A new member of parliament stands up for the first time and asks a plain question: why do legislators have subsidised alcohol available before they vote? It is a reasonable thing to want answered. What comes back is not an answer. The benches grumble and talk over her, the reply recites unrelated policy, and it closes with a mocking line about a rival party. The question is not engaged. It is absorbed and dispersed, the way a defended system disperses anything it does not wish to take in.
Hold that scene next to a different promise being made in the same week. Advancing quantum computing, we are told, will let us solve our hardest problems: new cancer drugs, more food from less land. The two scenes seem unrelated, one a failure of manners, the other a frontier of capability. They are closer than they look. Both turn on the same confusion about what kind of difficulty we are actually facing.
Most of what gets called complex is not complex in any technical sense. It is contested. A complex problem, in the strict meaning, is one whose causation is hard to trace. A contested question is different in kind: it is morally plural, it asks us to choose between goods that cannot all be honoured at once, in the sense Isaiah Berlin meant by value pluralism, and it needs a legitimate decision rather than a better calculation. “Cure cancer” sounds like a problem for compute. But which cancers, for whom, at what cost, with what tolerance for risk, on whose consent: none of that is settled by processing power. The mathematics is the morally empty part. The value judgements are the whole point, and they do not yield to faster machines.
This matters because the word we choose decides who is allowed to speak. “Complex” sounds tractable and politically neutral; it hands the matter to experts and systems. “Contested” admits the thing needs open argument and the sharing of power. So labelling a contested question complex is not a description. It is a move. It launders a question about values into a question about capability, and quietly removes it from the room where it ought to be fought out. This is the same manoeuvre that productivity-is-the-wrong-word traces in the language of efficiency, and the same blindness that lets a board believe its data problem is technical when it is really about who gets to decide.
Technology makes the move easy. It lets you relocate a contested question into a domain where it looks like a neutral engineering task, then hand it to a vendor or a model. Consider health data placed in the keeping of a company whose ethical limits are set only by its contracts. It may genuinely shorten waiting times. But the question of whether it should hold the data at all is not complex. It is contested, and the answer most people would give is fairly clear. What is missing is not insight but the institutional nerve to set and enforce a limit. The value judgement does not disappear when it is handed to the system. It still gets made, only now invisibly, by whoever built the thing, accountable to no one who voted. A chamber, for all its theatre, can in principle be answered to. A model cannot.
Seen this way, the jeering chamber and the optimising technocrat are the same failure from opposite ends. Both are fleeing the work of legitimate collective judgement. One flees through performance and tribal noise; the other through outsourcing and optimisation. Neither does the actual job, which is to weigh competing goods in the open and own the result. The flight to compute is not the cure for the broken chamber. It is the same avoidance wearing a cleaner coat.
There is a reason everything starts to feel complex once this sets in. Complexity is what unresolved value-conflict feels like from inside a body that has lost the ability to resolve it. The problems have not grown harder. The capacity to hold a disagreement honestly and reach a binding, legitimate answer has thinned, and so familiar questions now present as impossibly tangled. “It’s complicated” becomes the standing excuse for not doing the thing that is merely difficult and uncomfortable. Some of this is honest accretion, the real tangle of legacy systems and procurement, a difficulty worth respecting. But that real plumbing is then stretched over the moral question to excuse it, and a defended system, as a frame that cannot see itself always will, mistakes its own avoidance for the difficulty of the world.
Machines can optimise an objective. They cannot set one, because setting it is a value judgement, and value judgements are the substance of politics, not a preamble to it. A society that has lost the capacity to deliberate does not recover it by buying compute. It only hides the judgements it is still making inside systems no one chose. The complexity is not in the world. It is in the gap between the moral richness of what we face and the impoverished institutional capacity we have left to meet it.
Further reading:
↳ Productivity is the wrong word
Garden notes
- Productivity is the wrong word: the same laundering of a value question into a technical one, in the language of efficiency.
- Your data architecture isn’t technical: a contested question dressed as an engineering task, and what that costs.
- The frame cannot see itself: why a defended body mistakes its own avoidance for the difficulty of the world.