SSM Conceptual Models
Dearest Systems Thinkers,
I can only apologise for the blog becoming quiet. There have been a combination of reasons. We’ve all become much busier at work these past months while our organisations go through a “turnaround” process, and our attendance at sessions has been a bit more sporadic. I’ve personally become even busier out of work too and have become involved in a few more projects that have diverted me. We’ve stuck with it though and have been plugging away at Soft Systems Methodology (SSM) and other things, but at a slower pace.
Return to the Root Definition
If you recall, the last time I posted on SSM, we had defined our Root Definition and extracted the key “Transformations” from it, with the intention of then building our top level conceptual model from them. I had some doubts about whether this was going to produce a coherent model. I was right.
A way to coherence
When we came to build a conceptual model with these Transformations, it didn’t fit together. There were not obvious links or relationships between each of the bubbles/transformations. It wasn’t coherent. This was a powerful insight, as it made us realise SSM and conceptual modelling in particular provides a way to check our thinking is coherent. We decided to dive straight in a build a model without referring to the Root Definition. Here, after a number of iterations is what we came up with.
Harder than expected
We found this pretty difficult to do. Because this type of modelling and thinking about things was new to us, because it’s quite difficult, and we also found that our 1 hour lunchtime sessions weren’t really long enough for us to get our teeth into it and build momentum. It always take us a little while to warm up and it seemed that as soon as we managed to begin to make progress, it was time to pack up and end the session.
Take your time
We eventually overcame this by deciding to get together for an extended session of 4 hours one Saturday morning. It was great. The extra time enabled us to really get into model building, develop our own technique and make proper progress. We got stuck into the work and redrafted and agreed our top level model, and then also completed 2 sub-level models. So, if you are doing this yourself then I’d recommend you set aside plenty of time and space for it. Here is a picture of some of us at work.
Keep it flexible
I mentioned we developed our own technique for doing this. That probably makes it sound grander than it is. We found that as we built models, we changed our minds about what the nodes should be included, how they should be titled and how they related (connected) to one another. That meant trying to do this with pen and paper as we tried at first ended up producing a mess of crossed out and redrawn stuff.
The method we found worked well, was to start out by brainstorming the nodes we thought might be appropriate by using a post-it note for each one and writing it’s name on it. A not dissimilar process to the first step in the creation of an Affinity Diagram approach we tried out previously. The target is to have between 5 and 9 (7 +/- 2) nodes per model, but we would create as many as we could and then start to try to put them together the model and discard those that didn’t belong and reword any that needed it in order to fit coherently. We would connect them with arrow shaped post-it notes and reorganise until we were happy with what we had. At that point, we would commit pen to paper and set out the model. Here’s a picture of us doing this.
During that Saturday morning session we completed the top level model and then 2 sub-level models. We found that after that, we were able to build 1 sub-level model in a 1 hour lunchtime session. Once we had figured out what we were doing, it was a lot of fun building these models. There was always a lot of debate as we went and we gained a deeper and broader understanding of the “problem space” we were looking at.
So, here are our models. I’ve drawn them up in Insight Maker and shown each individually. It would be nice to be able to embed each of the sub-level models into it’s owner node in the top level model and then have it display when you click on that top level node. Does anyone know a free piece of software that will do that?
Peeling the layers of an onion
We decided the next steps would be to then pick a few nodes in the sub-level models and create underlying models for them. You can keep on going down layers and getting into finer and finer levels of detail. The question is how far is it helpful to go, and that is a judgement call.
Returning to the Root (again)
We also want go back to the top level model and compare it to our root definition and see if the worked together and whether we needed to modify the root definition now we had learned more about the problem space the top level “system”. And then of course complete the remaining SSM steps, which include comparing our idealised conceptual models against the reality of the canteen set-up and identified where changes might be made.
How to handle management???
Another thing we weren’t quite sure how to handle is the general management and then monitoring functions. The various tasks and people involved in the canteen will need to be managed. It wasn’t clear to us how to integrate that task/function into a conceptual model. It needs to sit in all of them at all levels one way or another, but that would also seem wrong. We figured that perhaps the management and also the monitoring of performance is somehow an external function that wraps around the whole thing. We’re not confident that is right though and this will need more research.
We’ll be back!
We’ve put that on hold for the moment though and have been following a few other paths. I’ll try to write up more about that, when I can. There have been some interesting developments with the canteen though and our work is not done there. I promise to write more.
Enjoy your weekend!