Meaning Guide

Seeing the system

There’s something missing.

The newly released UK apprenticeship standard for Systems Thinking locates systems practice “in arenas where complex problems exist that cannot be addressed by any one organisation or person, but which require cross-boundary collaboration within and between organisations.” This is great. It neatly identifies why the language of systems is coming back into vogue: As the world becomes more complex, we’re waking up to the fact that problems can’t be simply pinned down to one person, one team, one organisation, one population.

But I want to look at this quotation more closely, because it highlights what for me is a big gap in the systems world. In particular, I want to pull out the two concepts of “complex problems” and “cross-boundary collaboration”. Firstly, let’s just pause to appreciate what a wonderful thing it is to be able to read these two phrases in the same sentence in a government-backed standard! So how do systems practitioners actually create “cross-boundary collaboration” to address “complex problems”? How does it actually work in practice? Well, one of the things a systems intervention will invariably involve is some form of collaborative modelling. I’m using modelling here in a very generic sense; even if nothing is written down, and the intervention simply amounts to a series of conversations across organisational boundaries, this will still have the effect of shaping the mental models of those involved in the conversation.

But that’s not most people’s experience of systems modelling. Across most of the major methodologies, the model will be co-constructed into an explicit, usually visual form by the participants. Here are a few examples:

Example causal map from SODA (Strategic options Development and Analysis) methodology
Example Causal loop diagram from System Dynamics methodology
v lifecycle model
Example conceptual model from Soft Systems Methodology
Example rich picture from Soft Systems Methodology
Another system dynamics model – perhaps the most vilified Powerpoint slide in history?

Now, even to the uninitiated, it’s obvious just from looking at these kinds of models that they are addressing complex problems. In fact, that’s often the point; their creators want them to appear as convoluted as possible in order to emphasise that there are no easy solutions. For most people though, the complexity is just off-putting – the most obvious thing the examples above have in common is that they all look a bit like spaghetti1.

And this is where I think the systems world is missing a trick, because to my mind, once you want to tackle a complex problem with cross-boundary collaboration, and you try to articulate the complexity through a visual model of some kind, then surely you have to address the issue of legibility.

Étienne Wenger, in his ideas on communities of practice, popularised the concept of boundary objects – the artefacts that span the gap of meaning between different communities. When you are involved in a system intervention, you are typically trying to get a representative of as many different parts of the system as possible (the communities involved in or affected by the complex problem) into the room at the same time in order to co-create the model. So for the duration of the session, the model is a boundary object, a focus of shared meaning for those taking part. And in my experience, that’s exactly what happens. When you start with a blank sheet of paper and slowly build up a map of the system (obviously assuming skilled facilitation that’s keeping everyone attentive and involved), the model becomes a record of the shared insights that emerge. The group listens to perspectives they haven’t heard before, overlays them into a composite form, and watches as patterns emerge that no one group by themselves would have predicted. When run well it’s a hugely rewarding experience, and you end up with a map that is rich in shared meaning.

So what’s the problem? The problem is that the meaning stays in the room! The boundary object only holds the shared meaning for as long as the group that creates it stays in one place. It’s practically illegible to everyone else. Even if it’s laid out in a relatively clear form, with clear handwriting and not too many tangled lines, the simple appearance of complexity is enough to put off virtually everyone who wasn’t in the room from even attempting to navigate it. And even if they do, how well equipped are they to determine what the group meant by each of the words they left behind?

So there’s a gap. I think systems practice is missing a whole sub-discipline, which is how to make models of systems meaningful across the system. How to make models that cross organisational boundaries, where you don’t need to have been present at their creation to understand what they are saying, but which nevertheless spark systemic conversations because they don’t just reflect one group’s perspective of the whole. How to help the whole system see itself, if you like, not just the representatives who turn up to systems interventions. What sort of things might fill this gap?

I’ll not try to answer the whole question here, as the main point of this post is just to draw attention to the fact that the gap exists, and to be honest, I don’t think anyone is remotely close to understanding how to fill it. But just to give an indication of the kinds of things that might be involved, I would say we need a much greater appreciation of the extent to which humans naturally make sense of the world through stories, rather than systems. The two are not independent, though. The brain is effectively a pattern-spotting device, tuned to recognise and respond to different situations. A system is an inter-connected set of situations that recur in a predictable way. A story is a collection of these situations told in sequence. So here’s at least one rabbit hole to jump down: What if the next step you took with the kinds of models I showed above was not to turn them into Powerpoint slides, but to turn them into sets of visually interconnected stories? What if instead of just using the output to show everyone how very complex the situation is, you built up to the complexity by laying multiple narrative threads on top of one another, until the overall pattern became unmistakeable?

If this sounds like an interesting idea, you might want to read some more thoughts on it here.

  1. Some people, criticise these kinds of diagrams in the other direction i.e. that they are reductionistic, fail to adequately capture the emergent nature of the system’s behaviour, interpret the complex as it were merely complicated, imply that the practitioner is not part of the system, and so on. While there’s validity in these concerns, I’d repeat my earlier point about modelling: Even the forms of words people use to voice these criticisms of the model are themselves models. There’s no escaping the need to model – as ever, the question is whether or not the model is useful.


  • Hi Steve, interesting article and thoughts.
    For what it is worth, I am currently working with social care and community health – across two large public sector organisations One of the issues the team that I have faces is how to understand the interrelated elements of what we come across. We have taken 95 cases so far, worked with those people, and understood them in detail, and attempted to take them on a better health journey than they would have done in the current service.
    So, the issue we have is how to understand and communicate this complexity and interrelated connections.
    The answer that we have is in two parts.
    The first is that we dont, not exactly anyway. We record down the flow that the person takes on their path to health. The path they take is unique to them, and it is not particularly complicated. It is complex and we also want to express the system around them. We use a simply flow like picture for the flow, and a picture of the value created, and a picture of the persons interrelationships.
    The main learning here is that it cannot truly be documented, that knowledge resides in the brain of the main person who is dealing with them.
    The second answer is that the ‘system’ of the NHS and social care what is traditionally sought is simply a very very complicated picture of all the possible interrelationships that can possibly happen. But no simple person goes down that route, so why map it?
    The third thing, out of two, is that the team create stories to communicate what they are doing, taken from the outside-in perspective from the person, as to their journey, what matters to them, their interrelationships and how it fits together in a time line. If we can add in resource used, handoffs, people involved, then we do if the leaders are looking for links with their current business model.

  • As noted here, this is a separate problem from developing insights about a complex system and I don’t think it has received very much attention.

    It’s s statement of the obvious but worth saying that the reason we wind up with such large models and we can only create them by bringing together many disciplines and viewpoints is that they span too much for one person to understand easily. Having a big messy picture will not overcome that.

    Two strategies drawn from a different line of work, understanding uncertainty in very large capital projects (mines, mineral processing and infrastructure), are:
    1. Break out subsets of the overall picture to communicate to each interest group the things they want to know immediately, with the option for them to explore other areas if they wish;
    2. Work top down, condensing dozens of components into s smaller number of group items with high level relationships between them that are backed up by second level pictures of the components and relationships within a group, always having the entire network available for reference as people work their way into it.

  • Yes, the meaning stays in the room, hence one reason why SSM, was extended into SIS to enable the group to design sndxevaluate the options and also develop an implementation strategy. SIS system intervention strategy accomodated a range of modelling methods, eg, SODA, SD, systems mapping etc etc to accomodate the maturity and culture of the group but also the organisation.

  • So there are at least two problems here.

    The legibility problem points to the limits of visual sensemaking. Visual graphs help up to a certain limit of visual complexity and then they start interfering. Mathematics and software modelling tools are the best way to go beyond the limits of crude and static qualitative representations. Mind mapping tools such as XMind can sometimes create a bridge by helping the reader view the graph dynamically from different perspectives.

    The problem of meaning staying in the room is a universal problem of explanation. You can summarize expertise, but the summary can never replace the richness and depth of understanding that come from figuring it oneself and struggling with all the raw data and multiple perspectives along the way. Visual sensemaking graphs are even worse, because they pretend that the nodes and links are meaningful to readers when they aren’t necessarily. They really only work well for the people in the room who participated in making them.

  • Steve,

    You make some good and interesting points. Let me fill in some blanks and consider a challenge or two…

    First – the development of maps and their use for planning and decision making is unquestionably beneficial. They are the only way (so far) we can hope to capture the rich complexity of “how the world works.” To get the most out of our maps, we’ve found that the nodes/concepts should be measurable and that the connecting arrows should be causal. Otherwise, in creating a map of a muddle, we simply end up with a formalized muddle.

    Along those lines, we can also use a “structural” approach to show what is missing from our maps (and, so how useful each map might be for making decisions and reaching goals). For a simple example, consider a chain of “A causes B causes C causes D.” What is missing? One possible answer is “D causes A.” These kinds of “missing connections” can sue us to look for more knowledge (and more stakeholders) to make the map more complete before charging off in the wrong direction.

    Second, we are getting better at refining our maps. Yes, as you say, maps often end up looking rather messy. One way to address that issue is to have an “expert facing” map (where a team of experts, each perhaps focusing on one part of the map) for use in on the research side. Then, to have a simpler “client facing” map to support descriptions, conversations, and stories.

    More sophisticated techniques are emerging for reducing the complexity of the map without adversely effecting its usefulness. Those include:

    Wallis, S. E. (2019). Orthogonality: Developing a structural/perspectival approach for improving theoretical models. Systems Research and Behavioral Science, in press. doi:10.1002/sres.2634.

    Wallis, S. E. (2014). Abstraction and insight: Building better conceptual systems to support more effective social change. Foundations of Science, 19(4), 353-362. doi:10.1007/s10699-014-9359-x.

    And… as you note… mapping can be useful for interdisciplinary work. I would suggest that it can be more: Wallis, S. E. (2020). The missing piece of the integrative studies puzzle. Interdisciplinary Science Reviews, (in press).

    As far as stories go, however, I am not in complete agreement. No story can communicate as much as a map. And, stories must focus on few things to the exclusion of others. Also, stories have been in use for years… and have not given us much progress (stories may be used to support good as well as evil).

    On a side note, our book on “Practical Mapping” helps people learn about systems without using systems jargon (a research textbook that is 99% jargon free). It would be great to get more “pre-systems” books/movies/conversations/stories into the public sphere to help prepare minds for more complex systems notions.



    • Thanks Steve. The issue for me is not so much stories vs maps, but how the two could be better integrated.

      To me, stories correspond to how we gather information about the world, maps to how we store it. Take geographical maps as an anology: A city map isn’t very meaningful to you if you’ve never visited the city before – what you need is a guidebook with walking routes that take in the main sites. Each excursion you make using the guidebook is like a little story that unfolds over the course of the day, bringing the city to life in a way that the map doesn’t. After a few days of following the guidebook though, you transfer your attention to the map as the primary point of reference, because you understand where you want to go based on your lived experience of what you’re interested in, rather than deferring to the creator of the guidebook. The map becomes the container of meaning in a way that wasn’t possible without the series of stories you lived out to get you there.

      So as I see it, telling each recurring dynamic in a system as a story that people can relate to should build familiarity with the patterns that underlie all of the stories, in the same way that following a series of guidebook excursions around a city builds familiarity with the underlying geography. The analogy obviously breaks down for lots of reasons, but the important point is that there doesn’t seem to be any equivalent methodology for how to visually link the two together in the systems domain, so that the general audience who have not taken part in the systems intervention don’t just hear the stories arising, but see how they inter-relate on a map to generate more complex recurring patterns. The ideal outcome would be for the general audience to use the map to tell their own stories, to seek out aspects of reality that are under-examined and generate new possibilities, just as a tourist you might scan the map to seek fresh vantage points that are off the guided tour routes.

  • Steve, I fully agree with your amazing essay here. I am a advocate, trainer, consultant and practitioner or visual systems thinking and modeling as you describe, working with primary aged school kids up to government ministry level policy makers here in Asia. I think you hit the nail on the head in your assessment and potential solution to the “lost in translation” effect that systems mapping has once is it cleaned up and put into a visual diagram (even with some narrative explanation of the big picture and causal dynamics and behavior). I am working with my Chiang Mai community to try to create a system of indicators and information based on shared concerns and a shared vision of a more sustainable and great place to life in the future for our city. WE use the Compass of Sustainability framework to help simplify and organize our conversations and inputs, then will apply systems thinking to help tell a story of cause, effect and feedback, as well as intervention and change. I would love to have further conversations with you on this topic. I am very interested. I have shared your post on Linkedin and my FB far and wide. This is totally what is missing, or as you say… the missing piece. I AGREE.

    • For me the key to unlocking this is revelation…..if a study is done or a group have an experience they get it……They can tell the story…..that can be powerful…..I’ve done that a lot with CLDs I’ve created and then told the story and thrown the diagram away…..but more powerful is either recreating that experience or asking questions which enable the person to make their own revelation/discovery. Now that is really powerful……It can be via a “thought experiment” or a physical experiment…..set up in such a way that learning can be gleaned…..then it becomes exciting…..what will happen?…..There is wonder!! But there is no “quick fix”……

Subscribe to new content:


Subscribe to new content: