What is a stub road, and why does your basemap need to track it?

A stub road is a short segment of pavement that ends abruptly, built that way on purpose because the plat was drawn with a future connection in mind. Walk to the end of one and you'll usually find a turnaround, a barricade, or just gravel and weeds where the asphalt gives up. The right-of-way keeps going past the curb, though. It's reserved, even though nothing's built on it yet.
For a basemap production lead, that distinction matters more than it looks. A stub road is a temporary dead end in the routing sense, not a permanent one. It's a placeholder for a connection that the county, the developer, or the next phase of a subdivision will eventually finish. Treat it like a true cul-de-sac in your network and you'll route traffic into a wall the day the connection opens. Miss that it exists at all and you'll leave a gap between two developments that should already show as linked.
Why subdivision phasing creates so many of these
Most stub roads trace back to how subdivisions get built. A developer rarely builds an entire network at once. Phase one goes in with its own loop and one or two short stubs pointing toward the property line, toward where phase two will eventually tie in. Sometimes phase two gets built the following construction season. Sometimes the back parcel gets sold to someone else entirely, and the stub sits there for eight years before anyone touches it.
The plat itself often already shows the connection: a dashed line labeled "future street," or a reserved right-of-way with no centerline symbol yet. That's a platted but unbuilt road segment, and it's distinct from a stub road in one respect. The plat shows intent on paper. The stub shows a few hundred feet of built pavement reaching toward that intent. Agencies that pull plats as a source will sometimes capture the planned connection before it's graded, let alone paved, which is useful but creates its own reconciliation problem when the imagery and the plat disagree about what's on the ground.
What to watch for when you're maintaining the graph
A stub road worth flagging usually has a few tells. The pavement width matches the rest of the subdivision instead of narrowing down to driveway width. There's a temporary turnaround, a hammerhead or a paved circle, sized for emergency vehicles rather than parked cars. The right-of-way width in the parcel data extends well past where the asphalt stops. And there's often a neighboring parcel, platted or just vacant, sitting exactly where the extension would go.
None of that tells you when the connection actually gets built. That's the part that matters for a routing graph, and it's also the part a one-time capture misses. A stub road driven or digitized once goes into the network as a dead end and stays that way until somebody notices the pavement's moved. A dead-end road with a likely future connection needs a different edit cadence than the rest of the network, because the whole reason it exists is that it's about to change.
Road Extraction pulls a vector diff of the road network from high-res imagery on a quarterly cycle, which is roughly the pace subdivision construction moves at. A stub that was 200 feet of pavement and a turnaround in the spring shows up as a connected through-street by fall, and that new segment arrives as a layer you merge into the graph rather than a note in a field log. For a team tracking which developments are still phasing in, that turns a guessing game about whether a stub has connected yet into a quarterly check against what the imagery shows.
If stub roads are a recurring edit item on your network, it's worth seeing what a quarterly diff catches in your own subdivisions.