Context-space mapping and enterprise-architecture

(This series of posts explores a concept of ‘problem-space’ versus ‘solution-space’ which in part demonstrates alternative uses and interpretations of the Simple / Complicated / Complex / Chaotic categorisation originally described in the Cynefin diagram. It must be emphasised that this is not about the Cynefin Framework; for details on Cynefin, please contact Cognitive Edge.)

This post represents yet another attempt to describe certain fundamental differences in approach from twf (aka ‘That Welsh Framework‘ – so-called because we’re no longer allowed to use its official name at all) and to find an alternative term that might reduce the ongoing friction in that quarter.

To do this, we need to go right back to first-principles: the core concept of context-space, which eventually leads us to context-space mapping.

(Another long-ish post: more after the ‘Read more…’ link.)

Before any notion of order or unorder, or even of disorder, there is simply ‘the everything’: everything and nothing, all one, everything and nothing connected to everything and nothing else, a place-that-is-no-place that incorporates within itself every possibility. It’s not ‘chaos’ – it simply is.

There are all manner of names for this ‘active no-thing-ness’: Lao Tse called it ‘the Tao’, for example, whilst the ancient Greeks described it as ‘the Void’. For the more business-oriented purpose of enterprise-architects, though, we’ll need to constrain the scope of this ‘the everything’ somewhat, and we’ll also need a more ‘business-like’ label. So let’s call it context-space – the holographic, bounded-yet-unbounded space that contains every possibility within the chosen context.


In previous posts I’ve split this context-space into problem-space – the context in which things happen – and solution-space – the space in which we decide what to do in relation to what’s happening. But ultimately there’s just the context: “the only true model of a system is the system itself”.

Yet to make sense of anything, we need to impose some kind of structure. One place to start would be to filter ‘the everything’ in terms of its variability. Perceived-repeatability is one example of a variability that we might use (which we’ll come back to in a moment), but there are of course many others.

Context-space - variability

Initially this gives us a finely-graded spectrum of variability. Yet interestingly, most human sensory-perception is not very good with smooth gradations: it works much better with firmer boundaries. Hence most sensemaking will usually attempt to place some kind of ordered structure upon what may initially seem like unbounded chaos.

Context-space - bounded variability

When we look at the physical world, of matter and material, we can see both of these processes in action, even within matter itself. There is a fairly smooth gradation of variability, primarily linked to temperature; yet there are also explicit ‘phase-boundaries’ where the internal relationship of matter undergo fundamental changes. Significant amounts of energy (‘latent heat’) can be absorbed or released in the ‘phase-transitions’ between these modes. In effect, these present as four distinct states of matter, traditionally described as Earth, Water, Air and Fire, for which the more scientific terms are respectively Solid, Liquid, Gas and Plasma.

Context-space - common domains

Looking at the internal structures of matter within each of these states, we would typically describe the respective structural relationships as Simple, Complicated, Complex and Chaotic, as phases or domains within the context-space of matter. This categorisation along a single axis represents a simple first-order map of that context-space – hence context-space mapping.

Much the same applies to just about any other view into the overall context-space. If we take almost any type of gradation, we will be able to identify distinct phase-boundaries that can be used to partition the context-space into distinct regions along that axis: the nominal split of the visible-light spectrum into Red, Orange, Yellow, Green, Blue, Indigo and Violet is one such example. But perhaps the most useful split of all for enterprise-architecture and business-architecture is along an axis of repeatability, dividing the inherent uncertainty of context-space into regions that we could describe, in parallel with those states of matter, as Simple, Complicated, Complex and Chaotic.

Which brings us, unfortunately, into the same conceptual space as twf (That Welsh Framework) – though we’ve arrived there via what is, in very literal sense, a fundamentally-different route. And unlike twf, we can now see:

  • how and why we’ve arrived at those particular categorisations
  • how and why to use any specific axis for such categorisation
  • what the boundaries between the ‘domains’ in the categorisation will look like
  • how, why and when the nominally-Simple boundaries between categories may move (Complicated), blur (Complex) or fragment (Chaotic).

This provides a layered, recursive richness that is largely absent in twf. It also provides a means to link right across every possible view into context-space, rather than solely a specific set of interventions that focus primarily on a set of views into the Complex domain.

A first-order (single-axis) context-space map – such as the Simple-to-Chaotic ‘stack’ – is not all that much use in practice. To make it more useful, we’ll need to add other axes as filters for sensemaking, to enable relevant information to fall out of the respective comparison. And we make it more useful again by selecting a related set of axes to provide a multi-dimensional base-map upon which other filters can be placed. (Two-dimensional base-maps are the easiest to work with, for obvious reasons, but three or more dimensions are entirely feasible – the tetradian is one example of a four-dimension frame compressed into three-dimensions for use as a base-map.) To do this, we choose axes which force the domains of the original single-axis spectrum into relationships of opposition and similarity with each other. For example, if we use ‘levels of abstraction’ as the core axis, and overlay that with timescale in one direction and a ‘value-versus-truth’ spectrum in the other, we arrive at the following base-map and its ‘cross-map’ of interpretive text-overlays:

Time, interpretation and abstraction

Here Chaotic and Simple are in opposition over their means of interpretation, but similar in terms of timescale; Chaotic and Complex are similar in their means of interpretation, but opposites in terms of timescale; Simple and Complex, and Complicated and Chaotic, oppose each other on both axes; yet all domains are related in terms of layers of abstraction. The central region (‘reality’) is essentially a reminder that the domains represent related yet arbitrary views into what is actually the total ‘hologram’ of context-space – everything else is actually an abstraction from the real.

We then layer this recursively to apply to the nominal boundaries between each of the domains, so that these too may be considered to be fixed, movable, porous or fragmented or transient. An axis based on a simple binary true/false categorisation (in other words, a Simple boundary) will split the the context-space into two domains along that axis; if both overlay-axes have relatively-Simple categorisations (or movable two-part categorisations, in Complicated style), the overall context-space is split into four regions – which aligns well with the ‘matter’-type categorisation of Simple, Complicated, Complex and Chaotic. Likewise a smooth gradation along both axes pushes the context-space into four regions with Complex or even Chaotic boundaries between them.

Because of this,  a four-region base-map is likely to be the most common and most useful two-dimensional type – hence, we may note, the twf is often shown paired with two-axis overlays. But other layouts are possible and sometimes useful: for example, a pair of tri-value  axes would typically be used to align an eight- or nine-domain primary axis, such as seven-colour plus infra-red and ultra-violet.

The result is a consistent structure for base-maps that are simultaneously bounded and not-bounded, and that describe the whole of a context-space by structured views into that context-space that also acknowledge that the context-space ultimately has no actual structure. Hence the importance and validity of the assertion that even though twf is often shown paired with two-axis overlays, it is not solely a two-axis matrix. The other point, though, is that this indicates that twf is merely one instantiation (or set of instantiations, rather) of a generic class of context-space mappings that has been around and in general use for a lot longer than twf itself.

Hence to avoid further clashes with twf, I suggest that in future we use the generic term context-space mappings to denote base-maps and derivatives that use this type of structure.

Once we’ve cleared that particular road-block, we should be free to concentrate more on practical applications of context-space mapping for whole-of-enterprise architecture, but I’ll leave it there for now. As usual, any constructive comments, ideas and suggestions would be most welcome 🙂 – over to you, if you would?

Previous posts in this series:


11 Comments on “Context-space mapping and enterprise-architecture

  1. I need time to absorb this which I won’t have in the next few days. In a quick speed-read I did pick up one ‘nit-picking’ observation. I try and avoid characterising twf as a matrix because it has curvy boundaries and the disorder domain in the middle.

  2. No rush – thanks a’much for commenting at all.

    Will take your point about the ‘nit-picking’, but it is actually dealt with above. I do explicitly state there that twf is definitely not _in itself_ a two-axis matrix – exact quote is “even though twf is often shown paired with two-axis overlays, it is _not_ solely a two-axis matrix”. I guess you missed that in the speed-read – my apologies that I write too long… 🙁

  3. Err. well..I did read that sentence :). Had I written it, I would have deleted the word ‘soley’ and replaced it with ‘actually’.

    When mapping things onto to Cynefin Domains I try wherever possible to use an ‘official’ Cognitive Edge graphic that shows all 5 domains, even if I don’t refer to the 5th one explicitly. (There is also a wiggly bit at the bottom of the vertical axis , but I’m not sure of the significance of that)

    As I understand it, the fact that the boundaries are curvy is an important part of the ‘sense-making’ nature of the model. It’s similar to the idea that all the diagrams in Peter Checkland’s books on Soft Systems Methodology are hand-drawn to illustrate the points that they are not models of reality but mental constructs to be used to explore the nature of reality.

    So I don’t think it’s a matrix at all. Hope this clarifies where I’m coming from.

  4. I’m trying to be as careful as possible here, because I don’t want any further trouble from from That Quarter.

    “As I understand it, the fact that the boundaries are curvy is an important part of the ’sense-making’ nature of the model.”

    Yes, I’d agree, but we need first to understand the nature of boundaries, and thence the nature of a model’s domains. Look at the point above about the nature of boundaries: they too can be Simple (static and non-porous), Complicated (non-porous but movable), Complex (porous – i.e. spectrum) or Chaotic (fragmented and/or quantum states – simultaneously present and not-present).

    If we then apply that to a four/five sector context-space map such as ‘twf’, we end up with a range of boundary-options:
    1. Simple two-axis matrix
    2. Complicated two-axis matrix with movable regions
    3. Complex ‘domains’ with a sort-of two-axis matrix
    4. Chaotic structure in which domains simultaneously exist and do not exist

    Trying to use type-1 boundaries for a four/five-region model clearly won’t work: in that sense, it’s true, ‘twf’ is not a Simple two-axis matrix. In fact, the presentation usually shown for ‘twf’ is somewhere between type-2 and type-3: it’s a two-axis matrix with type-3 blurred boundaries pushing outward from a central region where a ‘none of the above’ condition applies (i.e. ‘disorder’), onto which type-2 movable non-porous boundaries have been placed. Hence the ‘curvy’ boundaries in ‘twf’: it’s the best way to depict a type-2/3 boundary.

    To ask a very simple question, how much use is actually made of ‘twf’s central ‘disorder’ region, by most people who claim to be using ‘twf’? Short answer is ‘not much’: which in practice means that they’re mainly using ‘twf’ as a type-2 two-axis matrix, or even a type-1. (The whole concept of a ‘boundary’ between domains is actually either type-1 or type-2 – there is no identifiable point of transition ‘between’ in type-3, and no real boundary at all in type-4.)

    The basic layout and structure of ‘twf’ actually is a two-axis matrix, around ‘order’ vs ‘unorder’ (horizontal) and a slightly-odd usage of ‘layers of abstraction’ (vertical), without any explicit depiction of the unifying continuous-variability that links all of its domains together, using type-2/3 boundaries to mark out the domains. (The alternate assertion would be that the layout is entirely arbitrary, which I don’t think is the case.) Almost all of the overlays I’ve seen for ‘twf’, including those provided with the ‘official’ model, are simple two-axis matrices, or derivations from the cross-maps between the overlays and the base-model. (Some of the overlay-sets also reference the undocumented unifying-axis that links all four domains in an arc, as shown in the diagram above: one example of this is the mapping of cause/effect relationships in each domain.) The fact that it uses type-2/3 boundaries makes it possible to use it as more than solely a Simple two-axis matrix, but it really does not go much further than that.

    For example, ‘twf’s set of ‘official’ documented cross-maps is really quite limited; there’s very little cross-mapping between different cross-maps; there’s no apparent concept of recursive nesting; there’s only a limited set of pathways (e.g. consisting solely of ‘take control’ or ‘collapse’, between Chaotic and Simple); and there’s no means to link to any other base-map than the ‘official’ four/five-domain frame. It tackles one specific requirement very well indeed – understanding how to move from other domains into the Complex, and how to operate within that domain – but despite a specific person’s protestations it does _not_ tackle anything else very well at all. The documented techniques in ‘twf’ all emphasise how to derive information about the collective, in order to act on a collective space: they do _not_ address uniqueness other than as a source of point-information to be merged into a Complex-domain collective. (The web-seminar indicated some work on uniqueness of outliers, but again still primarily or exclusively within the context of the collective.) Context-space mapping (aka ‘tinc’), by contrast, needs to be able to cover movements and actions between and within _all_ domains – and that explicitly _does_ include working with inherent uniqueness in the Chaotic domain.

    In essence ‘twf’ represents just one specific categorisation using one fairly simple form of context-space mapping – and there are many, many others. It’s a very useful categorisation, in a very useful layout for a base-map: but the notion that anyone would think that they possess proprietorial ‘rights’ over one such specific depiction in the midst of a vast general class of depictions to me makes no sense at all.

    “(There is also a wiggly bit at the bottom of the vertical axis , but I’m not sure of the significance of that)”

    The tiny blob shown on lower-centre in the Wikipedia graphic for ‘twf’ is, I believe, an artefact of placing four apparently-arbitrary ‘methods’ onto the context-space. To me it’s certainly not a mistake – though some people apparently believe that it is – but is a correct representation of a possible minor outcome of the categorisation/boundary process: conceptually speaking, it’s actually another ‘none-of-the-above’ region (i.e. part of ‘disorder’), cut off from its central ‘parent’ by the sweep of the type-2 boundary imposed on the type-3 spectrum of variability.

    “The diagrams in Peter Checkland’s books on Soft Systems Methodology are hand-drawn to illustrate the points that they are not models of reality but mental constructs to be used to explore the nature of reality.”

    My point exactly. A model is only a model, a tool for sensemaking. The moment we try to build it up as being anything more than that, we’re in real trouble. That’s why it worries me a great deal that we’ve had such bizarre arguments about what is or is not the ‘proper’ depiction or usage of ‘twf’ – despite the claims of science and suchlike, it’s not ‘the truth’ or anything like that, it’s just a useful made-up model.

  5. First: I like this post very much as it explains and clarifies the ‘structural concepts’ behind TINC, even in terms that ‘my mother’ would be able to understand. Great post!
    With Sally I sense (feel) something ‘lacking’ with regards to the important difference between TINC and the constant awareness that ‘a model is only a model’, as Tom feels is evident, as I do (duh), but yet … so I gave this some thorough thought and came up with the following: what the models (TWF and TINC) ‘do’, whether by straight lines or curvy lines, is provoking the ‘conclusion’ that a typical ‘either – or’ is assumed. A problem is either in the ‘complicated’ or in the ‘complex’; it is by its form an invitation to position ‘it’ … somewhere. This happens at a very early stage for that is what the models invite to. And yet, this is not what the philosophy behind TINC wants us to ‘fall into; rather: to the contrary.
    So I borrow from medicine (again) and would like to introduce the concept of ‘differential diagnosis’ and superimpose this on the model-as-a-tool. As in mental health care, the diagnostics of even the highest level (normal, aberration, disorder, disease) is not a thing to be established before… we can start treatment. In fact the treatment and the diagnosis happen together iteratively for a very long time. Meanwhile normal medicine uses the very same approach in cases where say child abuse may (or may not) be the cause of something as simple as a bruise or a broken arm. The fundamental assumption is that, if I transpose this to TINC, a ‘problem’ as well as a ‘solution’ is supposed to be in every quadrant of the whole map (simple plus complicated plus complex plus chaotic) and can/will only be dis-guarded from any ‘spot’ when elimination (the opposite of determination and conventional diagnostics) allows for it.
    What I am saying is (I guess): using the TINC effectively requires differential diagnostics ( for both the problem- and the solution-space. This is the only way to put the vision behind the fuzzy lines in the quadrant to real practice. And it will imho also allow for the conclusion that we can have quantum-problems that reside in more then one quadrant, which will allow for interventions with regards to the problem scope even. Now, I am not sure if I ahve been able to put my thoughts on this as well in to understandable words as the post that I am responding to, so I will leave it for now and wait for either red or white flags to be raised 🙂

  6. Paul – “I will leave it for now and wait for either red or white flags to be raised” – actually, I was thinking more of a green flag… (unless my differential-diagnosis of your ‘red or white flags’ is even more mistaken than usual… 🙂 ) – in other words, great comments, and good challenges.

    “what the models (TWF and TINC) ‘do’, whether by straight lines or curvy lines, is provoking the ‘conclusion’ that a typical ‘either – or’ is assumed.”

    Yes, that’s the risk – in other words, using the terms in my last reply to Sally above, a ‘type-1’ or ‘type-2’ boundary is assumed (a non-porous ‘either/or’), when it could equally be a type-3 or type-4 ‘both/and’. Part of the sensemaking process consists of identifying what kind of boundaries are involved, and at what level. 🙂

    (As you’re likewise discovering, this is extraordinarily hard to express in words or even in diagrams, even though it’s actually quite easy to visualise… 🙁 )

    The other point is that we don’t really situate at any single point in the model, or even at multiple places in the model. (By the way, most real-world contexts – perhaps all? – will include _some_ components that are ‘in’ every segment of a twf-like model.) The point is that this is _dynamic_, not static – about pathways, not positions (and multi-layered pathways at that – kind of like a giant n-dimensional version of a child’s snakes-and-ladders board-game). Individual techniques may be situated at specific places or regions (depending on the model-overlays in use), but the actual process of working on/with the ‘problem’ moves _through_ those ‘places’ – i.e. _uses_ the technique, and then moves on to something else. And yes, it’s often iterative – in fact always will be for any wicked-problem or the like, because it can only ever be ‘re-solved’ rather than solved.

    If you think in terms of pathways, we’re always moving towards the complex or complicated as soon as we have any time to think about what we’re doing – thinking-time is a luxury that we don’t have in the true chaotic, which is why we practice so hard with simple responses so as to be able to implement algorithmic (complicated) or pattern-based (complex) real-time interactions. But it’s a pathway, not a place: eventually we _do_ have to return to the real-world, along the very edge of the chaotic/simple axis.

    I strongly agree with your point about differential-diagnosis and the like, though note that technically that’s just another Complicated-domain technique (literally analytic, and assuming a single ‘right answer’). Other options – as described in the Wikipedia article – include gestalt (complex) or intuition (nominally Chaotic, but in practice more likely a muddled mixture of Chaotic and overly-Simple – i.e. what Snowden describes as ‘premature patten-entrainment’). The real point is that ultimately each patient/client/whatever has _some_ unique characteristics: the danger with conventional differential-diagnosis is that it can lead very quickly to what I call ‘the Meaning Mistake’, where missing a single step can result in finer and finer analysis of the wrong decision-pathway (see the ‘Disciplines of Dowsing’ book). What I’m looking for here are ways or techniques or whatever that can warn us that we’ve done so, and work our way back up to the missed branch. (The British Medical Journal had a great article a couple of weeks back on synaesthesia, warning that it’s relatively harmless and is much more common than most people realise – about 1% of the population, for all forms combined – and should not be confused with more serious problems such as schizophrenia.)

    What I’m saying, I guess, is that TINC or whatever is not a static ‘this is it’ model: it’s more like a map for a toolkit for creating maps for navigating in all surprisingly-common yet crazily-confusing conceptual/contextual spaces.

    Dunno if that makes it any more clear, or have I just muddied that waters even more? 🙁 🙂 I suspect that this is one of those topics that’s very hard to sort out via this medium, but might be much easier in person with a large recording-whiteboard and an unlimited supply of caffeine? 🙂

    In any case, thanks again – this kind of constructive challenge really does help.

  7. Thanks Tom for your elaboration. And yes, it is indeed difficult to describe 3D in terms and language of 2D (which is as yet all we have within the ‘logical’ realm of non-art and the from-A-to-B discourse 🙂 ).
    Yet I will attempt to put some more coal on that fire… A way out of ‘the Meaning Mistake’ is, indeed, essential. It is our developed (not natural) assumption that ‘steps’ are to be taken, a more-or-less linear process is inherently part of the road ‘from problem to solution’. It is this assumed logic that forces us back into the 2D world even when we focus on the 3D and even 4D realms.
    What-if the methodology behind TINC assumes, or even prescribes, a deliberate (rational) counter-approach..? I am thinking about one very good example of how this counter-rational behavior of a ‘therapist’, with the hallmarks of dowsing at the core (intuitive), gives us an example of how this should/could ‘work’: Frank Farelli. In his world (psychiatry) he introduced, with love and very intuitively, the counter-acting (to the established ratio) and with enormous impact. I do not know if you know of him, and of his work that has become known as ‘Proactive Therepy’ (, but it has all the hallmarks of a process that imho would be highly usable and effective in the field of TINC too: iterative, high involvement of the ‘subject’, not only in the diagnosis, but also in the solutions, and also in the iteration between problem and solution… Even more importantly this would very much assume that the ‘subject’ is totally involved in the whole ‘process’ of getting from un-awareness to awareness of the here-and-now… And maybe, just maybe, that value-now-place is the closest thing to reality we can get-and-work-with, that is as soon as an identified subject is placed (back) in the true context: chaos.
    To the more practical: what if the interaction between subject and interventionist is ‘ruled’ by the same rules of engagement as provocative therapy? Would not, then, the subjects reveal its true ‘nature’ with far better accuracy then any conventional client-consultant/architect interaction? And would the very unpredictability of such an approach nor be closest to chaos, or even unorder, that we possibly could get?

  8. Sorry for the typo’s in my last post. Also ‘Proactive Therepy’ should of course read “Provocative Therapy” – ‘Proactive’ was a Jungian typo 🙂

  9. Hi Paul – Yes, Provocative Therapy looks interesting, though I can’t comment any further until I’ve at least had a chance to look at the website.

    I’m being slightly careful about this at the moment, and say that ‘tinc’ (for which I’m suggesting the slightly more permanent name of ‘context-space mapping’) is primarily about _mapping_ the context and process of intervention, rather than specifying any particular mode of intervention. (Yes, I acknowledge Snowden’s dictum that “any diagnostic is also an intervention”, etc, but the point is about what I’m emphasising here – i.e. sensemaking rather than intervention as such.)

    Interestingly, if we take a strict ‘chaotic’ view (i.e. a type-4 quantum-like boundary), then there both is and is not a boundary between subject and object, observer and observed, ‘problem’ and ‘solution’ – it’s a _simultaneous_ both/and, never _actually_ an either/or.

    A strict ‘chaotic’ view would also align well with Deming’s dictum that the place most likely to have the knowledge needed to ‘solve’ the ‘problem’ is at the point of action.

    Better stop there for tonight – well past time for bed, will look at it again in the morning. But thanks again, anyway.

  10. Let me know when you’ve written this using a model less flawed than Cynefin — the continuum was never Chaos-Simple, it’s Chaos-Order with the paradox represented by Complexity which relies on the balance where the ‘optimal simple’ reins.

  11. Hi Paula

    I’ll admit I don’t quite see the difference, in that to me they’re the same spectrum: Simple is Order with tightly-constrained assumptions, and Complicated just a bit further round the spectrum towards Complexity.

    But it actually doesn’t matter: if you don’t like Cynefin, don’t use it. 🙂 The whole point about what I’ve termed ‘context-space mapping’ is that you can use whatever base-map you want, that to you fits the need for sensemaking in relation to the ‘real-world’. More on that in the next article in this series, that I’ve posted about half an hour ago. (I originally thought you were replying to that one, not this.)

    Many thanks for joining in with this discussion, anyway.

Leave a Reply

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