At the CETIS 2009 conference on Wednesday we built a consensus model on the floor — or at least, made a lot of progress towards one — connected to competence. Not many people turned up in the end — we had more booked onto the session than came — but that was more than compensated for by the quality of those that were there. As promised, we talked a great deal, but not as a whole group, and used the traffic cones, string, paper, pens and staplers. I transcribed the “maze” onto CmapTools, and have posted it on the session wiki page, so no need to reproduce it here. We did indeed make as much progress in three hours as some groups seem to need years for, though it was clearly not finished.

There was a lot of very interesting discussion going on. As I had suspected, this generalised well from my own experience, recounted in a previous post, that one-to-one discussion is much better for helping one’s own model to develop than is discussion in a larger group. So, in the meeting, I discouraged conversations that were either across the floor, or threatened to involve the whole group, even though, as it happened, we probably could have got a long way with these as well, because of the small numbers involved. But it was more important to trial the method properly, so we can be more confident that it will work when scaled up to, say 20 or so people working simultaneously.

Personally, I will take a close look at the output, put together with my recollections of all the great conversations I had over one or other traffic cone, and apply them to the extension of my developing conceptual model. If anyone else wants to take the CmapTools file, and elaborate it with their own ideas, I’d be very interested to see the result. During the conference session, I had very consciously held back from putting in ideas from my own conceptual models, at least in the first half, so that what developed was independent of that. I did participate in the second half, working on what had been built up, and trying to play a normal role of a collaborative participant. Now, I’d like to repeat the exercise again, with this, or in related domains.

I’d keep the traffic cones as they are. Mark Stubbs commented that they were a good size. They aren’t the biggest ones you can get, but they are proper normal full-sized traffic cones. The string, paper and fibre-tip colouring pens were workable — not very neat, but adequate, and perhaps a little untidiness helps to keep the informal atmosphere that in turn helps people relax and discuss deeply and openly.

But I’m driven to incorporate the ideas from the top-level ontology I’m developing, as mentioned in the most recent post here. Perhaps we could decide which top-level categories are most helpful for collaborative conceptual modelling, and pop different coloured sports cones on top of each traffic cone, depending on the type of thing it is representing. In the models in earlier posts, I have used a four-way distinction:

• material things, including agents and non-agents together
• real instantiated processes that are proceeding, or have completed
• repeatable patterns of all kinds
• expressions, including assertions of fact and predictions

It might be helpful to distinguish agents from non-agents, and assertions from predictions. It’s not clear what will be most helpful in the practical situation of this “floor-based conceptual modelling”. There need to be enough categories to help people recognise what they and others are meaning, but not so many that they are themselves difficult to understand, or confusing. I’ll be thinking about it, talking with people, and waiting for relevant comments from readers here…