I like the idea of moving towards Caves of Steel, but with parks on top and atrium areas where streets currently are.

It might also be possible to use a heat-stack effect to draw the air out of the city as a whole.

If we could adopt a combination of subways, dreaded e-scooters, bikes and golf cards for elderly and people with disabilities we could make this happen. Some European cities have huge pedestrian zones and it works.

> But it's extremely expensive and disruptive to go rebuilding existing cities, and all the construction itself would have an environmental cost. Surely air conditioning is cheaper?

It's at best a cheaper up-front investment. And while it is one thing to think about the limits of retrofitting old buildings, that is no excuse to not do better with newly constructed ones.

First thing to do is stop making the problem worse.

Last I checked we are still building our cities. Mine is in the process of knocking down all the one story commercial space and replacing it with six story mixed use buildings.

Increasing density in urban areas by building up is the right thing to do, compared to expanding urban areas which would require more paving, more roads, and more inefficient use of land.

It's the way we design those new buildings that's important.

Increased density means fewer low density buildings and paved surfaces elsewhere. Urbanization is a massive net positive for the environment.

It's a lot easier to insulate multi story buildings. Volume grows faster than surface area.

You're just trading off energy spent cooling denser cities vs. increased transportation in less dense cities.

Why is that bad? people complain about the enormous costs of accommodation, to which the main solution is simply to build more.

Your proposed alternative is to increase urban sprawl?

What? No. My point is that each new building is a chance to do better. Claiming it’s too late because we have built up areas is helplessness.

If you double the size of a town and build the new half well you solve more than half of the problem.