On SCAN, PDCA, OODA and the acronym-soup

What’s the relationship between sense-making and decision-making? And how does SCAN help with this?

This is a kind of ‘double follow-on’ to a comment by Stephen Law on sense-making and decision-making; and another query by Cynthia Kurtz on the structure of SCAN.

The structure of SCAN

I’ll address Cynthia’s question first. She writes:

I love it that time can stand in for simplicity, which is realistic. … The only thing that bothers me now is the confusion of the word “simple” being in two places. It’s like saying cat sometimes means cat and sometimes means grey cat. Confusing. Which is what you are trying to get away from right?

Okay, it’s clear I still have some work to do on a better description of both of those axes, but in essence it all boils down to just two questions:

• How the heck do we make sense of this, so as to decide what to do?
• How much time do we have, before we have to decide what to do?

The horizontal axis in SCAN deals with the first question, “how do we make sense of this?“. It’s actually about the type of logic – or reasoning, rather – that we use in ‘making sense’: a simple true/false categorisation of ‘is-a’ versus ‘is-not-a’ over on the left, and a more fluid modal-logic of possibility and necessity over on the right. In principle it should be a continuous spectrum, but in practice there’s that crucial Inverse-Einstein test: if we do the same thing, do we get the same results, or different results? Hence, is it simple, or not? The notion of Simple versus Not-known is actually a recursion on itself, applying a Simple true/false partitioning to that spectrum of possibility.

Another variant of this is digital versus analogue. In most cases, reality is actually analogue; digital is an abstraction of analogue, an arbitrary partitioning of analogue according to some chosen schema. So whilst analogue can (almost) always replicate digital, a digital representation is at best a simplified simulation of analogue: the simulation can be close, especially if done at very high speed, but it’s not actually the same. Yet because digital is a simplification, it’s much easier to do, and to replicate; whereas exact replication of analogue can be almost infinitely complex. Trade-offs, again: always trade-offs…

The vertical axis in SCAN deals with the second question “how much time do we have before deciding?”, by applying that same kind of spectrum-of-possibility to ‘time-available-to-decide’. The more time we have, the more possibilities we’ll seem to have, and the more possibilities we can assess – though note that, again, that horizontal axis still applies to how we can ‘make sense’. But as time runs short, we get fewer and fewer possibilities, fewer and fewer choices, until eventually, we have just this choice, right here, right now: and at that point, we must decide.

[There’s always a choice before action, a literal ‘response-ability’: remember that ‘deciding to do nothing’ is still a decision… 🙂 ]

There are two classic approaches to that point of action: follow the rules, or follow the principles, some kind of ‘guiding star’. (This gives us CP Snow’s classic ‘The Two Cultures‘, the seemingly-separate worlds of the sciences and the arts: in reality the boundaries between them are always somewhat blurred, especially these days, yet in many ways that useful distinction still stands.)

For enterprise-architecture, though, the really important point is that whilst real-people can take either approach – usually according to natural preference and the like – most machines and IT-systems can only follow the rules. There are several crucial corollaries to this:

• some people prefer the clarity and certainty of rules, others prefer the fluidity and flexibility of guidelines – which means they probably have different views about Simple versus Complex
• the same applies to types of work – some require certainty, other require flexibility, and most require an often varying mix of each
• in principle, people can always replace machines, but machines can’t always replace people
• in almost every context where there is inherent uncertainty, we are going to need to back up the machines with people who have the appropriate skills to take over where the machines’ rules no longer make sense
• because rules are an abstraction, the rules may need to be changed whenever the context changes, so as to still ‘make sense’ in practice
• it takes time to reassess those rules – so if we have no time to rethink, or if the changes happen faster than we can rethink, we may not be able to use rules, and hence not be able to use machines for that part of the task

In enterprise-architecture, look around at the business-rules space, the security-space, the process-management space, the ‘enterprise 2.0’ space, the service-management space, the knowledge-management space, and so many others: you’ll see example after example after example of contexts where those corollaries have been ignored, almost invariably with expensive consequences…

In other words, this matters, folks…

A couple of other points about that vertical-axis. One is that it’s much more of a true spectrum, without any obvious point at which to place a boundary: in that sense, those boundaries between Simple and Complicated, and between Ambiguous and None-of-the-above, are actually only arbitrary choices, which is why I’ve shown them as dotted-lines on the SCAN core-graphic.

The other is that in many ways there’s a direct line or spectrum of repeatability, all the way from Simple – which expects everything to repeat exactly – through Complicated, and then Ambiguous, and thence to None-of-the-above – which expects nothing to repeat at all. Yet there’s also the very real impact of ‘time-available-to-decide’; and also that definite point of the Inverse-Einstein test which applies as ‘the boundary’ between Complicated – which may be complicated, but ultimately unambiguous – and Ambiguous – which may not be all that complicated, but is always somewhat ambiguous.

So in a sense we might be better to draw the SCAN core-graphic like this:

It’s just more compact to do it the way shown at the top of this post, that’s all. The four-square layout also makes it easier to translate across from a Cynefin-style layout for the SCCC-categorisation: we need to do a left-right mirroring, but that’s about it.

[A reminder again, though, that this is not Cynefin: it’s a different framework, different underlying structure, different purpose. More detail in this post, if you’re interested.]

So to answer Cynthia’s question, it actually is correct to have ‘Simple’ in both places, because that region is Simple in terms of both sense-making – simple true/false logic – and in its decision-making – no time for anything but the simplest decision. (If you like, it’s kinda ‘Simple-squared’… 🙂 ) ‘Complicated’ is simple in its sense-making logic, but the time-available allows for more complicated reasoning; ‘Ambiguous’ allows modal-logic and more time; whilst ‘None-of-the-above’ is requires non-simple sense-making yet allows no time to think about it – hence radically different from Simple.

I don’t see anything particularly precious about those ‘domain’-names – in fact that’s the whole point, that different people have different meanings for these ‘domains’, which in their own way are all equally valid. In some ways they’re not even domains, they’re just ways to describe particular aspects of the sense-making / decision-making space – the context-space, in the terms of context-space mapping.

So if you want to use different terms for any of this, do: it’s up to you, really. 🙂 Just be aware that changing things around may make it more difficult to explain it to others, but that’s about the only restriction.

SCAN and the sense-making / decision-making loop

Let’s turn to Stephen’s question, about how sense-making is put to practical use:

I was just wondering how this might help with decision making (after sense-making). For example when doing EA work and confronted with a new situation and the nature of it. Is it something simple, that can just go ahead, is it complex and I need a plan, is it actionable but I need some form of governance, should I refactor the problem in an attempt to make it more simple? … if the framework helps shape thinking then it does not need to be completely exact. A new creative thinking tool?

The essential point to get here is that, on its own, sense-making has very little value: it’s of literally ‘academic interest’, but not much more than that. It’s only when we couple it with decision-making, and action, that it starts being genuinely useful.

It’s also not much use as a one-off: sense-making, then decision-making, then action. In almost all real-world practice, we need it to form a continuous loop: after the action, we loop back to sense-making, to start the same sequence again.

The two classic versions of this are the PDCA loop ‘Plan, Do, Check, Act’, and the OODA loop ‘Observe, Orient, Decide, Act’.

PDCA arose from the quality-management disciplines, primarily in manufacturing production. In SCAN terms, it assumes a relatively low cycle-speed or, to put it the other round, expects and allows a fair amount of time-before-decision; it’s about tactics and operations, but not real-time decisions as such – sense-making, decision-making and action all tend to have distinct separation in time. The Plan phase is, in effect, the task-oriented side of the decision-making part of the cycle; then Do, the action-phase. After the action – perhaps assisted by an After Action Review or the like – there needs to be a reassessment of what happened in the action, and what can be learnt from it to do differently next time: in other words, a sense-making stage, which starts in the Check phase and carries on through into the Act phase – about decisions and actions to change the capability – and thence back to Plan for the next task.

[Note that ‘Act’ in PDCA is not about action to do a task, but action to change the capability, ‘the ‘ability to do work’ – as distinct from setting up the next task of work, in the following Plan phase.]

OODA arose from the US military, specifically for air-combat, but has since been adapted for many other purposes. In SCAN terms, it can be applied at any cycle-speed, right the way through from relatively-long-cycle strategic assessment to real-time decisions in combat at literally supersonic speed. Observe and Orient are respectively about the inputs to sense-making, and the sense-making itself; Decide is, obviously, decision-making; and Act, equally obviously, is the moment of action. We then loop back round to Observe the results of that action.

[That’s a very simple overview, of course: there’s a lot more depth in practice, around each of those phases. For example, the Orient phase, even at combat speed, is expected to include ‘depth’-themes such as culture and training of the opponent; also we need to build an awareness of the opponent’s OODA-loop as much as our own.]

What SCAN usefully adds to the party is high-speed use of systems-thinking and the like within the sense-making phase – and to some extent the decision-making phase as well – in these types of continuous-loop processes. Remember those seven principles outlined in that earlier post:

• rotation – such as switching between views or between the items in a checklist
• reciprocation – balance across a system
• resonance – feedback-loops in a system
• recursion – self-similar patterns at every scale
• reflexion – the whole contained within the part
• mismatch – cognitive-dissonance to break ‘out of the box’
• serendipity – allow ideas to emerge ‘at random’

That checklist itself is a Simple-style rotation, as is the visual pattern of the SCAN ‘domains’ – which means that they’d be available to us at very high speed. Each of those items is very simple in itself: it doesn’t take long to memorise what they each mean in practice, and how to scan quickly through their implications in each domain of the context – fast.

Hence, as per Stephen’s examples above, flicking rapidly through different options, pushing anything ‘Not-known’ into that labelled box so that we can pick it out again for review later when we do have time. That’s the idea here: ‘business-speed‘ sense-making, to support ‘business-speed’ decision-making.

But it’s also not just about the Simple side of that real-time balance: the other side is about working with the real-time chaos – rather than somehow trying to pretend that it doesn’t exist. For example, this quote from the Wikipedia article on OODA puts it well:

the proper mindset is to let go a little, to allow some of the chaos to become part of his mental system, and to use it to his advantage by simply creating more chaos and confusion for the opponent. He funnels the inevitable chaos of the battlefield in the direction of the enemy.

An improv actor, or a salesperson on the front-desk, would do much the same, though preferably without viewing the audience or client as ‘the enemy’!

I’ll describe this in more detail in later posts, but I’d probably best stop here for now. Hope it gives you something to start with, anyway?