architecture (it is not EITA – link to http://bit.ly/fMD31C is spot on)
  • RSessions: @markjappleby I admire his passion.
  • pbmobi: @markjappleby @tetradian #entarch is misusing the term architecture, real architecture is about breaking rules, uniqueness & creativity
  • MartinHowitt: @pbmobi disagree. Real architecture cannot break the laws of physics :) #entarch
  • tetradian: @pbmobi agree w your view of architecture – sounds like my ‘business-anarchist’ theme :-)
  • pbmobi: RT @tetradian @pbmobi agree w your view of architecture – sounds like my ‘business-anarchist’ theme :-) =>that’s me: business-anarchist :-)
  • pbmobi: RT @MartinHowitt  disagree. Real architecture cannot break the laws of physics :) #entarch => http://bit.ly/hN9Qf0
  • b_kratz: interesting discussion between @tetradian and @rsessions about EA today.Twitter imho may not be best platform for such discussions.add blog?
  • RSessions: @b_kratz @tetradian I think you are right, Twitter doesn’t work well for this. Blog now open for business: http://bit.ly/elBKfg

Do read Roger’s blog-post just above, and especially the comments-interaction between him and Chris Bird – the all-too-evident misinterpretation by Roger of what Chris has explained illustrates yet again the ways in which IT-architects (such as Roger) completely miss the point that whole-enterprise architects (such as Chris) are trying to make.

And I do emphasise that it’s not fair to single Roger out in this – as I say, he’s rightly one of the leading lights in his profession. It’s the whole darn IT-architecture profession that is our problem here. :-(

Oh well…


I knew it wasn’t just me that gets upset about this, but it was good to see the following comments from Nigel Green a few minutes after I posted the above:

  • taotwit: @tetradian I have found this so frustrating that I have given up using EA to explain what I do/how I earn a living hence I use #bcdesign [business-change design] // perhaps #bcdesign is a ’style’ of EA that applies more to big change rather than BAU [BusinessAsUsual]

And, to make the point, he followed up with this: a real-world example of the dangers of ‘EA’ not covering Enterprise Systems. Useful indeed.