My first city was Conan Doyle’s London, in the company of Holmes and Watson. My mother gave me a two-volume omnibus edition when I was 10. London was a vast, cozy, populous mechanism, a com­forting clockwork. Foreigners and criminals served as spices, highlighting the assumed orderliness and safety of the Empire’s capital (assuming one were sufficiently comfortably placed in society, and in Doyle one tended to be).

I lived in rural southwestern Virginia, the nearest cities several hours away and those were smallish cities. Relatively little of what I saw on television conveyed much sense of urban reality, perhaps because it was still inherently difficult to film in large cities. Except for Los Angeles, and I saw a lot of that, and Los Angeles never did become much a part of my imagination’s map of cities.

I reverse-engineered a concept of urban life from Doyle’s rich and intriguing (and cozy) construct. I walked through my hometown, imagining it a city. What I was imagining, I now see, was an increase not in size but in number of choices.

Cities afforded more choices than small towns, and constantly, by increasing the number and randomization of potential human and cultural contacts. Cities were vast, multilayered engines of choice, peopled primarily with strangers.

You never know whom you might meet in the city. In a small town, you’re less likely to encounter people or things or situations you haven’t encountered previously. These people or things or situations may be wonderful or horrible, in either city or town, but cities have the numbers, the turnover. To a writer of fiction, this is extremely handy, a city being able, more or less believably, to mask excessive coincidence, producing, as Doyle taught me, whatever the narrative might require.

Should the populous mechanism of the fictive city fail to produce phenomena of sufficient weirdness, our literature of the fantastic often turns, quite reflexively, to dead cities, our most profoundly and mysteriously haunted artifacts.

Many deserted cities probably never were engines of choice. To stand in the vast plaza of the pre-Columbian Monte Albán, for instance, is to know that Monte Albán was about decreasing choice, narrowing it. Monte Albán was a control machine, an acoustically perfect environment with magnificent lines of sight: a theater of power. We don’t know why Monte Albán was as abruptly deserted as it may have been. Perhaps the show failed, finally, to come off, and no other was available, or possible, within that inflexible, uni-purposed structure.

That’s the danger of choice reduction, of top-down control. And the curse of gated attractions, the ultimate fate of every Disneyland: you can’t repurpose a theme park. Cities, to survive, must be capable of extended fugues of retrofitting. Only the most pubescent of cities have never witnessed, to whatever extent, their own ruins. Berlin has, Rome has, London has, Tokyo has, New York has. Relative ruin, relative desertion, is a common stage of complex and necessary urban growth. Successful (which is to say, ongoing) cities are built up in a lacquering of countless layers: of lives, of choices encountered and made.

The most crucial layers are those of various essential technologies, all of which must in some sense be present and functional for a city to endure. We didn’t begin to build cities until we could secure adequate supplies of food, which generally meant growing and storing it. Growth beyond a certain size requires mastery of sewage-disposal technologies. The city evolves as a pyramid of technologies, some essential, others incidental.

Cities can be at their experientially richest during periods of relative disjunction. Cities that are somewhat dysfunctional in one sense can be brilliantly functional in others. The city you want, as a young creative person, is partially ruined, marked by areas semimoribund in real estate values. Low rents, minimal policing, casual welding allowed on sidewalks. Manhattan in the 1970s, a place and time people my age now regard with mixed nostalgia, was fraught with ruins, with buildings abandoned, nights lit by insurance fires. On first observing this, in 1979, I suggested, half-seriously, that the Japanese be allowed to sort the place out, given their way with urban real estate. New Yorkers smirked at my bumpkin naïveté, knowing the Bowery would always be the Bowery.

Today the Bowery is nothing like the Bowery.

Cities can do that, reversing out of disjunction, throwing themselves into a different gear. Although in doing so, they run the risk of Disneylanding themselves, of building themselves too permanently into a given day’s vision of what they should be. Paris feels that way to me, lovely as it is, with New York and London hurrying to catch up.

Meanwhile, though, some of the world’s largest human settlements are now not only places where one can weld on the sidewalk but places that have bypassed many of the ways in which Europeans and North Americans have assumed cities necessarily need to grow: Rio, Mumbai, Nairobi, Istanbul, Mexico City.... Vast squatter conurbs, semi-neo-Medieval in their structure and conditions. The future will emerge from such cities as surely as it will emerge from the Disneylanded capitals of an Old World that now includes North America.

The future of cities will consist of two different modalities combined within the ageographical and largely unrecognized meta city that is the Internet.

As a boy, I took myself away to cities as quickly as I could and have lived in them ever since. When I travel now, I travel mainly to cities, and I tend to return to those I know, taking a deepening pleasure in the serial experience. The idea of visiting a fascinating city only once saddens me, and I seldom leave a city I’ve come to know without wondering if I’ll see it again. But in our ageographical existence, I am never entirely not in London, entirely not in Tokyo.

We all inhabit the meta city now, regardless of physical address.