Parametric-washing: Hype vs Reality

Greenwashing is when an organisation spends more on looking sustainable than on being sustainable - a coat of green over an unchanged business. Parametric-washing is the same move with a different paint: spending more effort on looking computational, data-driven and optimal than on genuinely letting evidence shape the design. The rendered district is captioned 'generated' and 'optimized', the deck is thick with dashboards and heat maps, and the overall message is that objectivity has replaced opinion. But strip the imagery away and you often find that the fundamental decisions - how dense, how tall, who the district is for, which land was included, which people were not counted - were made by developers, officials and designers for ordinary human reasons, and the computation was wheeled in afterwards to decorate and justify them.

Why does this happen? Partly because computation genuinely is impressive and clients want to be seen using it. Partly because a generative render is a superb sales tool - it looks like the future and it forecloses argument. And partly because 'the model found this' is a wonderfully convenient way to avoid owning a contested choice. The result is a systematic inflation of what computation actually contributed. A parameter was tuned; the slide says the city was optimized. A daylight analysis was run on a massing someone had already drawn; the slide implies the massing itself was derived from first principles. A single scenario was rendered; the language suggests thousands were searched and this one won.

The competent response is not cynicism - that would throw away the real value with the hype - but literacy. You learn to hold two questions at once: *what did computation genuinely do here* (often something real and useful - a network analysis, a solar study, a scenario comparison), and *what is the imagery claiming it did* (often far more). The gap between those two is the parametric-washing, and naming it precisely, without dismissing the genuine work underneath, is the skill. In India, where glossy smart-city and new-city visualisations circulate widely and carry real political weight, the ability to separate the substance from the render is not academic - it shapes whether the public can meaningfully contest what is being built in their name.

The danger of parametric-washing is not intellectual, it is aesthetic and emotional, and that is exactly why it is powerful. A high-resolution generative render bypasses argument and speaks to the eye. Its smooth curves and glowing metrics feel like the future has already been settled; disagreeing with it can feel like disagreeing with progress itself. This is aesthetic seduction, and it does specific work: it converts a proposal into an apparent fact, and a contestable political question into what looks like a solved technical one.

Several features amplify the effect. Photorealism lends unearned certainty - a thing rendered this convincingly feels already real, though nothing has been built and no one has been consulted. The dashboard borrows the authority of science: numbers to three decimals, green ticks, a walkability score of 87, all implying rigour whether or not the underlying model deserves it. The word 'optimal' is doing enormous rhetorical labour - it quietly asserts there is a single best answer, that the model found it, and that alternatives are therefore simply worse, when in reality 'optimal' is meaningless until you name the objective, and the objective was a human choice. The absence of people - or the presence of identical, cheerful, abstract figures - lets the image avoid the messy question of who actually lives here, who is displaced, who was never counted.

Recognising the seduction is a defence, not a cure - the images still work on you even once you can name why. So the practical discipline is to slow the image down: to refuse to let a render function as evidence, to insist that beauty and confidence are not the same as being right, and to remember that the more polished and certain a computational-urbanism claim looks, the more you should ask what work the polish is doing. The worst top-down 'drawn' cities were also sold with gorgeous perspectives that looked magnificent from the air and were desolate at street level. Parametric-washing is that same seduction, now automated and gilded with a false sheen of objectivity - and just as in need of a sceptical eye.

Here is the core practical skill of the lesson: a set of questions you put to any computational-urbanism claim, which reliably separate the genuine analysis from the parametric-washing. Ask them in order and the gloss dissolves, leaving whatever real substance is underneath visible and assessable.

What was actually optimized, and against what objective? 'Optimal' is empty until the objective function is named. Optimal for daylight? For developer yield? For car travel time? Each is a different city, and the choice among them is not technical. If the claim cannot state its objective plainly, there is nothing to evaluate. Who chose the objective and the weights? Multi-objective results depend entirely on how competing goals were traded off, and those weights encode someone's priorities - almost always the commissioner's. What did the computation actually decide, versus decorate? Distinguish the parts genuinely derived by analysis (a network measure, a solar study) from the parts a human drew and the computer merely rendered or scored. What is not in the model? This is the sharpest question. Every model has a boundary, and the most important things - community, memory, justice, the informal settlement on the site, the livelihoods that do not appear in the land record - usually sit outside it. What data was used, how good was it, and whom does it miss? Data is never neutral; it over-counts the legible and formal and under-counts the informal and poor. What alternatives were suppressed? A single confident render hides the space of options and the trade-offs a democratic process is entitled to weigh.

None of these questions requires you to be a computational expert; they require you to be a critical reader. Together they convert a shiny, unfalsifiable claim of optimality into a set of specific, answerable, contestable statements - which is exactly what a binding urban decision needs before it goes to the planning authority, the affected communities and the public. The goal is never to debunk computation, but to hold it to the honest standard it can genuinely meet: a tool that informs and opens up debate, never one that closes it with a picture.

The deepest form of parametric-washing is not the render but the metric itself, because a number looks neutral while smuggling a whole worldview inside it. Take 'walkability score', which sounds like an objective property of a place. But someone chose what walkability means - proximity to shops, or safety, or shade, or the presence of a crowd - and chose how to weight those, and chose the data that stands in for each. A score of 87 presents all of that buried judgement as a single objective fact. The same is true of a 'liveability index', an 'efficiency' measure, or a 'density optimum': each is a compressed argument about what matters, wearing the costume of a measurement. To read computational urbanism critically is, above all, to unfold metrics back into the value choices they encode - to ask, every time, whose idea of walkable, whose efficiency, optimal for whom.

This is where the module's through-line first appears clearly. Computation optimizes the *measurable*; the choice of what to measure is where the values hide; and parametric-washing is the art of hiding those values so well that a political choice arrives looking like a scientific result. The informal market that a 'liveability' model scores as blight, the slow street a 'traffic-efficiency' model wants to widen, the tight-knit basti a 'density optimum' would clear - each is a value judgement disguised as an optimum, and each is exactly the kind of decision that must stay democratic.

The honest alternative is not to abandon metrics but to use them transparently. State the objective and the weights in plain language. Show the alternatives and the trade-offs, not one triumphant render. Name what the model cannot see and treat those absences as first-order, not as rounding error. Present computation as evidence that informs a public, political decision - deferring the binding planning and land-use choices to the planning authority, the participatory process, the affected communities and the governing law - never as an oracle that ends the argument. An urbanist who can read a claim this way, and who insists on presenting their own work this honestly, is the antidote to parametric-washing: someone who keeps computation genuinely useful precisely by refusing to let it pretend to be more than it is.