Dave Snowden

Options in uncertainty

RSS Feed

In my last couple of posts I have been talking abound boundary conditions in Cynefin. I now want to extend that a into understanding operations in the complex domain. I’m not sure about this picture and I doubt it will survive a few presentations and discussions but it is a starting point. I’ve taken the complex domain and extended the boundary state, hopefully with a degree of fuzziness or ambiguity. This is the first time I have put together different capabilities within Cognitive Edge so bear with me here.

The long established approach within Cynefin for complex issues has been to seek out all coherent hypotheses and then construct a series of parallel safe-to-fail experiments, one of each hypothesis. Those are managed as a portfolio with some basic rules: contradiction between experiments to ensure scanning, one that uses a naive asset and possibly the odd oblique experiment. In practice those experiments generally merge, mutate or fail and a solution or solutions emerge. Historically we have left it there in terms of teachable method. But I now want to move on, adding some options, including some different decision points and in effect extending the boundary conditions to both the chaotic and complicated domains.

The simplest way to determine that a situation is complex is to ask a simple question: Does the available evidence support competing hypotheses? If so then after testing for coherence safe to fail experiences take place in parallel. These are show as yellow hexagons on the picture below, with one in blue stripes to remind readers of the need for at least one of which should use a naive approach.

Now I want to add a new test here, one were there is concern that there are insufficient coherent hypotheses, or more important insufficient contradiction. If that is the case then we should be concerned that we may have insufficient range of experiments and may be at risk of missing a key weak signal, or signals. If that is the case then we now have an additional option. We create a basic infographic to describe the situation, possibly using alternative timelines, counter factual or pure fiction. The MassSense capability of SenseMaker® can then be used to get a diverse range of human sensors to interpret that infographic and to add their own narratives. From that we can draw a fitness landscape which shows the diversity of views.

I’ve show one of these to the right and it shows three clusters or perspectives, within which there are some outlier events. If we have set up a sensor network in advance (what I have termed networks for ordinary purpose that can be activated for extraordinary need) then this can be done more or less instantly. These maps lead us into decision point two on the map. If we have lots of outliers or disparate patterns then we move into safe-to-fail, but if we have coherence we can skip that stage and move directly into the complex-complicated boundary zone.

More on this as I develop the idea, and hopefully some better illustrations. The ability to take this route does however require creating the network, ideally in advance. I’m increasingly arguing that effective management of the complex domain requires preparation of decision support capacity with no knowledge of actual use. Probably less effort than a conventional strategy approach, but different in nature.

Anyone interested in experimenting here; let me know. One of the drivers here is a (to use the jargon) a pivot we are working within Cognitive Edge on training programme. The goal is to scale that work and scaling requires codification. More on all of that in future posts.

The opening pictures by the way are from my recent walk on Carmarthen Fan, light on a mountain tarn; ambiguity and uncertainty but a degree of coherence within the text, then broadening out to a wider perspective in the banner

  • Marcus Guest

    Interested in experimenting.

    • Dave Snowden

      Lets talk next week in Geneva

  • http://www.Broadleaf.com.au Stephen Grey

    Do you have any plain English alternatives to “Does the available evidence support competing hypotheses?” to use with clients who don’t see the world in terms of hypotheses?

    • Michael Hill

      Explanations (for the available evidence and it’s meaning)? Tentative Theory or understanding (of/concerning the evidence)? Courses of Action (to exploit the evidence)? I’d think the alternative chosen would depend on the client and how well you understand both their language for viewing the world and Dave’s use of hypotheses.

      I enjoyed catching up on his blog yesterday, especially the parts on untranslatable Welsh words. It seems you may be in a similar quandary, although I’d think you could ‘bring them along’ if you understand them well enough. It’s not that difficult a word, somewhere between guess and established theory or accepted but not concretely proven understanding and definitely not a “Law.”

      • http://www.Broadleaf.com.au Stephen Grey

        There may not be an opportunity to ask a client to stop and contemplate the terms in which they understand their context, no matter how sensible we think it would be for them to do so. Very few will agree to take time out from a stressful situation to be introduced to Cynefin for instance.

        The ideas at the front of their minds are attached to language rooted in the physical and information systems they are trying to work with.

        I can explain to someone that an incremental exploratory approach is useful where stakeholders are loosely aligned but remain independent and will seek to promote their own objectives, which might change as they see work starting to take place. That is one line of questioning that can be useful – “tell me about your stakeholders and the extent to which they can exert an influence on the direction you take”, for instance.

        I expect that most people with an interest in this field use such strategies. I’d be interested to know if anyone has made them explicit.

        • Michael Hill

          My experience is different. I don’t mean that in a bad way, just to put my comments in context. My experience is often following failure of other outsiders who never learned the clients language. I will usually persevere until I find the analogy or example that works for them and then bring them toward the language of what I’m offering, unless the language doesn’t really matter in which case I’ll try to remember to use their term rather than mine. It likely depends on the role you are in: are you a troubleshooter brought in to help fix it; a trainer or instructor to impart skills or knowledge they need to fix it; a facilitator brought in so they can fix it themselves, etc. I’m usually in one of the first two roles, so I put the burden of learning their contextual language on me. Does limit me to English as my other language skills are survival basic at best, that may be a bug in the global economy but I haven’t corrected it (yet). If the client didn’t see the world in terms of hypotheses, I’d start by trying to learn how they do see the world and how they deal with their ‘facts’ that can be explained more than one way. If they are ‘one answer’ people and are locked on to a view based on their gauges and indicators, then it might be inauthentic disorder (to borrow from Dave), and they need to see that despite all indications of pressure from the pump and perhaps even flow, the shower has no hot water, so what are the possibilities between here and there? The complex domain should be a rich field for spinning a couple hypotheses to the client without even using the word hypotheses. From there you’ll probably hear their language: ‘alternative theories’, ‘competing models’, or ‘dangerous ideas’ and can use it in the future. Sorry, can’t help you with a list of explicit tactics – I’m sure someone’s selling a book somewhere, but the skill seems very tacit to me. Codified knowledge of it would probably only provide some little scaffolding for the apprentice or journeyman. Experience combined with feedback from the master or for the apprentice, the journeyman, would be the real power source of increasing knowledge flow of how to do this, more than a laundry list of “best practices.”

          • http://www.Broadleaf.com.au Stephen Grey


        • Michael Hill

          Apologies, formatting fell out of my answer.

  • https://twitter.com/chrizbot Chris Butler

    “I’m increasingly arguing that effective management of the complex domain requires preparation of decision support capacity with no knowledge of actual use.”

    There is still a purpose of the organization, isn’t there? It seems that all new endeavors will start in the complex domain when involving people that have some problem the organization wants to solve. That seems like knowledge of some use. Or am I misinterpreting?

    • Dave Snowden

      My point was creating a capability that could be utilised without knowing in advance what it would be needed for. So yes I think you are ….