Mixing Uses & Programme

Streets, blocks and massing are the container; programme is what fills it with life. Programme means the *uses* a city holds - living, working, buying, selling, making, learning, worshipping, eating, meeting - and, more than any single use, the way they are combined. A district that is all one thing - all housing, all offices, all shopping - is a machine for one activity, busy at one hour and dead the rest. A district that mixes uses is alive across the whole day: the offices fill the lunch places, the homes keep the streets watched at night, the shops serve both, and the sheer variety of reasons to be there produces the density of life we recognise as urban.

The classic argument is Jane Jacobs's, made against the single-use planning of her time. A great city street, she observed, needs a fine mixture of uses to generate safety (eyes on the street at all hours, because people are there for many reasons around the clock), convenience (daily needs within a walk), economic vitality (a mix of enterprises supporting each other, room for the small and the new), and the unplanned encounter that is the deepest pleasure and the real productivity of cities. Mono-functional zones fail on all four: the office park dead at night, the dormitory suburb dead by day, the mall that sterilises everything into one managed use.

Crucially, mixing has a *grain*, exactly as the fabric does. Coarse mixed use puts a housing block here, an office block there, a mall beyond - technically "mixed" at the district scale but mono-functional at the scale you actually walk. Fine-grained mixed use weaves the uses together at the finest scale - different uses side by side along a street and stacked vertically in a single building (the shop at street level, workshop behind, homes above), so that every hundred metres holds a dozen kinds of life. Fine-grained mixture is the real thing, and it is the signature of the world's most loved streets - and of the traditional Indian bazaar and mohalla, where home, trade, craft and worship interleave at a grain few planned cities achieve. Programme, finely mixed, is the life of the city; and as we will see, it is precisely the quality that computation finds hardest to hold.

To see why mixture is fragile, look at what the twentieth century did to it. Modern planning, in much of the world, was built on single-use zoning: the legal division of the city into zones each permitting only one use - residential here, commercial there, industrial elsewhere - connected by roads. The intentions were partly sound (separating genuinely noxious industry from homes) but the result, applied wholesale, was catastrophic for urban life: it made the fine-grained mixture that keeps cities alive *illegal*. You could no longer put the workshop behind the shop and the home above; the corner store in the residential zone became a zoning violation; the city was sorted into mono-functional pods that forced everyone to drive between the separate zones of their lives, drained each zone of life for most of the day, and dissolved the walkable, mixed street into traffic.

The reaction - Jacobs's critique, and decades of urbanist argument since - has been to rehabilitate mixed use as a central goal of good planning. But here is the subtlety that matters for this course: not all "mixed use" is equal, and the coarse version can quietly betray the fine one. A great deal of what is marketed as mixed use is *coarse* - a tower of flats over a podium of chain retail, a "mixed-use district" that is really adjacent single-use blocks - and it delivers little of the life the fine grain produces. The mixture that actually makes a place live is fine-grained: many different uses, many of them small and locally owned, interwoven at the scale of the individual plot and floor, able to change one door at a time. That fineness is hard to mandate and easy to lose. It depends on fine grain in the fabric (Lesson 4.2), on small units and low thresholds that let a small enterprise start, on rules permissive enough to allow a home to become a workshop to become a cafe - and it is mostly *emergent*, the product of thousands of small decisions no plan made. Zoning can *permit* fine mixture, or forbid it; what it cannot do is *manufacture* it. And a computational model, as the next section shows, is far better at allocating coarse zones than at growing fine, emergent, living mixture.

Now the crux of the whole module. The fine-grained mixture that makes a great city is, of everything in urban form, the single hardest thing for computation to handle - and understanding *why* is understanding the optimization trap at its purest.

First, the fine mixture is emergent, not allocated. It arises from countless small, self-interested, unpredictable human decisions over time - this tailor opening here, that tea stall surviving because of the office across the road, a home quietly becoming a workshop. It is the product of a complex adaptive system, not a plan, and complex emergent order is exactly what a model that *allocates* uses cannot generate. A computer can distribute "40% residential, 35% commercial, 25% office" across a plan; it cannot conjure the living, self-organised, ever-shifting mixture that emerges door by door - because that mixture was never allocated by anyone.

Second, the categories are too coarse to hold it. A model works in land-use categories - residential, commercial, mixed - but the reality is a single building that is home and workshop and shop and shrine, changing by the hour and the season. That fineness does not fit the categories, so the model flattens it: the vibrant hybrid becomes "mixed / other", a hundred kinds of life swallowed by one tidy label - and what the model cannot represent, the plan tends to erase.

Third, and deepest, the value of the mixture is unmeasurable. You can measure a jobs-housing ratio, a use-percentage split, a walk-score to amenities - and a scheme can hit all of them while being dead, because what the fine mixture actually produces is safety, spontaneity, opportunity, belonging and life, none of which is in the metrics. So an optimizer told to "provide mixed use" will satisfy the measurable proxy - the right percentages, the mixed-use label, the amenity counts - and deliver coarse, lifeless, allocated mixture, having hit every target while missing the entire point. This is the optimization trap in its most concentrated form: the fine mixture that makes a great city is exactly what is hardest to compute, easiest to reduce to a hollow proxy, and easiest to zone away - and a naive generative masterplan, left to it, will reliably produce a perfectly "mixed" city that is dead to live in.

So what is the competent stance? Not to abandon computation - it has honest uses here too - but to know exactly what it can and cannot do, and to defend the fine mixture deliberately, because nothing else will.

Computation *can* help with programme in real ways: it can *analyse* an existing city's mixture and reveal where uses are dangerously segregated or where a dormitory zone has no daily life; it can measure walking access to amenities and expose neighbourhoods starved of shops, schools or jobs; it can test whether a plan's coarse allocations are even internally coherent; and it can model the *conditions* that tend to allow fine mixture - fine grain, small units, permissive rules, connectivity - so you design the soil in which mixture can grow. That is genuine and useful. What it cannot do is *generate* the fine, emergent, living mixture itself, or tell you it is present from the percentages. So the discipline is: use computation to analyse mixture and to create the conditions for it, and never mistake a good land-use metric for a living place.

The Indian dimension makes this urgent. So much of Indian urban life *is* fine-grained mixed use - the bazaar where trade, craft, home and worship interleave at a grain few planned cities on earth achieve, much of it informal. This is world-class urbanism, and it is exactly what the model cannot see: the informal shop, the home-based enterprise, the hawker, the live-work hybrid do not fit the land-use categories, so a naive computational masterplan can zone them out of existence - imposing single-use order on some of the most alive fabric anywhere, and destroying the livelihoods and vitality that the fine mixture sustained. The twentieth-century zoning mistake, now automatable, threatens precisely India's greatest urban asset.

Hence the course's stance, at its sharpest here. Treat the fine mixture as something to *protect and enable*, not allocate - defend permissive, mixture-friendly rules and the fine grain that carries them, and defend the informal mix the model erases. Use computation to analyse and to prepare the ground, never to manufacture life it cannot make. And keep the binding choices - what uses to permit, how to zone, whether to protect or clear a living mixed neighbourhood - with the planning authority, the affected communities and the statutory process (in India, the master-plan and development-plan process, the applicable development-control regulations and the National Building Code of India), because the mixture that makes a city live is a human, social and political achievement, not an output the algorithm can compute.