More keywords for SCAN

Some notes that came up for me almost a month back, on the SCAN framework for sensemaking and decision-making, and that I hadn’t gotten round to documenting until now.

I was reminded that I hadn’t posted these notes when I saw Mark Foden refer to the Cynefin framework in his talk at the Integrated EA conference, to illustrate the difference between a context that’s merely complicated, and one that’s genuinely complex – the latter often because it includes a variety of ‘people-issues’ and suchlike. The Cynefin framework is quite useful for illustrating that difference.

Yet over the past few years – as I’ve described in quite a few of my previous posts, such as ‘A human view of Simple, Complicated and Complex‘, and ‘Ensuring that the Simple stays simple‘ – I’ve become increasingly wary of the term ‘complex’: the problem is not in the term itself, but in the fact that people interpret it in so many different ways. For example, much of what IT-folks might call ‘complexity’ – because it’s, well, complex, isn’t it? – would be described instead by aficionados of complexity-science merely as high levels of ‘complicatedness’ – the key distinction being that it expects a predictable and repeatable result. Hence a lot of unnecessary confusion and, occasionally, heated argument, that really doesn’t help anyone at all. That’s one of the key reasons why, in SCAN, I switched over to using ‘Ambiguous’ instead: it’s perhaps not as precise a term as ‘complex’, but at least it’s unambiguous about the existence of the ambiguous! 🙂

Which brings me back to those notes, because I’d been building small collection of alternate keywords for each of the ‘domains’ in SCAN. None of these keywords replace anything: it’s just that, following the same principles of context-space mapping that underlie SCAN itself, they add extra layers of richness to the sensemaking by providing a kind of contrast, cross-reference, cross-check and suchlike. Anyway, here’s the list so far:

S‘ domain: Simple and Straightforward (high-repeatability, at or near moment of action):

  • Step-by-step
  • Speed [don’t feel, don’t think, just do]
  • Switch [as in predefined flowchart, ‘no thinking required’]

C‘ domain: Complicated but Controllable (high-repeatability, at distance from moment of action):

  • Correct
  • Calculable
  • Certain
  • Controllable
  • Confirmable
  • Conformity
  • Compliance
  • Certifiable

A‘ domain: Ambiguous but Actionable (low-repeatability, at distance from moment of action):

  • Adaptable
  • Adjustable
  • Amenable
  • Assessable
  • Ask [as in the uncertainty of experimentation, rather than the (apparent) certainty of calculation]
  • Angst-laden! 🙂

N‘ domain: Not-known, None-of-the-above (low-repeatability, at or near moment of action):

  • No idea
  • No time to think
  • Newness [as in (near)-unique, therefore no certain rules]

Any other suggestions that you’d add to this?

10 Comments on “More keywords for SCAN

  1. A = Adoption-led (As in LiT Adoption Engineering), Agile-practice, Answers-emerge
    C = Capability-constrained (Expertise required), Constrained (bounded), Calculated (formula applied), Conforming (to rules)
    N = Nascent, Non-conforming, Next-practice, Not yet, Non-compliant
    S = Sorting, Sifting, Stacking, Shelving (as in archive), Squirreling (filing)

    • Brilliant, Nigel – many thanks indeed!

      You’ve actually added a whole new dimension to the usage for SCAN. The first two usages were about sensemaking (the original version of SCAN, and why it’s called SCAN), and decision-making (the crucial distinction between ‘rational’ decisions-at-distance versus decisions in real-time). What you’ve added here is “what does this look like in real-world practice?” – and that’s really going to help a lot in building practical adoption.

      (In a way, not at all surprised that you’ve come up with this: you always have been much better than I am at framing things in terms of concrete business-terminology pragmatics – and it’s one area where I know I do need help. So again, many thanks!)

      On your distinctions between ‘S’ and ‘C’: yes, that’s it exactly. ‘Sorting, Sifting, Stacking’ etc all presume that there’s a predefined frame in which to put things and a predefined method for doing it; ‘Constrained’ etc expect a certain amount of ‘freedom to move’, but a ‘bounded-freedom’ – yet in some case not even knowing that the constraints are ultimately arbitrary.

      On your ‘A’, I really like ‘Answers-emerge’! – and you’re right about the others, too.

      On ‘N’, yes, ‘Nascent’ is a really good one, and ‘Next-practice’ fits well with the first stage (‘Idea’) of the ‘scientific-development’ sequence of ‘Idea, Hypothesis, Theory, Law’.

  2. Another build…

    I think the bottom axis would be better described as Repeatable >> Experimental/Experiential rather than Simple to Not..
    – given the lefthand cell is already using the ‘Simple’ tag and the upper lefthand is by definition *not* simple (requires levels of expertise in formulae etc.) but is highly repeatable (engine-like). And the right hand side takes us more towards discovery, change and the generally chaotic.

    Cheers, n.

    • You’re right again, of course – I should have fixed up those labels a long time ago, because it’s probably well over a year since I linked that axis to a spectrum of repeatability-to-uniqueness.

      That’s a straightforward fixup in the graphic, though: in this part at least, it’s those two labels that are wrong/inconsistent/out-of-date, rather than the structure of the framework itself.

  3. Another part 2…

    Is there something insightful in changing the vertical axis to ‘Operative >> Expert’ (bottom to top) and the horizontal to ‘Repeated >> Uncommon’?

    S quad = Highly operational and repeatable behavior (standard procedure)
    C quad. = High specialist expertise required and highly repeatable results (formulaic method)
    A quad = High sense-making expertise applied to uncommon behavior (adaptive method)
    N quad =. High operational impact (reality of the coal-face) but no manual to follow (operatives unguided decision)

    just riffing, thoughts? N.

    • A great riff, Nigel, and again many thanks – though this is one point where I do have to disagree with you.

      I’d strongly agree that “there’s something insightful in changing the vertical axis to ‘Operative >> Expert'”, because that is how most business-folks would see it. The trap is that it leads us straight back into Taylorism, where the purported-‘expert’ is one who is further and further away from the point of action.

      Remember that, like Cynefin, SCAN is a two-axis frame and not a two-axis frame, both at the same time – and I did choose those two axes with a lot of care! 🙂 The key point in this case is that if we overlay the skills-development sequence ‘Trainee / Apprentice / Journeyman / Master’ onto the frame, it actually forms a curve:

      • Trainee = real-time/repeatable/low-theory = Simple
      • Apprentice = repeatable/high-theory (but learns theory at distance from practice) = Complicated
      • Journeyman = some-uniqueness/high-theory (learns theory of uniqueness at distance from practice) = Ambiguous
      • Master = real-time/uniqueness/informed-by-theory = Not-known

      In other words, the highest ‘expertise’ is actually back down at the base of the vertical-axis – not at the top.

      If you reframe your choice of vertical-axis as ‘Practice >> Theory’, then yes, it does work: in fact I do use that framing sometimes. If we do that, then the key point about the ‘S-N’ pairing is that ‘S’ is practice-without-theory, and ‘N’ is practice-fully-informed-by-theory’. ‘S’ doesn’t know anything other than ‘the rules’, whereas ‘N’ needs to know the rules so well that it knows when and how and where and in what ways to bend or break them.

      Or, to put it another way, ‘N’ needs to know how to dance with uncertainty, a constructive ‘chaos’ – whereas ‘S’ tends to collapse into panic, the colloquial sense of ‘chaos’.

      That last note is important in another sense, because it’s actually the point where I found myself forced to part company with Cynefin. As I (mis?)understand it, Cynefin uses much the same vertical-axis as you suggest above, from Operative to Expert, and the key difference that it draws – which is, I emphasise, an important distinction – is between Complicated and Complex, the differences in tactics and methods that we need between ‘repeatable’ and ‘probabilistic/uncertain’ at the ‘Expert’/high-theory level. (Interestingly, this probably makes the framework more ‘saleable’ to managers and the like, because it doesn’t challenge Taylorist delusions about the purported centrality of the manager’s role. :wry-grin: ) Yet in exactly the same way that there’s a phase-shift in tactics between Complicated and Complex, so too there’s another phase-shift as we approach true-uniqueness, because not only do Simple or Complicated rules not work/make-sense there, but neither do Complex patterns: the unique is unique, no rules, no patterns, because there’s nothing else with which it can be compared. And although Chaotic is the correct term in a technical sense here (and a shift from Complexity-math to Chaos-math is helpful here, too), unfortunately it’s too easily poisoned by the colloquial sense of ‘chaos’, which gives us tactics that are often entirely the wrong ones to use here – the crucial distinction between a panicked “Don’t just stand there, do something!” (aka ‘Act / Sense / Respond’) versus a more Zen-like “Don’t just ‘do something’, stand there!”.

      (From one of your HK tweets that quoted Snowden as talking about ‘Control <-> Complex <-> Chaotic’, it seems that Cynefin may at last be beginning to move in that direction, but I hasten to add that it’s none of my business as to whether or not it does: I’m just concentrating on making SCAN more usable and useful. The only part where I would be concerned is if there’s an attempt to ‘sell’ Cynefin – or any other framework – as a tool for sensemaking and action in uniqueness if it can’t actually handle it, that’s all.)

      One last point, coming back to “is and is not a two-axis framework” etc, SCAN also is and not a categorisation-framework. (Snowden makes the same point about Cynefin, and I’d strongly agree with him there.) Yes, we sort-of-categorise, because yes, the ‘domains’ or ‘cells’ do indeed represent sort-of-categories. Yet first, and crucially, the ‘boundaries’ are actually quite fluid: there are critical and identifiable transition-points on both axes in SCAN, yet exactly where they are and where they take effect in different contexts and different people can vary a lot. And secondly, the sole point of categorisation is not to put things in nice, neat, predictable, safe, Simple boxes, but to identify appropriate tactics for use in different parts of a context. As I understand it, every context will include various elements that are Simple, Complicated, Ambiguous and Not-known/unique: the tactics we need to get the best out of that context are different for each, hence we need those different tactics, applied at the respective skill-levels, with the respective governance/review and suchlike, to the right elements in that context. Which makes it an architectural concern, as well as a pragmatic one. Which is where I started all of this in the first place, of course. 😐

      Hope that makes a bit more sense? And any further comment/questions/critique on this, if you would?

      Thanks again, anyway – much appreciated!

  4. Tom

    I have not really attempted to “come to grips” with SCAN yet, but it looks interesting and valuable.

    Reading through this post and subsequent comments, I could not help but think about what I have learnt from “Adaptive Leadership” by Heifetz et al in relation to technical problems vs adaptive problems. Seems to me that this relates quite well to the distinction between complicated and complex.

Leave a Reply

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