Modernism versus pragmatism
|The design of legal products
Modernism vs. pragmatism
- Vertex vs. edge
- Text vs. meaning
- Algorithm vs. heuristic
- Formal vs. informal
- Tool vs. application
- Innate vs. emergent
- Obvious vs. subtle
- God vs. Darwin
- Simple vs. complex
- Quantitative vs. qualitative
- Calculated vs. interpreted
- Static vs. dynamic
- Stocks vs. flows
- Structure vs. interaction
- Nouns vs. verbs
- Trees vs. wood
- Permanent vs. ephemeral
“I should explain that in the Soviet scientific community in those days, mechanistic determinism held sway over all other approaches. Researchers believed that the natural world was governed by the iron law of cause and effect. This mentality was a product of the political environment.”
- — Cixin Liu, Ball Lightning
A running theme in the JC is the distinction between top-down and bottom-up of organisation models, particularly where it comes to dealing with existential risk.
We are in the swoon of an obsession with data, technology and the algorithm. Thought leaders perceive an inevitable, short, path to a singularity where everything can be calculated, everything planned and we will no longer have to rely on irrational, costly, inconstant, error-prone meatsacks. Only crusty old refuseniks can’t see it.
Now I am just such a crusty old refusenik, and while that is largely borne of self-interest — I am an irrational, costly, inconstant, error-prone meatsack, after all — before mortgaging our futures to the machine, it is worth nutting through the digital prophecies to see if they hold water.
We start with a fundamental, philosophical divide between, on on hand, modernism: in a deterministic world wherein the causal principle holds it follows that, in theory, we can calculate all outcomes from first principles. In this world the main challenge is outright data processing capacity and the sophistication of our model. Encouraged by the success of artificial intelligence in solving problems not long ago considered intractable — chess, alpha go, self-driving cars, facial recognition, chatbots and so on — modernists extrapolate to a world where risk is atomised and calculated out of existence.
On the other hand is pragmatism: whether or not the causal principle holds, and however good robots may get at chess, they cannot manage non-linear interactions and dynamic environments which present “wicked problems” and complex systems. Not only can formal logical tools not deal with rapidly emerging risk situations, they can’t even see them. It’s better to live with uncertainty, deploy experts, proceed with caution, keep slack in the system and use practical rules of thumb to which we have no great metaphysical attachment to address contingencies. We should live on our wits and figure things out as we go. There is no certainty, but humility is no bad defence.
Determinism begets modernism and sees centralisation and automation as the highest good. We should aggregate and optimise processing power. The main role of the executive is orderly administration and maintenance of a machine which, by cold operation of logic, will dispense optimal outcomes by itself. The less we interfere the better.
Pragmatism begets systems thinking and aspires to decentralisation: the world is fundamentally unpredictable; it is best dealt with by experienced experts; management’s main function is to empower and equip experts and optimise their ability to communicate. Pragmatists prioritise relationships and interactions and over equipment and structure.
So; centralised algorithms versus distributed heuristics.
Perfection versus good enough.
Determinists are from Mars; pragmatists from Venus.
“Modernists” view organisations, and systems, as complicated machines. Formal design is important, and follows (centrally-determined) function; the more efficient your contraption is, the better it will navigate the crises and opportunities presented by its environment — the market. Modernism regards the market as an extremely complicated mathematical problem: hard, but— theoretically — calculable. Modellable. Should a model not work, one must refine it.
Shortcomings in current technology mean we cannot — yet — fully solve that problem. We still need humans to make sure the machine operates as best it can, but the further humans are from that central executive function, and the better the algorithm gets, the more humans resemble a maintenance crew. Their task is to ensure the orderly functioning of the plant. As technology advances, human agency can be progressively decommissioned.
Business, and government, suffers from a kind of physics envy. It wants the world to be the kind of place where the input and the change are proportionate: everything is numerically expressible and the amount you spend on something is proportionate to the scale of your success.
The modernist narrative focusses on what it can see, which is necessarily limited to the formal inputs and outputs of its own model. There are at least two consequences of this.
Modernism only sees what it can see
Firstly, modernism cannot “see” informal, but vital, interactions between components of the system that are not in its model: random acts of kindness, the star seller’s sales technique, the time every staff member spends building lateral relationships, the value of those relationships, the necessary work beyond the service catalogue, the work-arounds that the machine going; the ad-hoc tricks that make up the difference between meaningful performance and work-to-rule. The critical call that clinched the deal.
Theory: it is not an organisation’s form or structure, but its interactions that determine its outcomes. A badly organised firm that interacts well with its customers will perform better than a perfectly organised firm that interacts poorly:
“You are in a queue. an operator will be with you shortly. Your call is important to us.”
An organisation is a system. Systems are comprised of stocks, flows and feedback loops. Key are their interactions: components that don’t interact are not in a system. A system’s behaviour emerges from how its components interact, not what it is made of or how it is arranged. In other words, a system is defined by its “flows”, not its “stocks”. Flows create feedback loops. Feedback loops create, or deplete stocks. Stocks — formal, structural capacities — only facilitate flows. But formal structures are easier to see than interactions. Yet modernism focuses on formal structure, organisation lines — mostly vertical — not functional communications, which are mostly horizontal.
Modernism as a negative sum game
Secondly, thanks to its physics envy, modernism is a negative sum game: its baseline is immediate, costless, faultless performance. Positive variance from this baseline is not possible: the goal is to lose as little energy as you can. But friction, gravity, heat, entropic energy loss means in the real world, the system loses energy. We can minimise entropic loss with engineering and environmental control but it remains practically impossible to conserve energy, and theoretically impossible to create it.
Human operators create a great deal more entropic loss than unattended algorithms. If the only gauge is accurate, efficient execution of instructions, humans must be worse at that than machines. Modernism cannot give credit to insight, diagnosis, model revision or reimagination because there is nothing to reimagine. There is no such thing as a valid alternative model. Economics is a kind of physics. There are no “alternative facts”.
Now, if bettering an algorithm is impossible, it stands to reason: meatware is expensive and inconstant: the largest risk to the organisation is human error, thus the strategic direction of an organisation’s development is to eliminate it where possible. Where that is not possible, human activity should be constrained by rigid guidelines and policies to reduce the probability of mishap, and monitored and audited to record and correct those errors that do happen top prevent them happening again. To the modernist, malfunction and human error are overarching business risks.
This worldview is one that appeals to many people in business management. Others might find it it rather desolate. But desolation is no argument against it if it is correct.
When a man throws a ball high in the air and catches it again, he behaves as if he had solved a set of differential equations in predicting the trajectory of the ball. He may neither know nor care what a differential equation is, but this does not affect his skill with the ball. At some subconscious level, something functionally equivalent to the mathematical calculations is going on.
When you get too close to your material, sometimes you can’t see an absurdity even if it pinches you on the nose. Not only does a person catching a cricket ball not solve differential equations, or anything “functionally equivalent”, but she can’t. She would need the inputs for every differential equation in play. Just to determine a trajectory is: Y = H + X * tan(α) - G * X² / 2 * V₀² * cos²(α).
To be clear: it is’t simply that she is doing this subconsciously, somehow, in a way she can’t apprehend: she doesn’t know, even subconsciously, the velocity, angle, vector, or starting coordinates of the ball. She would need all of these just to perform that differential equation. She is doing something else: she’s using a subconscious heuristic: keep your eye on the ball, run towards it, and keep the angle of your gaze as constant as you can. It is called the “gaze heuristic”. No cosines or tangents required.
Bottom-up models are, for want of a better world, “pragmatic”. They see the organisation as a constantly changing organism operating with incomplete, ambiguous information in an environment that is also constantly in flux. To survive, firms must respond dynamically and imaginatively to unpredictable, non-linear interactions in the environment which is constantly shape-shifting into new configurations in unexpected, and unexpectable, ways. For a pragmatist, practical control must be exercised at the points where the organisation interacts with its environment. A firm should have talented, empowered, well-equipped people — subject matter experts — to handle those interactions. Those in the central management function have a holistic view of the environment and can provide aspiration and tools to the subject matter experts, but real decision making is done by those experts at the edges, not the the management function in the middle.
Intellectually, the battle ought to have been won by the pragmatists long since (systems theory, complexity theory, even, for all its obsession with algorithms, evolutionary theory line up with pragmatism), but modernism keeps devising new ways of getting itself back in the game, and over the last twenty years has been winning. What with the giant strides of the information revolution, the forthcoming singularity, technological unemployment, the abolition of boom and bust in 2005, and the effective management and distribution of financial risks through sophisticated financial derivatives (amirite?), it is easy to be lulled into a sense of security.
Getting down amongst the elephants and turtles is not to everyone’s taste, but if you do it helps to see the planet on top of it more clearly. Here’s a distinction to draw: between things and interactions between things. Nouns versus verbs.
The illusion of permanence and the Ship of Theseus
We see that even many of the markers we treat as formal, fixed and permanent are really temporary: the Dread Pirate Roberts effect: the personnel comprising a corporation change over time. Likewise institutions: corporations merge, change business models, change locations, move into different markets. IBM of 2021 is very different from the IBM of 1971.
But the individuals may be fleeting and transitory; the residue they leave behind is not: The corporation’s devotion to the formal means that successive individuals become progressively constrained by their predecessors actions and decisions — even if, in the mean time the dynamic considerations that led to the decision no longer prevail. A rule that has been there for a long time, but that no-one knows the provenance of, acquires a kind of mystical quality. I think this is the inverse of the “Lindy effect”.
The illusion of significance
- ↑ The Selfish Gene, 2nd Ed., 95 — see it on Dawkins’ website.