During my presentation today using TM-derived METAL data, I picked Vermont as a demo to find closest pairs of points, and it came up with an NMP pair that looks to me like it should be lined up at I-91/US 2/US 2 Truck (US 2 is off from the other two). They differ by 0.000001 in both latitude and longitude so it gets the "looks intentional" treatment, but I don't see why it would be intentional.
