Systems Functional Needs

As well as dallying about at conceptual / philosophical levels with cybernetics and systems thinking in recent decades, I’ve had several abortive attempts at getting some useful tools going. In fact in those decades I was also working with and for software developers / solution providers on “generic” tools mainly in the “capital facilities” business.

Could never get the right levels of abstraction taken seriously in any products, pragmatists are always very focussed on (apparent) immediate business needs … but now I may have the right systems thinking language …

And now, my needs have doubled – I have real selfish need of tools to develop and describe / document the models I’m developing, as well as toolsets I propose for generic wider value to humanity. So …

30 years ago before ontological (graphical) languages were fashionable I was a big fan of IDEF0 – and I still am. I know now the reason for its attraction are twofold.

One – it is process focussed, yes it represents things, but all the links are flows of information and stuff between functional objects. (My whole world ontology is a information process / computation metaphysics these days.) Not surprising  quite a few flowcharting tools – yes even Visio – include IDEF0 templates, but what NONE of them have is …

Two – it naturally nests levels of abstraction as sub-systems non-exclusively within super-systems – but NONE of the tools supports this functionality.

Related, but quite independently, diagramming languages have evolved for the information models and ontologies themselves. Express-G and UML – the latter dominating for a long while – but ultimately these are just networking graphics where rules determine what semantics are carried by nodes and edges, either or both. Unfortunately as people have been driven towards standardisation (a good thing), that semantic choice has been baked-in but not necessarily in the best way. So …

The new (ISO) standard for network graphics is ArchiMate and there are already many “Archi” tools, if not many good working examples. The Architectural root of Archi was very encouraging, but the results are disappointing. Unfortunately in the new standard all the semantics are in the edges – generally a good thing – but it means the sub-super-system abstraction (a feature of the nodes) is lost. As with previous ontologies, all the management structures have to be imposed by a meta-data overlay (aka eg “Templates” in ISO15926-p7) representing the “emergent” sub-super collections – but in the real world – not that of the modeller – these are the data that “matters”. We need to re-introduce the IDEF0 functionality to the ArchiMate language.

As well as generic tools for general use (with libraries of standard objects) I now have personal need for documenting the (process ontology) models I am now proposing and describing in my own writing.

Latest candidate tool in a systems thinking / active inference context is cadCAD – complex adaptive system diagrams (with dynamic “simulation” behaviour) … so, hopeful. And that “Active Inference / Markov Blanket” thought gives us a language for what is missing – if all semantics reside in the edges, the nodes have no thingness. See also footnote to previous post. It’s perfectly OK to capture semantics in relations involving the thing, but let’s not lose the thing itself.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.