One of the problems of an arcology, which Hans seems to be overlooking, is that 3 dimensional cities allow far more interaction per person than 2D cities.
Also, crime rates scale with population density; violent crimes and gangs rise proportionally with density and inversely with average income. Average education doesn't seem to factor directly (tho' it does link to average income with a contributory relationship). Also, studies of social interaction show that the same 2D density in a 3D space has about the same social effects as that same density in a 2D space; Areas of Tokyo with a density equal to that of downtown NYC have a similar rate of both crime and social distancing behaviors. (I can't vouch for how they measured social distancing, as I read the paper 4 years ago during my MAEd program; I recall it measured the number of social pleasantries per hour in a given area, but not how counted.)
It's interesting to note that the densest pre-industrial populations also were among the most socially formal; Byzantium, Feudal Japan, Urban Rome... Even the Imperial City of China was said to be socially rigid in form.
The idea of social rigidity in custom is, in point of fact, one of those areas where lots of North Americans find the whole idea of formal interaction abhorrent; US cities, rather than formal interactions, tend towards non-interaction; Japanese tend to fall to rigid forms of limited interaction and "false privacy" (People willfully ignoring individuals until the correct social forms are followed).
The social rules needed where a 1km movement can bring one into contact with 10 million people are far more stringent than those where it can only bring 10,000, and those more so than when it's 500 people. As someone who's lived in range of about 5k, and now am in an area where it's about 2K, the social pleasantries are in fact far more common; it's only 16km distance, but the social environment is VERY different; 3km away, it drops even more, and people are FAR more friendly there.
So I would expect that Fenrock, at 33M on 6km^2 (in 4.5km^2) would be 5M per km^2... 100 times that of Manilla, the most dense on the list on Wikipedia, and 100 times that of Mumbai... would be stiflingly formal. You can run into too many people. I would expect clear social indicators, and clear restrictions on who goes where. Even if not properly uniforms, I would expect clear social class dress modes. All things that make it horrific to me. And, from the corner, you've got 5 million people within a kilometer. Oh, and that's only 136.36 cubic meters per person... INCLUDING industrial space. Assuming a nice 3m ceiling, that's 45m^2 per person, or 6.75x6.75x3m each.
That's pretty damned cramped there, Hans. 22x22 feet each. That's my living room per person, counting work and accessway spaces. Using a general rate from MT, 10% will be LS and power to run it; and half the quarters space will be outside the quarters, that's 20m^2 per person in quarters, and 20m^2 per person in commons, presuming only LS is accounted for; if there's industrial space, it gets MUCH worse.