Private road vs public road: how a routing graph should flag the difference

A private road that routes drivers through someone's driveway causes real problems: an angry dispatcher, a delivery truck stuck in a cul-de-sac with no turnaround, a resident calling to ask why strangers keep following their GPS past the front gate. Get the public/private call wrong often enough and a routing graph stops being trusted, even when the geometry underneath it is fine.
Why this isn't just a tagging exercise
Access control on a road segment feeds things downstream that have nothing to do with drawing a line on a map. Emergency dispatch routing weighs private roads differently, because fire and ambulance crews need gate codes or an alternate approach. Freight routing avoids private roads outright, since a semi working its way through a gated community's internal loop is a liability nobody wants on their books. Turn-by-turn navigation that sends a driver onto a private drive and strands them at a locked gate generates the kind of support ticket that gets escalated.
The private/public attribute changes how a segment gets weighted, whether it's eligible for certain vehicle classes, and whether a dispatcher even treats it as a through-route option. That's a lot riding on one flag.
What carries this attribute, and what doesn't
Most routing graphs lean on an access field (private, no, destination) layered on a highway or fclass tag, plus a separate maintenance or surface flag for roads that exist on paper but haven't seen a grader in years. These come from different sources and drift at different speeds.
Legal access, meaning public right-of-way versus private easement, usually traces back to county plat records, HOA filings, or permitting data. No amount of imagery tells you that on its own. Physical maintenance is a different story: a road that's grown over or rutted looks obviously different from one a county crew still services, and that part is visible from above. Existence, whether this segment is even on the ground yet, is the one imagery settles fastest. A gated community's new internal loop or a private estate access road gets cut and paved well before any county record catches up.
That last gap is where most basemap teams lose ground. A developer builds a new private road inside a subdivision, it's paved within a season, and it sits invisible on the graph until a field crew happens to drive past it or a resident submits an edit through a crowd-sourced queue that may or may not get reviewed. Months pass. Every delivery routed near that development either falls short of the new addresses or gets sent down a driveway that didn't used to be a through-street.
Catching the segment before anyone has to classify it
Imagery can't tell you a road is legally private. It can tell you a road exists where your current graph shows nothing, which is usually the slower and more expensive half of the problem. Once a new segment shows up as geometry, classifying it as private, gated, or unmaintained is a quick desk check against parcel boundaries, or just a look at the imagery for a gate, a cul-de-sac bulb with no connecting through-traffic, or a surface that never got paved to county spec.
Road Extraction runs that first half on a quarterly cycle: a vector diff of new and changed road segments pulled from high-res satellite imagery, delivered to merge straight into an existing network rather than replace it. It won't tell you which new loop belongs to a homeowners' association. It will put that segment in front of your team months before a field survey or a crowd-sourced edit queue would have caught it, so the private/public call gets made on a road that's actually on the map.
Run a quarterly diff over a subdivision-heavy service area and see how many new private loops turn up before anyone's filed an edit for them.