Megaregional Change
Just as transportation networks can create a new megaregion, so can they work to destroy one. Back in 1910, the golden age of manufacturing centralization, the US had a megaregion, namely the core formed by the quadrilateral whose vertices are Boston, Milwaukee, St. Louis, and Baltimore. Outside this quadrilateral, there were almost no large industrial cities, and the main exceptions were either right outside and represented extensions (Kansas City, the Twin Cities) or in California (San Francisco), representing a separate manufacturing network.
The formation of the Northeast megaregion was really a replacement of the old megaregion with a different system. The US used to have a general core area and a periphery. It was never perfect – the North’s dominance of the South was only complete after city regions developed elsewhere, so there was never a total dominance of the manufacturing belt. But it was still far richer than the periphery, and each city had a separate industry to specialize in. In contrast, what we see today is different: there are still core and periphery regions, but instead of one giant core, we have a spikier core, spread relatively evenly nationwide. The Northeast’s four major cities are all core, but there are core cities everywhere, so in a sense we get a core and periphery in each major region of the US. Megaregions occur along clusters of core cities surrounded by their nearby dependencies.
Just as the new, smaller megaregions are bound by high-speed rail or by regular rail and freeway networks, the old one was bound by the emerging freight rail network. Lower transportation costs made it easy to manufacture everything in one place and ship it elsewhere. Once a core-periphery model sets in, it takes a large difference in income to make it worthwhile to start locating factories elsewhere.
I forget which book I got this from (I think The Bottom Billion by Paul Collier), but it took until the postwar period for first-third world income disparity to grow to the point that American companies started offshoring factories. Until then, what is now the first world consistently had higher economic growth than what is now the third world, even absent colonial relationships. Of course within a country the costs of moving factories are lower and so the wealth disparity required to change the core-periphery dynamic is different, but the principle is similar. For decades, there wasn’t much city growth in the South. California grew very quickly, since it had gold and then oil and needed to manufacture its own goods since transportation costs across the Rockies were too high for it to import everything from the East. But, as Jane Jacobs quotes Henry Grady on the situation of the region around Atlanta:
I attended a funeral once in Pickens county in my State. They buried him in the midst of a marble quarry: they cut through solid marble to make his grave; and yet a little tombstone they put above him was from Vermont. They buried him in the heart of a pine forest, and yet the pine coffin was imported from Cincinnati. They buried him within touch of an iron mine, and yet the nails in his coffin and the iron in the shovel that dug his grave were imported from Pittsburg. They buried him by the side of the best sheep-grazing country on the earth, and yet the wool in the coffin bands and the coffin bands themselves were brought from the North. The South didn’t furnish a thing on earth for that funeral but the corpse and the hole in the ground.
Of course, something similar to what happened in the US between about 1910 and the 1960s is happening on a global scale: the dominance of the US, Europe, and Japan is in decline, and third-world economic growth is now consistently above first-world growth.
Paul Krugman suggests that this change in agglomeration patterns comes from further income growth, noting that China today looks a lot like the US of 1910. While he talks about it in terms of city specialization, this is equally true in terms of the location of the core. China’s core today is a contiguous region stretching from Guangdong to just north of Shanghai, plus Beijing. It’s entirely possible that in fifty years it will look different – that the small cities between the megaregions around Shanghai and Guangdong will rust while new cores will develop around large interior cities.
Pedestrian Observations 
The lesson I take from this the USA’s northeast re-invented itself (and remained prosperous) following industry’s exit back then…. and therefore so should the (whole) USA do so today.
Mind you, that’s a simplifcation. New York was prospoerous in the 1950s and today, but it went through a bad patch after Things Changed around the 1960s.
The Northeastern Megaregion began to form in the 1830s. They were busily building railroads so you could go from Washington DC to Boston. Before that you had to watch the shipping news in the local paper and hope to catch a sailing ship that was headed your way. Built the Long Island Railroad because learned engineers told them it wouldn’t be possible to build a railroad through Connecticut. Took them until the 1880s to make it possible to do without getting on a ferry. By the 1880s they were building competing railroads to capture the volume of business. By the 1890s they are building tunnels in Baltimore with electric trains and running luxury trains over them. In 1917 the Hells Gate Bridge opens and gives the NEC as we know it.
Part of the reason there’s Senate was counterbalance to the states with the big cities along the NEC…
It wasn’t really just the Northeast, though. Early on, the in-state transportation projects, i.e. the Main Line of Public Works and the Erie Canal, were more important. If you look at American accents, they even have traces of which East Coast city the region was settled from: the inland northern accent (from Syracuse to Milwaukee) is areas settled from the Erie Canal, the northern midlands accent is areas settled from Philadelphia, and so on. By the time there were tight north-south connections in the Northeast, there were equally tight connections going as far west as Chicago and St. Louis.
The formation of the Northeast megaregion was about something different. It was about the decline of the east-west connections and strengthening of the north-south ones, along the coast. Part of it is that the railroads kept improving, yes. Another part is that between the East Coast and Chicago, the cities all went into terminal decline. In 1920, Rochester was a medium-sized city in a megaregion with varied connections to other cities; today, it’s a rump colony of New York. A colony that gets subsidies, as all colonies do, but doesn’t really benefit from the arrangement. (Britain invested a lot of money in the colonies while it starved them to death.)
On a nitpicky note, the Senate was a counterbalance to Virginia with its large population, not to the North, which was poorer than the South in 1789. Virginia wanted both houses proportional to population, Connecticut wanted one-state-one-vote representation. If only Virginia had gotten its wish…
The various Post Roads are pre-Revolutionary. They are doing overnight service between Philadelphia and New York before the Revolution too. They aren’t doing that to exchange wedding invitations, birth announcements and Christmas Cards.
Almost as soon as they are laying the tracks for the first railroads they are finagling ways to lay a second set for a competing railroad. They aren’t doing that for the sheer joy of experimenting with steam engines, they expect there to be enough traffic to make money. They carve through the Palisades so the train from Philadelphia can get to the ferry landings in Jersey City. Ferry from Jersey City to Brooklyn and connection to the Long Island Railroad – which was built to get people from New York and Brooklyn to Boston via the North Fork and a ferry to Rhode Island.
Alon, I think you can’t analyze the urban morphology of megaregions ignoring the urban x rural balance that changed so drastically in the period.
In a process that was and still is (China) replicated, areas with a high prevalence of rural population have different dynamics that delay the concentration of people in bigger cities.
Moreover, some advantages to centralization that existed in 1910 are negligible today, like having all departments of a company concentrated within a limited area to facilitate interactions and reduce internal transaction costs.