Every planning committee meeting about housing eventually arrives at the same exchange. Someone says the scheme is too dense. Someone else replies with a figure — dwellings per hectare, floor space index — that demonstrates it is comparable to a district everybody likes. Neither person changes their mind, because they are not discussing the same thing.

The objection is not usually to the quantity of housing. It is to a set of experiences that tend to accompany it: overlooking, wind, shadow, noise at the wrong hour, a ground floor that presents a blank wall for forty metres. Those are design outcomes, not arithmetic ones. It is entirely possible to build the objectionable version at a modest density, and it is possible to build a great deal of housing that nobody objects to at all.

The number is a poor proxy

Floor space index on its own tells you almost nothing about what a place is like. The Dutch work on this — the Spacemate method developed at Delft — makes the point precisely: you need floor space index and ground space index together, because the same quantity of floorspace spread across a large footprint at four storeys and concentrated on a small footprint at sixteen produces two entirely different environments at the same headline figure.

The Amsterdam canal belt runs at a floor space index that would alarm most committees if it were proposed today. Barcelona's Eixample is denser still. Neither is experienced as oppressive, because the density is delivered in mid-rise perimeter blocks with generous streets, continuous active ground floors, and a courtyard behind. The tower-in-a-car-park that gets proposed to "achieve density" frequently achieves less of it, while feeling like considerably more.

This matters practically. If you accept that the number is a weak proxy, the argument in the room changes from how much to in what form — which is a conversation an architect can actually win.

What makes density feel soft

Six or seven storeys rather than fifteen. It is high enough for real density and low enough that the street keeps its sky, and it is the height at which people on a balcony are still connected to what is happening below.

A perimeter block with a courtyard, so that every home has a quiet side. Dual aspect is the mechanism: one face to the city, one to something calm. It is difficult to make a home feel crowded when half of it looks at a garden.

A threshold at the ground. Front doors on the street, raised half a metre, with a metre and a half of defensible planting in front. It costs frontage and it transforms how a building is read from the pavement — the difference between housing that faces the city and housing that turns its back on it.

Variety in the section rather than the elevation. Setbacks at the upper floors, a taller ground storey, occasional double-height openings through the block. These do more for perceived scale than any amount of facade articulation, and they do it honestly.

The trade that makes it work

Softness costs area, and the area has to come from somewhere.

At Linnaeus Courtyard we took it from the apartments. The thirty-four homes are around fifteen percent smaller than the market norm for their bedroom count, and the recovered floor area was collected into three shared rooms: a kitchen and dining room opening onto the courtyard, a workshop in the basement, and two guest rooms at roof level that residents book.

The residents had agreed this principle before we were appointed, which made the project unusually easy. Our contribution was deciding how much and where — and the where turned out to matter more than the how much. The shared kitchen sits on the only route from the street to the stair, so every resident passes it several times a day. Shared facilities that require a decision to visit do not get used. Shared facilities you walk through do.

The guest rooms are booked around seventy percent of nights, which is the clearest evidence that the trade worked: residents are no longer keeping a spare bedroom empty for eleven months a year in order to house visitors for one.

Where this argument runs out

I do not want to overstate the case. Perimeter blocks and mid-rise will not deliver the densities required on every site, and there are locations — directly above transport interchanges, on very small plots — where towers are the right answer and the objections have to be argued with rather than designed around.

Nor is the courtyard block automatically good. A courtyard that is too small becomes a light well and a noise chamber; below about twenty-five metres across at six storeys, it stops working as amenity and starts working as a shaft. A perimeter block with a dead ground floor is just a wall with homes on top.

And there is a real cost. Mid-rise perimeter development is generally less efficient per square metre of site than a tower with a podium, and the shared facilities at Linnaeus are floor area that generates no rent. Someone pays for that, and in this case it was residents accepting smaller homes with their eyes open, which is not an option available on every project.

The useful reframing

When a scheme is called too dense, the productive response is rarely to argue about the figure. It is to ask which of the actual complaints is driving it — the shadow, the overlooking, the blank frontage, the wind at the corner — and to establish whether the density is causing it or the form is.

In our experience it is nearly always the form. Which is good news, because form is the part we can change.