Evolution is the preservation of contingency

There is a difference between prediction and regulation. I suspect also there is a difference between anticipation and prediction. An anticipatory system - which all biological systems are - does not simply make bets on the future. Of course they make choices, but in the process they remain open to contingency. Predicting machines do not do this. They make choices and throw away contingency. 

Evolution is the name we give to the ongoing process of maintaining contingency while making choices. Taken as totality, evolution is a process of regulating homeostasis, at the level of organism, species, ecosystem, universe. Exactly how this is done we don't know. Our AI systems do not do it, that's for sure. The many different biological theories (for example, Friston Free Energy) are deficient because they too discard contingency. 

Music is a phenomenon (or at least the human relation with music is the phenomenon) that does preserve contingency. Studying its mechanisms for doing this can perhaps help reveal how it does it... But we are a very long way from understanding it. It is analysable in terms of information flows, as I've argued many times, and in those flows, no information is discarded - every element of contingency is necessary in order to construct the conditions for closure of music. It is a highly complex evolving system whose evolution occurs in front of us, unlike evolution itself. 

But we do not live in a musical world. If we did, we wouldn't have so many problems. We live in a world of language - and language does not have the property of music. As St Anselm said in his "De Caro Diablo":

We ought not to be 
held back by the way 
in which the improprieties 
of speech hide the truth, 
but should rather aspire 
to the precision of the 
truth which lies hidden 
under the multiplicity 
of ways of talking

Music gets closer to the truth. 

I'm currently working on a project which tries to get beyond the improprieties of speech. I don't want to dismiss LLMs - they move the bar in terms of the realisation of new technical explorations in ways which would previously have been impossible (or perhaps improbable). But it is possible to explore the deeper music of language.. not it's sound, but it's dynamics. 

We use words to refer to concepts without really having a clear idea of what a concept is. But we can say a few things about concepts which sheds light on these dynamics. Gordon Pask regarded concepts as stabilisations in recursive processes (there's recursion in the idea of explaining concepts!) - more technically they are eigenvectors produced in the process of organising a cognising system in an unknowable environment. Pask saw the perception of objects no differently - "reality" is stabilisations of recursive circularities. The process of eigenvector formation is difference and similarity which arises in the multi scale conversational dynamics between actors. Another way of putting it is that in these processes we detect states and transformations. 

We can't know directly what drives these dynamics except that the words that we use and the concepts they refer to leave a "trace" which  challenges a biological system to adapt. When we analyse conversation we are analysing the adaptation to all these traces. The interesting question is what happens between one trace and another. The in-between moments of traces sees a dynamics of dealing with information and contingency. This is the energy of what is going on. These dynamics are very similar to music. 

Our present AI can help us get off the starting blocks to analyse this. But it cannot really help us harness the energy of language because it throws away information - it is a prediction engine. This is probably why the LLMs are not particularly practical in real life - present obsessions with them are likely to fail. The increased efficiency and productivity of organisations depends of better regulation of their operations. That in turn depends on conserving contingency. Where we are at the moment is we have AIs which discard contingency by making predictions, and meanwhile idiotic managers, transfixed by the technology, will further discard organisational contingency by replacing humans with robots. It's not going to end well. 

We are going to need a new kind of technology which doesn't generate language, but somehow helps us to feel language and it's organisational context more deeply. It should help us to feel the energy of speech. It should tell us how and when meetings are pointless and dull (as if we need telling!), or identify the energy of creativity, or spot the telltale signs of psychopathy. 

This technology will be useful, and more importantly, it won't replace people, but in fact enhance the potential of effectively organised human systems. That is what maintaining contingency really looks like.

"The Old Oak" and utopian nonsense

Astrid and I watched Ken Loach's recent (2023) movie "The Old Oak" last night. Don't know how I missed it when it came out. I admire Loach's work deeply because it reflects the world we are in in a way which preserves the grittiness of daily life, but retains the possibility of magic which is the preserve of art. 

The conversations among reform voters, aggressive and angry people, young and old, the disappearance of hope, and the sheer logic of their grievance needs to be set out dramatically. This is not abstract. No warm empty words from privileged classes - politicians, educators, philosophers - can possibly get close to the rawness and realness of emotion. However, there is space for intelligent analysis of the context within which the drama plays out - the role of social media in whipping up hatred, the dilemmas of people who cannot escape hypocrisy, the sheer fear of refugees whose traumas in the UK are just the latest stage in a litany of trauma.

I have academic friends who talk of how things are getting better, how technology will bring us together, how the education system can be fixed. What nonsense. Get them to sit in the pub with these desperate people and try to explain that... "Technology will revolutionise society and make things better!" they might try to explain. "For whom?" Will come the most polite of the responses. 

"The problem is we have a false ontology!". Fuck off! - people have got no money! 

I want to think not of causes - it's too complex, and yes, ontology is part of it - but it's practically impossible to have a conversation about that even in universities! Much better to think about how to coordinate interventions. There is no one-size fits all solution, although there may be an ontological root. But when there are so many needs and demands, and so much diversity, and there are so many possibilities for addressing them, how can we select and coordinate interventions. Most tantalisingly, might we be in a better position to do this now than we once were, given the technology at our disposal?

That's going to require some ingenuity. And much of it will not work. But if we can coordinate interventions better then there may be ways we can address these deep issues more constructively, and in way that fits peoples' real lives much better than we currently do.

Creativity tomorrow

I find myself in a transition phase at the moment. I've spent the last few years trying to wake academics up to what was once the "impending wave of AI", and spending quite a lot of time frustrated by the fact that most academics thought it was irrelevant. Now everyone's a sodding expert! Well, that's academia for you...

So I think I can back away from this AI mania for a bit and let people get on with it. My university has bought 60000 licenses for Microsoft copilot. Personally, I would have done nothing, but worked hard with the staff to accept that our educational environment has changed and we must change - but buying an AI platform is not the way to do that! The locus of control of technology must sit with the individual student (and academic) - even if that now amounts the locus of control of the cost of technology.

Management anxiety has led to the institution to take this decision. They want to be seen to do something. Others will watch and see what happens... "For god's sake, don't just do something - stand there!" would have been more sensible advice.

Of course what the institution won't be able to do will be to keep up. I attended a fascinating "creatives" session at the RNCM the other day. There were film-makers who are making a name for themselves with AI video. They liken the present situation to the rise of cheap sequencers and production software in the 90s which meant that kids in their bedrooms could make music that only professionals could do a few years before. That's now happening with movies. No need for actors, props, lighting, sound, etc. Artistic input is still there but it coordinates activity by "worker bots" rather in the way conceptual artists produce art with assistants. 

That's here to stay. We're going somewhere different now. 

Most fascinating is the ability to generate analytical dashboards based on real-time discussions. I've got quite good at this. I was in a meeting with a public health academic this morning talking about cooperative housing. While he and another friend were talking, I made an app to distil the parameters of the conversation and demonstrate the interrelations between them. Quite amazing - and they were a bit gobsmacked. 

Of course AI is not making any money, there will be a crash, the energy, etc. But the energy problem will be solved (as many other issues will be) - its like early steam engines. 

So on the one hand I'm backing out of institutional AI. But I'm moving forwards with AI. The institution will get left behind. But that may need to happen...

The Reality of Charles Ives

Charles Ives famously joked once "Are my ears on wrong?" His music has always confirmed to me that if his ears were on wrong, so were mine. I was introduced to Ives by my dad who once led a production of Waiting for Godot, and had found the perfect music to accompany it - The Unanswered Question. Curiously a few days after my dad died (many years ago now), I went to a concert (trying to clear my head), and there was a remarkable performance of the unanswered question in Manchester. Strange how those things happen - it was a very meaningful experience - we imagine some kind of divine blessing at these moments.

Great artists tune-in to some fundamental principle of the universe. They often struggle to articulate exactly what that is, but it is clearly evident in their work. Academics are often arrogant enough to believe that they can unpick these fundamental principles - and make themselves look foolish in the process. In between the artistic expression and the academic "sense-making" is a process of loss of information. This is more pronounced in academic work which is "easier" to digest in the academy - that which divides things up into structural and formal relations, or carves it up on a spreadsheet. Its like how dissecting a frog destroys the living essence of frogness.

So if we have to academicise great art, there should be something as impenetrable in our scholarship as there is in the art. For me, the best theoretical work is like this, where theory has a similar structure to the thing it theorises. We tend not to think like this. We look to theory to "explain", where it can be something that accompanies us on a journey. It is a bit like how I described Gordon Pask's thinking about information the other day: not as a calculation, but a physical process. 

Music is a physical process. It is a process of physiological adaptation to perturbation driven by nature's tendency towards homeostasis. What drives that process we don't know, although there are emerging theories as to how it might happen. Much more difficult is to think how we might measure a process, or even to think what measurement actually means. 

Today we are used to using computers to measure things - often with statistical formulae. But what is a computer? What is a machine? So, here is Gordon Pask again on that with a statement I find very profound:

"The word 'machine' means a piece of hardware constrained, algebraically, to act as a computer" 

AI and Epistemological Correction

The capability of AI is likely to lead us to bad decision-making. But it needn't. What is remarkable in the statistical amalgamation of training data (particularly computer code) is the capacity to represent old thinking, which was probably never properly understood, in new ways which can be more readily understood. Some of this old thinking is deeply cybernetic, which also challenges present analytical techniques. 

One of those analytical techniques which straddles the cybernetic approach and present analytical approaches is Shannon information theory. Shannon entropy is a probabilistic calculation that is performed on a digital computer, but Gordon Pask's take on entropy was quite different from Shannon. He saw entropy as an enacted property by a physical machine, not as a calculation of a number of bits of information. The difference between high and low entropy was the amount of work that needed to be done in order for a system to establish stability following a perturbation. This is modellable as Shannon information, but it has a number of advantages over the Shannon equations. Primarily it places the emphasis on the whole system relationship which is not measurable as such, but which is enacted in the system behaviour. 

This is a much more biological way of approaching the whole issue of surprise. Surprise isn't a metric. It is an enacted systemic relationship. 

How can our present AI help with this? Partly because it makes it much easier to make "maverick machines" - those devices through which Pask explored his understanding (and in the process lost a lot of others!). Most importantly, those devices are analogue, not digital. But of course, present digital AI is quite good at creating (on-demand) analogue simulators. 

This really speeds up the process of getting up to speed with Pask's brain. I wonder if, in fact, very few of the other cyberneticians really understood him. Even his students became somehow dogmatic about his ideas rather than really thinking in the same way that he did. This was probably the fault of the pedagogical relationship between him and them (his airy Victorian engineer persona probably didn't help - and set a bad example to the succeeding generation). What if he'd been able to playfully create analogue computers on-demand to explain what he was talking about? (indeed this process was directly modelled in his conversation theory - the shared modelling environment at the bottom of his diagram). But there's more to it than merely the concepts. The analogue machine had to be coupled with a brain that operates with it.

There's something in these analogue machines that is missing in other cybernetic approaches. For example, Spencer-Brown misses something with the binary division of the mark, even if self-reference produces some interesting results. Shannon misses it by focusing on numbers for operationalisation. Maturana and Varela miss it because their "embodied cognition" (perhaps more Varela), while also embracing self-reference, isn't really embodied at all but metaphorical. Beer perhaps is the closest. For him, the organisation is the enacting device. But it is still difficult to envisage without simply falling back to Beer's demarcation of system 1, 2, 3 etc. and the march of his disciples.

With Pask we get the organic recursive system which must work on itself to establish homeostasis. The degree of complexity relates directly to the amount of work expended. This is why Pask's musicolour is particularly powerful. The organic system is us - and I think that is a clue as to how this enacted entropy might be operationalised. The key to an organisation's health lies in the physiological mechanisms of each of its workers. To understand those mechanisms we need to look to both cybernetics and biology.

Putting the details of that aside (pending a new paper!) that means that occupational health is deeply connected to organisational viability. Pask's enacted approach to entropy could be very powerful as an organisational tool focusing on the adaptation of each individual. Could this approach identify risks of pathological organisation or corruption? Maybe it could. 

What would W. Edwards Deming do with Higher Education today?

In July a report by KPMG and Mills and Reeve (Radical Collaboration: A Playbook | KPMG UK) was released about possible mergers and innovations in the future of Higher Education. It's good to have some blue-sky thinking - particularly at a time when universities are under such pressure. But looking through the ideas I was left wondering where  our strategic thinking is going wrong. The KPMG ideas include "Multi-University Trusts" in the same mould as "multi-academy trusts" in schools - an innovation which replaced local authority control of schools with corporations and CEOs on big salaries. 

This is well intentioned stuff - at best I think the authors believe that these structural changes will improve education and the experience of students. But what's noticeable is that there is rather little the report says about the experience of learning itself. Indeed, it's fascinating how little we talk about the experience of learning at all. We have professionalised the administration of the organisation of a particular kind of "education system" (with all its jargon which frankly I cannot keep up with), but we remain amateurs in our understanding of what it is to learn and to think. We tried to bureaucratise learning with ideas like "learning outcomes", but as Ron Barnett rightly points out, this was merely a kind of neoliberal conceit. In many ways, AI is highlighting this deficiency in actually dealing with the deficiency of thinking about thinking and learning. 

There is a deep asymmetry and disconnection between the design of the structural conditions of education from the consideration of learning and intellectual development. Almost all policy work focuses on the structural conditions, and assumes that the learning will happen if the structural conditions are right. The problem is that system design and learning experience are mutually constitutive. 

The deeper problem is that we have no methodological tools for making the connection between population-level study and individual experience. Yet every individual in education - learner and teacher - exists within a set of constraints through which each of us navigates in a step by step way: do a bit more of x, a bit less of y, and things move forwards. What Deming thought was that all organisations should work as organisms in a way where every unit (person) acts in concert with the whole. One way to conceive of that is to think that the constraints bearing on the whole translate to the constraints bearing on the individual, and that the interrelationships between these constraints can be made apparent. 

I was in a meeting the other day where high-handed decisions from the top were being enforced without any consideration for the constraints that those subject to them were dealing with. If this happens we get alienation and disintegration and, ironically, more top-down decisions as management seeks to reinforce control on a situation which resists control. 

But if we could see where we were, and where the high-level constraints were as a piece, I think this could be avoided. Deming's mantra was to study the whole system, rather than manage specific outputs or targets, under the broad heading of his "System of profound knowledge". His argument was fundamentally cybernetic - unmanaged variety was the problem, and variety needed structures and processes to manage it. Only by doing this could the whole system become more reliable and predictable.

Universities are interesting with regard to variety. They are very high variety institutions, but consequently management attenuates the variety with bureaucracy which becomes stifling. Management does not consider the constraints that each person - learner, teacher - works within, but imposes more constraints from above. This is partly because whatever chaos unfolds internally, the institution must present a predictable interface to society. 

Effectively what society sees is a top level aggregation of multiple constraints - an HE system which grants degrees and charges fees. It is subject to political forces. Management sees its duty to safeguard the interface to society - so its focus is on targets and external measures and not on internal organisational flow. I think this is why "heads of education" in universities are rarely experts in education. Perhaps if they were, things would be better, but the job as it exists is to manage targets which are externally inspected. 

The best of them understand that the targets can only be met if the internal flows work. It has to come down to listening to teachers and learners to work out where their constraints really are, and making local interventions to try to shift those constraints to make things work better. I think this is what Deming would do. I don't think he would impose blanket standardisation to force predictability on the system. That would merely mask it's pathology. Such a mask might temporarily be in the short-term interests of managers (who can claim their targets met), but ultimately it tips the institution in the direction of increased chaos.

Autonomy and Heteronomy - teaching students to make their own tools for learning

I've recently run sessions for summer school students from China and the US on public health. Occasionally you can get gasps from students who suddenly see that they can do something which they had never imagined they could do before. AI is great for this. To tell public health students (who by and large are not the most technical) "by the end of this session, you will be creating code for a public health app", and to be able to deliver on that in a way that surprised even me, was a great experience for everyone. 

I did an activity with them which asked them to design an app with focus on how the app would balance "what individuals can do for themselves autonomously" and "what is done for them". The biggest challenge in the exercise was actually to get them to think away from technology doing everything for users. When I asked one group who wanted to do something around vaccination about what individuals could do for themselves they said "follow the rules". We talked a lot about this!

The upshot of the exercise was that flipchart paper was filled with handwritten designs for their tool, what the system would do and what individuals could do. Taking a photo of this and putting it into an AI (we tried a number - copilot, chatgpt, deepseek), and asking the AI to write the interface code in HTML performed miracles - hence the gasps. Interestingly, by a long way, deepseek was  best at this. Its code was always correct (which copilot wasn't), but also far more detailed in its interpretation of the design than chatgpt (4o). 

But the discussion was more important than the tech. Behind my "what people can do for themselves" and "what is done for them" was the dichotomy between heteronomy and autonomy which Illich borrowed from Kant, applying it to technology arguing that technology reinforces the heteronomous side at the expense of the autonomous side. My little exercise may have done something to push things back to the autonomous side... albeit with large heteronomous AI models (although these could be accessed offline). All this is fascinating. 

But broadly speaking, Illich was right about the pressure to absorb everything on the heteronomous side, and he attacked most public services for doing this, from education to the health service. It is, I suspect, this heteronomous absorption which lies at the root of our current political, environmental and economic crisis. This challenge is consuming me at the moment - the real issue with AI is not AI - it is our approach to organisation which does what Illich complains about. From Silicon valley to our universities it is evident everywhere. 

40 years ago in Manchester, Enid Mumford was designing technical systems with users (for her, it was nurses). She insisted that systems should be made with people, not done to them. This was also a call for balancing autonomy with heteronomy. Frankly, it didn't really catch on. Increasingly - and perhaps out of necessity - computer system design became professionalised. But if people can create their own code with AI, perhaps there is a new opportunity for revisiting this. Some of the technical foundations for a shift are already there - particularly in Service-Oriented Architecture initiatives (my early work on Personal Learning Environments was based around users constructing their own tools around a Service-Oriented Architecture). AI gives us a new angle on this idea.

So one simple thing we could do to address the heteronomy/autonomy balance is to teach students to make their own tools for learning. This would be, I think, far more effective as pedagogy than any attempt at "AI literacy" which seems to be where things are going. All that does is create more power for the heteronomous power of the educators (who often are not experts in this stuff) to attenuate the creative forces which have been unleashed by the technology. Yet it is these creative forces which will be essential for the survival of graduates in the world as it is unfolding.

When the gen-AI thing broke, I said to university audiences (in Denmark, China, Manchester) that what people should worry about is what bright 13-year olds would be doing with the tech. By the time they come to University they are going to be hugely advanced not only technically, but also in their knowledge elsewhere. The world is such a weird and confusing place to be a child, and those children with the space and technology to ask questions will be exploring amazing things right now. I had a reminder of this when a friend who I invited to the Laws of Form conference last week told her 8-year old son that she was going to meet a world expert in topology. "I know about topology - that's about knots and space," he said, "I've been watching loads of videos on youtube. A circle is an un-knot - it's amazing". He then went on to tell her about the square root of -1. 

This may be one particular child. But I doubt it. Children have fresh brains and the world has lots of big worrying questions, and incredibly powerful means of finding things out and producing technologies. They are likely to make the most creative use of these tools. But they are then likely to encounter the heteronomous education system. The heteronomous side may win (again). But I'm not sure. And I'm pretty certain we should push against it.

The problem is that to rebalance heteronomy and autonomy is a real challenge to the way things are organised - particularly in the wake of a commodified education system (for which read, heteronomy-dominated). It is also a challenge to those who gate-keep the heteronomous side because it is in their interests to do so. But it is almost certainly not in the interests of the next generation.