Field notes: the ring road that took two counties three different names

I spent part of last month chasing a ring road that doesn't exist, at least not under one name.
Physically it's a single continuous loop, four lanes most of the way, built in two phases about a decade apart and finished by two different county road departments who clearly never got on a call together. For eleven miles it's Ridgeline Parkway. Then, at a county line marked by nothing more than a faded sign and a pavement seam you can feel in the car, it becomes County Route 9. Six miles later, inside a township that annexed its stretch sometime in the 90s, it's Old Mill Loop. The curb line never changes, the lane striping never changes, and nothing about the pavement itself marks a boundary, only the name does. Three names, two counties, one very confused attribute table.
This isn't rare. Every production team I've talked to has a version of it: the beltway that's "Route 9" to the county, "Perimeter Road" to the state DOT, and something else entirely to the township that annexed the last mile. County road naming conflicts like this usually started as a jurisdictional handoff nobody flagged for the mapping side. A road gets built in sections, each section gets named by whoever had the authority to name it at the time, and the geometry just keeps rolling through without caring that an administrative line crossed it.
The part that gets interesting is what it does downstream. A routing engine that trusts segment names for continuity will sometimes treat the handoff as three separate roads, which means a driver gets three turn-by-turn instructions for a maneuver that doesn't exist, because there's no maneuver, just a sign. Dispatch systems built on the same data inherit the confusion. And if your basemap pulls road names from different source layers at different update cycles, you can end up with a basemap attribute mismatch sitting right at the county line: one dataset still calling it Ridgeline Parkway a quarter-mile past where the sign changed, because nobody's edit queue flagged that specific stretch.
None of this is a geometry problem. The pavement is the pavement. A car on that road doesn't notice the county line, and neither does a satellite. What's actually broken is the naming layer sitting on top of a continuous physical feature, which is a different problem with a different fix, usually a human one: somebody has to decide which name wins, or whether all three get carried as alternate names on the same segment, or whether the segment itself needs splitting at the jurisdiction boundary regardless of what the signage says.
That's the piece worth separating out when you're triaging a network like this. Fixing who's right about the name is a production decision, often a political one between two county GIS offices that aren't in a hurry to agree. Fixing where the road sits on the ground, and catching it when a new interchange or realignment changes the geometry before anyone's resurvey schedule gets to it, is a different kind of problem, and it's the one that tends to sit untouched longest because nobody owns it between field surveys.
We built the quarterly vector diff behind Road Extraction for that second half: new and changed road geometry pulled from current imagery, delivered as a layer that merges into whatever basemap you're running, so the segments are current while your team argues about what to call them.
If your network has a loop road with an identity crisis of its own, we'd put money on the geometry being the easier half to fix.