Frederik Ramm wrote: > Our general rule is that things we map must be verifiable on the ground, > i.e. someone who goes there must be able to check that the feature does > indeed exist as described in OSM.
This message isn't really contradicting what you wrote later in the message, but I have to remind others that verifiability doesn't need to be easy, or it might not even be possible without expert knowledge, equipment or reference to "other" material not directly present in the location being mapped. As a counterexample, let's consider an underground water or heat pipeline in the city; these are constantly being repaired, improved or preemptively replaced. A mapper sees the open repair pit on a road, interpolates the pipeline's location and turns with other known features and imagery getting the accuracy down to, say, 50 cm. A month later, the next mapper would only see the patched pavement, and possibly some manhole covers far or very far apart; only the general alignment can be inferred. Then, a year later the whole street is repaved, and only the manhole covers remain, but one couldn't know (for some level of certainly) whether they belong to pipelines crossing the street, or whether one pipeline runs along the street, unless one surveys the whole neighbourhood. Nobody is however suggesting the pipeline should be removed, or turned into a straight line just because every casual mapper can't verify the form and attributes on site. In fact, even administrative boundaries can be verified "in place", in theory, but it would normally take a whole lot of time and incur expenses: build an illegal shack anywhere near the border, and wait for the officials to react, and read the papers you get to see which area you were in. And that could be costly. Likewise, national borders can be verified by doing something that attracts the interest of the enforcement authorities (police, border guard, military); when you see which country they serve, you know which country you were in. That's hardly "not present" on the ground. > "official" bike routes >... > (if not fully > signposted) by a national body, they were ok to have in OSM. They > wouldn't always be verifiable on the ground but it would be easy and > straightforward enough for anyone to verify them using existing material. Back on the topic of this thread, the unmarked but published routes, I would be on the inclusive side, (roughly drafting) "as long as the exact route is or has been used for the same purpose by several parties (n>>2) on different dates, and is not the only connection between two destinations" or something like that. This would rule out the personal dog walking routes, single mapping walks and other personal notes of the type "I was there", but would "allow" the routes of e.g. marathon competitions, long distance cycling races and routes of motorcar races on the public road network, when the event is not a one-off. The routes are very visible for some time each time the event is held; either a line on the ground, fences or other barriers, or just signposts that get removed later. If we record the data, it _could_ be reconstructed in the future even if the relation is later destroyed and eventually deleted, if we don't, it's soon gone forever. The last part of the draft sentence above is to say that when a single forest trail connects two villages, it's probably not a route in the sense of this discussion, but when two or more trails connectthe same villages but the overwhelming majority of pedestrians going from A to B use only one of those trails (for whatever reason), it 'could' be something in the DB; not part of a walking route network or anything, but not something that should be deleted straight away, either. If there's a signpost for it at the ends of the route, that's even better. -- alv _______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
