Algorithms versus the World
Modern mindsets about solving problems and finding knowledge (Part 1)
One sign of our current apocalyptic mood is an extensive genre of social and cultural commentary asking where things have gone wrong. One striking cluster of thinking has focused on ways our broader economic and cultural systems have become, somehow, inhuman. The criticisms take various forms, from denouncing neo-liberalism, to critiques of surveillance capitalism, technocratic mindsets, diagnosing our world as drifting into neo-feudalism to analysing about our governing mindset as embodying the Machine. The shock and societal impacts of responses to the covid pandemic triggered extensions of this thinking into concepts like the bio-security surveillance state.
When you find similar analyses from a range of divergent perspectives, there is clearly something going on. To identify the common elements, if any, it often pays to consider the underlying conceptual forces and philosophical assumptions at play. Given this is Humble Knowledge, our natural focus will be the epistemic attitudes within our modern world and their role in phenomena being analysed.
However, this is not a simple story as the epistemic attitudes need to be placed in the context of broader attitudes. This analysis will therefore be split into two posts. The second will focus on the epistemic attitudes. We first need to explore key features of how our modern, industrial culture thinks about the world.
Our modern mindset
The driving cultural force of that form of life we call “modern” is the idea, hope and desire that we can make the world controllable.1
With this starting point, Harmut Rosa, a German sociologist, begins his sociological analysis of the modern way of life in his short book, The Uncontrollability of the World, published in English in 2020. The book makes a compelling case that a central dynamic of modern life is the tension between this desire for control and the fact that:
it is only in encountering the uncontrollable that we really experience the world. Only then do we feel touched, moved, alive.2
One of the consequences of this is that:
the basic institutional structure of modern society can only be maintained through constant escalation …… in other words [it] requires constant economic growth, technological acceleration and cultural innovation in order to maintain its institutional status quo.3
If Rosa is correct, our driving desire for control gives rise to self-perpetuating modes of life and societal structures that are at odds with what really matters for us as humans. It would therefore be unsurprising that so many feel like modern society has become somehow inhuman.
There is plenty of evidence that belief in the controllability of the world is widespread in our societies, particularly within the societal groups that are culturally dominant. A few examples should illustrate the point and readers can likely provide their own.
Child-rearing has become for many an exercise in sophisticated project management to ensure our children are successful and turn out right.
There is a concerted, ongoing push to eliminate risks and make everything safe.
Governments and leaders are expected to solve any problems we come across.
If something doesn’t work or goes wrong, it is habitually taken as evidence of failure, not bad luck or an unfortunate accident.
What is hopefully clear in these brief examples is that, and this is the critical point for Rosa’s analysis and what follows here, there is a strong belief that the whole world, or at least everything that matters, is controllable.
The belief was especially evident in how people across the world talked about and responded to the covid pandemic. The implicit starting point in almost all public debate was that we could control or manage the virus and its trajectory. The disagreements were over how we do it. Was suppression or elimination the best strategy? Should we be using masks, or stay at home orders, or vaccines, or ventilation?
The alternative view - that we couldn't actually achieve much beyond helping the people who got sick along the way - was treated as outrageous in public debates and provoked huge amounts of scorn. Anyone who suggested it was accused of ‘letting the virus rip’ and ‘killing people’. Those who thought that the virus was not controllable, even though it has turned out to be an accurate belief, were treated like heretics.
This attitude also helps explain why public discussions of covid focused almost exclusively on interventions designed to control the virus, and downplayed or ignored treatments and mitigations. Vaccines are a method of controlling the virus and they were a clear priority; whereas Vitamin D, anti-inflammatories, exercise, or drugs like ivermectin were generally ignored or dismissed. Adopting these as an important part of our approach to the pandemic would have been an implicit admission that we couldn't control the virus.4
Limits on our control?
So far, we have skipped over a weakness in Rosa’s articulation of his starting point. The hope and desire that we can make the world controllable is not unique to people today, but has been part of human culture since the beginning of history. Cultural institutions like absolute monarchies, slavery and many religions aimed at control of the world, as did practices ranging from magic incantations, rituals and terraforming landscapes with fire.
While the desire for control is embedded in the human psyche, almost all historical cultures believed that our ability for control was circumscribed. They almost always had deeply ingrained concepts like Fate, the will of God, or events being in the lap of the gods. The modern mindset that Rosa explores, on the other hand, does not include any such concept. That is, we no longer recognise anything as, in principle, outside our human control. We believe that the whole world is controllable.
This is a striking and bold assumption that seems self-evidently false to many and yet so intuitive and natural to others that it is hard for them to imagine it not being true. It is therefore worth making a few partial observations about the source of this modern mindset.
The first source is unsurprisingly our success with modern science and technology. We have been able to control aspects of the world, for example through air-conditioning, modern nutrition and medicine, or digital technologies, that people in the past could never have imagined possible. We have been successful in so many ways that it is natural to imagine these successes can be indefinitely extended.
A second observation is that, despite it being an intellectual project specifically targeted against the modernist scientific mindset, postmodernism and critical theory have generally strengthened our belief in the controllability of the world. If power relationships and language significantly structure reality, then reality is much easier to change than if it is predominantly physical matter. As in our observations on misinformation, two dominant cultural traditions converge around this assumption.
These two sources also overlap with views on the balance between nature and nurture in determining who we are. The more we downplay the role of any inherent nature in determining our future, whatever the source, the more it feels like we can change ourselves (and the world) to match what we want them to be. In other words, the more we assume that nurture is what counts, the more controllable we see the world.
And finally, the advent of modern digital technology means that many of us spend most of our time inhabiting worlds and systems that are entirely constructed by humans. The digital world is perfectly controllable and infinitely malleable. We tend to transfer this mindset to the world around us and instinctively assume the real world is controllable.
Algorithms for control
There remains a question underlying Rosa’s diagnosis: by what means do we consider the world controllable? If we are confident that the world is controllable, this presumes that we have some means by which we can achieve that control (or at least feel like we do).
One obvious answer to this question is that we achieve control through modern science and technology. However, if we examine this in any detail, we realise that science is a collection of related but distinct methods and approaches - the practice of biology is very different to say physics - and we have many technologies rather than one technology. Clusters of varying techniques for varying uses are not a robust basis for a belief or feeling that control of everything in the world is possible. Instead we need something more universally applicable.
A compelling articulation of such a total approach, grounded in the practices of modern science and technology, was articulated by French thinker Jacques Ellul in his book The Technological Society, written in the 1950s and translated to English in the 1960s. Ellul based his analysis on a concept that he called technique. I should note that Ellul is well worth reading but can be difficult to understand. I first read this book as an undergraduate and it took me well over a decade before I felt like I got a decent handle on what he was saying.
Ellul’s style is to treat cultural and sociological trends and forces as independent agents which can cause change and which progress based almost on their own will. While this way of thinking feels uncomfortable as it leaves the human out, it can be a powerful way of analysing the inherent logic of cultural and social mindsets and beliefs. The logic within a set of beliefs will typically play out regardless of whether the individuals involved want it to. This, at least, is how I understand much of Ellul’s writing. I should emphasise that what follows is probably more an explanation of what I have learnt from Ellul than how he would explain anything.
To understand the total mindset he explored, we will begin with Ellul's definition of technique in his preface to the English translation of The Technological Society:
Technique is the totality of methods, rationally arrived at and having absolute efficiency (for a given stage of development) in every field of human activity.5
Technique is a totality - a mindset that is all encompassing and applies to everything humans do. It consists of a large set of rational and widely applicable methods, which Ellul often refers to as techniques. In Ellul’s language, techniques are individual methods while technique is the universal force or mindset that shapes how we interact with the world and has its own internal logic.
If we grant that Ellul’s technique exists, by his definition it will provide an effective and efficient means of control in anything we try to do. Thus, technique as a mindset would be sufficient to justify a belief in the controllability of everything in the world. To connect Rosa and Ellul: we believe the world is controllable through technique.
To understand technique, however, we need to understand how it is made up of, or grounded in, techniques or rational methods. Ellul gives many examples of techniques but never attempts a definition. From these, we can see that a technique is a rational method to deal with a situation or problem that is to be applied independent of the individual people involved. That is, it is an abstract set of instructions or processes that can be used universally.
We are all familiar with techniques and they play a valuable role in our life every day. What turns the use of techniques into the mindset technique is the conviction that the techniques have absolute efficiency, that is that they are entirely sufficient to properly organise that field of human activity and solve all the relevant problems. For Ellul, assembly lines, modern militaries and scientific management practices exemplify techniques organised in the mindset of technique.
This framing is unfamiliar and can be awkward at times. We can however connect them with more familiar concepts. One way of thinking about Ellul’s rational methods or techniques is that they are:
A finite set of unambiguous instructions that, given some set of initial conditions, can be performed in a prescribed sequence to achieve a certain goal and that has a recognizable set of end conditions.
Or
a precise rule (or set of rules) specifying how to solve some problem; a set of procedures guaranteed to find the solution to a problem.
These are literally dictionary definitions of an algorithm and adopting this as an alternative framing of Ellul’s account helps us more easily understand his analysis. To boil it down to a few points:
As a mindset, technique is the fundamental conviction that every field of human activity can be best managed by algorithms, or algorithmic approaches.
Achieving our goals or ambitions depends therefore on the ongoing pursuit of ever more efficient and effective algorithms; and we are convinced that getting the right algorithm will solve the relevant problem.
The internal logic of technique means that the goal of finding more efficient algorithms is itself generally achieved via algorithmic means. This creates a totalising approach in the sense that the internal logic drives changes without deliberate human direction.
Further, Ellul argues that the set of algorithmic techniques becomes the means by which we interact with reality: we increasingly function within the reality defined by algorithms and rules rather than reality as it is independently.
This is a bold and distinctive line of thought and I would recommend you read Ellul to understand it better. One important clarification is needed about the term algorithm in this context. While we increasingly associate algorithms with computing and digital methods, the concept as used here depends on its much older and broader meaning. The computing application of algorithms is nevertheless a useful subset and some of these points may seem more obvious if we restrict ourselves to digital algorithms.
Articulating Ellulian technique in this way allows us to better explain the connection with Rosa's analysis: the driving force of modern life is the conviction that the world is controllable by means of adopting efficient algorithmic approaches for all issues. To put that differently, we seek to control the world and solve problems by breaking things down into repeatable, effective processes that can be adopted everywhere.
Our algorithmic mindset
This fusion of Rosa and Ellul makes a very big claim that cannot be proven in one post. Instead, I can only give some examples and leave it to readers to decide whether it rings true or not. Note that the claim is that algorithms, finite sets of unambiguous instructions, are sufficient or all we need to solve the problems we face. Even if the instructions are often beneficial, the defining question is whether our mindset assumes they are enough in themselves to solve the problems. Some examples of this mindset include:
At the personal level, there are endless articles and books that outline the 10 steps to perfect health; or 6 easy steps to lose weight; or everything you need to do to be happy. All of these are based on providing a finite set of instructions (an algorithm) that is presumed to be universally applicable and will be enable you to achieve your goal.
In the workplace, safety and cultural issues are almost universally tackled by step by step training programs and compliance activities. In other words, they rely on the conviction that a precise set of rules will be sufficient to solve the given problem.
Governments primarily focus on solving problems through specific policies and programs - which are by definition a finite set of instructions or rules that are universally applied. The underlying conviction is that once we find the right policies, i.e. the right algorithm, the problems will be solved.
There was a proliferation of sets of instructions to deal with the covid pandemic. Governments and organisations worldwide worked on and produced long lists of rules and procedures and people by and large believed that by following these algorithms we would control the pandemic.
Many readers may find these descriptions of methods for solving problems to be intuitive and natural. Some may not be able to imagine there is any other feasible way of doing things. However, this mindset, and especially the belief that algorithmic methods are all we need to solve these problems, has only become dominant in the last century or two, and only in the industrialised parts of the world.
The existence of alternative ways of thinking can only be illustrated here by means of a couple of examples. These have been drawn from within the Western tradition but other examples will likely be even more obvious to anyone who has dealt with cross-cultural issues, especially with older, more traditional societies.
Historically, education was concerned to a large degree with developing good character. Today, we are predominantly focused on skills, competencies and proficiencies. These are the sorts of things that we have methods or algorithms for determining: we can make someone sit various tests or perform various tasks to determine whether they have the right skills; and we can come to a definite answer. Character, on the other hand, is a judgement about what we think a person is like and how they would act in a range of circumstances. We can neither develop an algorithm to teach it nor a definitive test to decide whether someone has a good character or not.
A related example is in the common experience of trying to borrow money from someone, typically these days from a bank. Once upon a time, the decision about whether to lend you the money would have been made by a person who knew you (somewhat) and made their decision based both on your financial situation and their assessment of you as a person, particularly whether you were the type of person who would pay back the loan. Today, the decision gets made by feeding your key financial data into a risk model (in essence, an algorithm) which treats you the same as anyone else with the same financial data. Even if you see a bank manager, they don’t get to make any decision themselves.
The point here isn't to debate which is the better approach. It is to illustrate ways that our modern approaches and mindsets are predominantly algorithmic. Moreover the alternatives that continue to exist seem to feel tenuous, as was neatly expressed recently by Mary Harrington:
I dare say there still exist longstanding practitioners of any number of occupations, whose main body of activity involves tactile, holistic, time-bound knowledge of this kind. What such experts know, and how they know it, is nigh-on impossible to convey in recipes or formulas. Because of this, it’s generally taught via apprenticeships rather than books. And it’s a kind of knowledge that can be temporarily eclipsed by machines, or by chemicals — but only at a price.
It is important to note that, as Ellul repeatedly emphasises, our algorithmic approaches (or techniques) are incredibly effective and useful. For example, at a systems level, two technological marvels of the modern world are the internet and our global transport and logistics system. One moves mind-boggling amounts of information and the other mind-boggling amounts of stuff around the world at ridiculous speeds, both at incredibly low costs and high reliability (at least until covid disruptions hit). Both systems are triumphs of algorithms and the relentless optimisation of algorithmic methods.
Modern epistemic attitudes
We can now go back to the range of societal critiques we began with and identify a common underlying cause for the range of ways they argued societies have become inhuman. All of the critiques identify ways that our society insists on using algorithmic methods to control the world and solve all our problems - and highlight the ways these are damaging people, societies and our Earth.
Most of these critiques also provide evidence for the limitations and weaknesses in our presumption of the sufficiency of algorithmic methods for controlling the world. We will leave the cultural, economic and sociological analyses to others and focus here on a set of important epistemic issues. If we want to solve many of the other problems, we need to consider the epistemology as that underpins our approach to the world.
As Rosa points out, “making the world controllable means …. making the world knowable.”6 Controllability presumes knowability: if we don't know how things work, we cannot control them. Thus, this modern mindset that presumes we can control the world depends on a profound epistemic confidence. This confidence has already been partially explored in the previous discussion of misinformation.
Nevertheless, it obviously runs counter to the arguments we have been building here at Humble Knowledge for epistemic humility: that knowledge is hard and we typically can't be sure we've got it right. We have covered some evidence for epistemic humility in the history of philosophy and quantum physics.
We have not yet explored the question of how we think we can find knowledge and be confident in its reliability. That will be the focus of the next post, the second part of our exploration of modern mindsets about finding knowledge. Building on the analysis here, algorithms will play a central role in the story. Can we use universal algorithms or processes to find out what is true? Or are we limiting ourselves by relying on them?
Hartmut Rosa, The Uncontrollability of the World, Polity Press, 2020. p. 2. Emphasis as in the original.
Ibid. Emphasis as in the original.
Ibid. p. 9. Emphasis as in the original.
NB: I make no claims about whether any of these treatments work. Some probably do and some probably don’t. The important point is the tone of the public conversation.
Jacques Ellul, The Technological Society, Vintage Books, 1964. p. xxv. NB: Ellul later thought that using the technological in the title was a poor translation that caused significant confusion.
Hartmut Rosa, The Uncontrollability of the World, Polity Press, 2020. p. 15.
So much in this. Terrifically interesting.
The thread of epistemic attitude is important and, I am conscious, is to be revealed further next time. But I wonder whether epistemic confidence needs to be unbounded for the dynamics you outline to exist. Confidence in things being more knowable via a direction of intellectual is different to confidence that things are ultimately knowable - especially in a holistic way. I wonder also whether some other important factors are at play. A drive for efficiency, for example, leads to the use of algorithmic tools even when we know the resulting answers will be substantially wrong some of the time. The requirement of accuracy is limited to being mostly right often enough to be efficient in a utilitarian sense.
The interaction between society and the individual might also be worth exploring. It is not clear that our confidence in controllability rests deeply at an individual level - indeed I wonder whether it is an acceptance of individual uncontrollability that creates the demand we see for books and governments which offer a path towards more controllability. This is an inherently democratic dynamic, which exists all societies (democratic or not). It is in the interests of 'the people' that governments all types respond actively to this demand. But with that positive comes negatives. One is the offer of false levels of control at a point of time, if not through time. Another, which is related, is the outsourcing of responsibility for managing individual uncontrollability to society.
Your good character versus competence basis reference is fascinating. I suspect it is true that ancient writings (Confucius, Plato etc) emphasised character over competence. Whether that reflected how society actually operated is less clear (to me at least). Both are clearly important. I dare say Confucius' superior man and Plato's philosopher king were intended to be of good character and good competence. For what it is worth, my read of societal direction today is that character is strongly reasserting itself as a demanded quality, even if this is expressing differently in China (social credit) and the West (me too, amongst others). I know this is not the main thrust of your discussion, but thought it was an interesting side path.