Coverage from National Geographic [1] (discussed at the time on HN [2]) provides additional detail, including that researchers chose the arbitrary number of 50 to constrain the number of cohesive regions.
At the time, I wrote [3] the following which still applies:
The arbitrary choice of 50 regions results in some fairly generous interpretations of 'regions'.
Namely, this method produces two kinds of regions: ones that are clustered around a single metropolitan area that has a higher-than-average commuter pull around its hinterland, and ones that are slightly too far from a major metro and lump together a chain of areas that form a loose "commuter continuum" of areas where commuters have two or more equally plausible choices to commuter to -- this chain then hops along interstate highways, grouping unrelated towns across hundreds of miles into the same region.
Take the one that they call 'Corn Belt', encompassing Des Moines, Davenport/Moline, and every single city in Illinois on I-74 (including Urbana-Champaign!). In no universe do people commute hundreds of miles along I-74. In truth, this is a polycentric area with many distinct loci which attract their own, distinct set of commuters: Des Moines/Ames; Cedar Rapids/Iowa City; Quad Cities; Peoria; Urbana-Champaign. The "cohesiveness" of the region exists solely in contrast to its neighbors: that more populous metros on the region's fringes are all too far out of sane commuting distance.
Or the Appalachians: Roanoke in Virginia forms a vital locus for much of west-central Virginia and yet is swallowed into a much larger region including Greensboro, Raleigh, and Wilmington(!), because people living midway between Roanoke and Greensboro (like in Danville, VA, or Martinsville, VA), have two equally plausible larger job centers to commute to. The same can be said about Charleston, WV, which forms a job hub in West Virginia, but is grouped together with Columbus and most of Ohio because smaller towns along the Ohio River are roughly equidistant from either.
It helps me to look at each of their regions and think "people rarely commute outside of their region", rather than the more natural interpretation of "people commute along their region".
Even in Boston/New England, it’s not really a single ring from the city—even if you include its suburbs and exurbs where many companies are actually located. Almost no one is commuting on a regular basis from Augusta Maine or Burlington Vermont to the Boston area. It does somewhat make sense as a region but you could easily subdivide it much further. Assuming the exercise makes some sense you probably have to arbitrarily choose some level of granularity.
At the time, I wrote [3] the following which still applies:
The arbitrary choice of 50 regions results in some fairly generous interpretations of 'regions'.
Namely, this method produces two kinds of regions: ones that are clustered around a single metropolitan area that has a higher-than-average commuter pull around its hinterland, and ones that are slightly too far from a major metro and lump together a chain of areas that form a loose "commuter continuum" of areas where commuters have two or more equally plausible choices to commuter to -- this chain then hops along interstate highways, grouping unrelated towns across hundreds of miles into the same region.
Take the one that they call 'Corn Belt', encompassing Des Moines, Davenport/Moline, and every single city in Illinois on I-74 (including Urbana-Champaign!). In no universe do people commute hundreds of miles along I-74. In truth, this is a polycentric area with many distinct loci which attract their own, distinct set of commuters: Des Moines/Ames; Cedar Rapids/Iowa City; Quad Cities; Peoria; Urbana-Champaign. The "cohesiveness" of the region exists solely in contrast to its neighbors: that more populous metros on the region's fringes are all too far out of sane commuting distance.
Or the Appalachians: Roanoke in Virginia forms a vital locus for much of west-central Virginia and yet is swallowed into a much larger region including Greensboro, Raleigh, and Wilmington(!), because people living midway between Roanoke and Greensboro (like in Danville, VA, or Martinsville, VA), have two equally plausible larger job centers to commute to. The same can be said about Charleston, WV, which forms a job hub in West Virginia, but is grouped together with Columbus and most of Ohio because smaller towns along the Ohio River are roughly equidistant from either.
It helps me to look at each of their regions and think "people rarely commute outside of their region", rather than the more natural interpretation of "people commute along their region".
[1] https://news.nationalgeographic.com/2016/11/us-commutes-reve... [2] https://news.ycombinator.com/item?id=13100209 [3] https://news.ycombinator.com/item?id=13101206