Okay, then about the "quaint little farms"... Labor-intensive permaculture, in addition to less quantifiable things like increasing biodiversity (and thereby nutrition, maybe?) and reducing a sense of alienation and even shortening supply chains, can produce enough food for maybe 10 people per hectare, or 1000 people per km². We can use this to estimate how large a food-autonomous city can be.

If everybody working the farms rides a bike there, a maximum of 10 km each way, there is a radius of 10 km around the city that can be worked. If the urban environment holds 80 people per ha, and each km² urban environment requires 8 km² of farmland to support it, we get...

(insert algebra sounds)

...about 145k population, in a circle of land that is 25 km across, but the city itself is 5 km across.

This is with side-by-side American standard city lots (1/6 acre, plus street frontage) that I'm doubling the buildings on to fit 2 one-story buildings within. With townhouses, you could easily double this density, and with apartment buildings that are still small enough to build with appropriate tech and to climb the stairs after a day at work, you could quadruple the density. If you put everyone in a narrow arcology tower with the same 10km radius, you could fit 314k people in the citytower.

Edit: With the sparser model (1k per sq km), you still can have all of humanity fitting into about 20% of the world's land area, including food production but not including fuel and other resources.