Are models objective? – Aidan Ward – Medium

Go to the profile of Aidan Ward

Are models objective?

Aidan Ward and Philip Hellyer

It is vital that the observer takes on responsibility for their observations, language, and action. The observer is inextricably linked to the object that they are observing. Heinz von Foerster

Taking responsibility for perception

Is an explicit model such as a mathematical equation or a theory in theoretical physics objective? Should it be treated as independent of the people observing it?[1]

My introduction to this question was many moons ago in St Andrews. I participated in a course about Larch, an algebraic proof language designed to do rigorous analysis of technical systems and programme code. The class solved equations for a ring of oscillators that excited each other. (Hmmm…) Anyway, the real excitement came when this rigorous proof system came up with two very different solutions. I think the instructor was genuinely concerned but worked out that the two solutions corresponded to two physical states: synchronous and asynchronous oscillation. For what it is worth, the cover of the Larch manual features an Egyptian holding a scroll and a pyramid being built upside down.

For half my life, I have held that this event shows that theoretical jiggery-pokery can lead to enlightenment about the “real world”. I think I have changed my mind.

Let me recommend The Order of Time by Carlo Rovelli. Carlo is a world-class theoretical physicist working on quantum loop gravity. He can write like an angel, so his deep and deeply confusing theoretical explorations are paralleled by highly accessible writing that keeps him engaged with fully human and properly subjective questions.

In Rovelli’s models, time is not universally flowing. The notion that time happened whether anyone was observing it or not, like Berkeley’s tree falling in a forest, is down to Newton who got it precisely wrong. Newton is very interesting to us here because his work was funded, essentially as a political programme, by the then equivalent of Atlantic dark money. Rich men wanted to change the world to their advantage and a thinker like Newton was what they needed to upend the power of the Church, amongst other things. Newton himself was ruined by this process, becoming bitter and feeling used despite his apparent success.[2] So much for being one of history’s great minds.[3]

Separation and connection

In The Master and his Emissary, McGilchrist makes a magisterial case for the separation of functions between the two hemispheres of our brains. The title holds the key to the true subject of the book. We have two fundamental (and fundamentally different) sets of thinking functions, one much more accessible to conscious thought than the other. McGilchrist makes the case that the conscious, languaging, modelling, classifying left hemisphere functions must remain the servant of the right hemisphere master, though they have not in western culture.

Remember Blake, always remember Blake:

May God us keep from single vision and Newton’s sleep.

Newton erected a model of time that was independent of the observer. It is a common trope that we have become slaves of time in our modern world.[4] I am in the process of freeing myself from that slavery and can feel it very strongly. That whole notion of the clockwork universe is not as separate from Newton and his enslavement to his sponsors as it would need to be if it were as objective as it was made out to be. I think this is the classic case of a whole world of illusion, taken up as rigorous truth by the likes of Michael Gove, who also has an enslaving agenda and is himself in hock.

Rovelli can be our counterpoint here. His presence in his writing is much more human; he explains that time can only be properly talked about as a partial ordering. We can tell, in a particular location, which event came before which other event and we can see local causation looking back and looking forward, but we cannot establish a global scheme.[5] The key equations for quantum gravity do not have time as a parameter and work equally well in either direction.

In this model, the emphasis falls on the observer, and what they can actually observe in terms of ordering. Consider the approach of the pin and the popping of the balloon. Indeed, the model says that the world is made of events, not of things, and the seeming existence of a thing (the pin or the balloon) is merely a drawn-out event. The world is made of events, some of which interact to provide the partial ordering. Think of this alongside Nora Bateson’s concept of warm data — it is all about relationship, and if there is no relationship there is nothing. Precisely not Newton’s objective, external world of things and laws.

Carlo Rovelli’s work is revolutionary but without Newton’s angst. If you follow this line of thought, you will end up with and experience of the world completely different from that which you have now. The work is based both on a mathematical rigour (unimaginable to Michael Gove) and on a deep humanity and respect for our lives (also unimaginable to Michael Gove). I am hammering on poor Michael Gove because I think he illustrates how easily we become dead wood. There but for the grace of God go I.

A world run by algorithms

Yuval Noah Harari in his 21 lessons for the 21st Century says that we have very little ability to escape our lives being subjected to algorithms. Whether or not we get a bank loan, whether or not we get accepted for uni, whatever.[6]An algorithm is just an implementation of a model of course. And in that sense models become objective for us because they are imposed without any control on our part and certainly without our help in making them make sense. They are also objective in the sense that once understood they can and will be gamed, which is a fundamental limitation of models and algorithms based on them.

Harari’s advice is just where we are with this blog post. He says the only chance of not being controlled by someone else’s algorithm, whether Amazon’s or Facebook’s or the government’s, is to know yourself better than the party attempting to manipulate you does. If they know you better than you do yourself, you will be in the position of a small child and will be subject to power plays and manipulations by the “adults” in your life.

So, the algorithms and the models they implement must remain the emissary to you as a master and you do that by knowing yourself, by organising politically, by outsmarting those algorithms in every possible way. You remain master by remaining master and there is no passive way to do that. No way you can do that by spending money. No way you can receive an education that will equip you, because all those masters have already been captured. If this blog can help you move towards freedom, I will be so grateful to the universe!

The fundamental, without-which-nothing move in knowing ourselves is to understand our own architecture a la McGilchrist. If we allow our left-brain emissary to take charge of the right-brain master we can never gain control over our own lives.

We are not thinking machines that are emotional, but emotional machines that think. Antonio Damasio

Models as language

Our language and languaging is not neutral. Our language has built in to its grammar a world of things. It is known to be hard to express our world as events and processes. I think you are supposed to read one of the A.N. Whitehead books several times over so that it can gradually reveal itself. Some people think that things that are hard to express must be themselves muddled and unclear, but it seems that our language has genuinely mistaken the nature of the world and how we experience it.

It is useful therefore to have dynamic models, ones that naturally express change, to share with each other as a communication mechanism. My first serious foray into this as a consultant led to me seeking advice about the consultancy process.[7] The advice I was given was that the model MUST be the clients’ model and if there were more than five elements in the model then the client would probably not understand it. It has to be said (before someone else points it out!) that the modelling consultancy work that Philip and I did together did not follow this advice, not even nearly. That however says as much about the nature of the relationship with the client and the nature of the contract as it does about Philip and me.

The most famous such model is probably still that produced by the Club of Rome in its 1970 report on the possible futures of the world. I quote this here only to shine a light on communication. It seems that most people looking at the dynamic model’s outputs and the text of the report treat it as a set of predictions, and are content 40 and nearly 50 years on to assess the quality of the modelling by how accurate its predictions were. But that is not what the model is for and it is not what the report says. The report is quite explicitly a set of policy scenarios: if the policy followed in practice is such-and-such then the implication in the model looks like this, and if some other policy is followed a different scenario of outcomes is generated. That is, the policy options are what is language so that the model can be used for communication. That the communication ensuing on the report did not really get engaged in policy debate is the point we want to explore here.

For a more recent example in the UK, the Munro report into Children’s Services also used extensive systemic models. Eileen Munro herself says that the government asked for most of the systems diagrams to be taken out of the report before it was published. And again, the debate on the report was not noticeably improved by having the models as a communication vehicle, though the analysis was good. It takes more than a good technical communication medium to have a good conversation!

Architecting-in conversation

Our brain certainly works in two separate hemispheres with a connecting bridge, the corpus callosum. That is its fundamental architecture. The two hemispheres have different functions. We are not used to dealing with functional architecture in this way, but I used to be a software architect in that sense.

Very approximately, our right hemisphere deals with the unpunctuated flow of experience and our left hemisphere deals with sensemaking. The whole schema is ever so plastic and parts of the brain get seriously repurposed when necessary. It should be obvious that the conversation across the connecting bridge is what brings the power of the architecture to life. When we privilege a model over reality as we experience it then we are making the left-brain emissary the master. (This is absolutely what economists do.)

In terms of signals within the brain, this bridge conversation happens several times a second. In terms of our ability to let reality correct our sensemaking, in our culture it can become vanishingly rare. One of the underlying vitally important questions is whether we can communicate with each other without resorting to the dominance of this language and modelling and sense-making. Everything we do that reinforces our cultural tendency to believe that language and models are real keeps us from rebalancing our own minds. Every empathic solidarity, by contrast, lets us know that the way we experience the world in flow is to some extent both shared and reciprocated. Truly when we are in love everything changes.

Truth, a big word indeed, is a process of staying in touch with our own reality, not an external, still less universal state. Carlo Rovelli is the gentlest of guides at arriving at why this must be so!

[1] I feel as though I ought to mention Cathy O’Neil and her criticism of algorithms in Weapons of Math Destruction. They all reflect the biases of the underlying data and the existing prejudices of society. For instance, whatever you think of the education system, it was made worse by the ‘generous’ intervention of the Gates Foundation and the resulting evaluation of teachers. But we get to algorithms later in this post.

[2] Let’s not forget that he was as fascinated by alchemy as he was by optics. The apocryphal falling apple just happens to have survived the scrutiny of history…

[3] Newton himself gave credit to his predecessors: “if I have seen further it is by standing on the shoulders of Giants”

[4] I never fail to chortle/shudder when I see UK corporations making declarations under the Modern Slavery Act. The formal protestations only disguise more subtle, endemic forms of slavery that are generally considered to be normal and natural.

[5] If I dare to cite Wilk and O’Hanlon, there may be an objective underlying reality, but we should not expect it to be consistent and coherent. Coherence is a sign of story-telling and abstraction from reality.

[6] In the US, increasingly, whether or not you get accepted for a job, even a McJob. The biases baked into these algorithms are almost literally incredible. My credit score is never so high as during times of personal financial crisis; what’s the logic of that?

[7] Not always a conversation that can be had. One of my local police was asked for advice by a colleague who’d just caught a 17-year-old with marijuana. The official line was (and so the advice had to be) that he needed to be booked. Ironically, had he been 18, an informal warning is now policy, but that doesn’t apply to minors. Had my local officer been the one on the scene, a quiet confiscating and forgetting would have occurred, but he couldn’t risk advising another officer to do the same…

Source: Are models objective? – Aidan Ward – Medium