Systems Thinking requires Doing

For a while I have been thinking about the ‘problem of landing’ - namely when the ‘Thinking’ part of Systems Thinking has to stop and some concrete changes are to be made (notwithstanding ‘watchful waiting’ which is a positive decision not the intervene) . (If the work does not result in decisions on direct action we are studying Philosophy or some other ology!).

While much of Systems Thinking and Complexity is concerned with ideas and concepts the maxim for the practitioner is “I am only interested in getting to know because I want to do!“. In much the same way that defining the purpose of a system is necessary to define the boundary and so engage the perspectives so settling on the practical ‘call to action’ is necessary to determine the learning objectives. A pragmatic stance which matches just because rather than just in case.

Physicality

My interests started with TRIZ which as a method is diligent in maintaining the links (back) to the physical world whereas some of the more Philosophically inclined approaches seem to be concerned only with dealing in abstractions often unconnected with a physical structure in which the envisaged behaviours are manifest. See for example How is the gap between the ontological view and the epistemic view traversed or search for physicality.

The Challenge Forward - Physicality and Consciousness

It is interesting that the AI community are now paying interest to this same challenge, as an explanation of ‘can AI be salient’ there is a key paper from DeepMind - ✨The Abstraction Fallacy - and a rebutal ✨Mastering the Abstraction Fallacy. The disappointment is that the new era of interest makes no reference to the foundational work and substantive thinking on exactly this topic. Like so many of the developments in Computing each new era of technology seems condemned to re-learn the painful lessons of the previous generation - STPrism itself is designed to implement a rudimentary version management and traceability for an AI environment (see the paper on Context Anchoring and the others in the series Patterns for Reducing Friction in AI-Assisted Development

The essence of ‘The Abstraction Fallacy’ is whether the associations and patterns found in information are ‘tethered’ to some physical instantiation,, if so information becomes more than a description of what has happened to a prescription of what might. If not, then the LLM is elegantly, but simply, producing an echo of the underlying data, and the argument is correlation is not causation. This is the same argument made by David L Abel in Developmental Biology (for choice contingency) itself derived from Pattee’s concept of the Cybernetic Cut.

I might add that the LLM by adopting a conversational idiom is perfectly suited to induce a form of ‘User Illusion’ - recall Feynman’s dictum “The first principle is that you must not fool yourself and you are the easiest person to fool”. I have written on the positive aspects of this User Illusion in ReMemex and the nagative aspects in The Guru Mill.

The Challenge Backwards - Consciousness and Information

Whilst on the subject to might be worth looking at the material from “I and Me” aka ontological/epistemic aka transduction aka the Cybernetic Cut aka the Origin of Consciousness in the Breakdown of the Bicameral Mind. This is not a quest for the ‘grand universal theory of everything’ but rtaher a principle centred was of ‘mastering the muddle’ of different framing and different language of what - could be - the same phenomenon. There is a long discussion on transduction but if you jump into the middle for the discussion on universalism - Luhmann And Relational Biology