[Tagging] Verifiability wiki page: "Geometry" section added

Kevin Kenny kevin.b.kenny at gmail.com
Sun Apr 28 20:04:43 UTC 2019

On Sun, Apr 28, 2019 at 3:24 PM Colin Smale <colin.smale at xs4all.nl> wrote:
> How about taking the maritime baseline (according to UNCLOS) as the location of the rivermouth? Then it becomes both credible and verifiable, as the baselines are deposited at the UN for the purposes of determining the limits of territorial waters.

The hard-liners would, if I am not misinformed, argue that the UNCLOS
maritime baseline is not verifiable, since it cannot be ascertained by
direct observation in the field. It's not obvious that it's what we'd
want to follow in any case - it doesn't make sense to me to say that
Connecticut has no 'coastline', yet the Long Island Sound is a
'juridical bay' and the US 'coastline' follows the south shore of Long
Island only.

> Don't we already (grudgingly) allow polygons to indicate maritime areas with fuzzy boundaries, such as bays and straits? Why should that be tolerated while the same concept on land is frowned upon?

The same hard-liners have argued specifically that the practice you
mention is not allowed and has never been allowed, and that all such
features that are in OSM are in OSM in contravention of clearly
established policy. The example that I gave of the Red Sea, for
instance, cannot have a (multi)polygon because of the uncertainty
about the boundary between it and the Gulf of Aden.

While the fire rages, I've refrained from mapping any such feature,
even though I would very much like to have them in the database so as
to be able to label the features on large-scale paper maps that show
only a portion of the larger areas, not including any point features
near their (necessarily approximate) centroids. Without having areas
in the database, it's surely unclear how to go about labeling
incomplete features. Having a separate data store to maintain the
'forbidden' multipolygon relations is fraught, owing to the fact that
for similar reasons, anything in OSM resembling a 'foreign key' and
allowing OSM objects to be associated with an external database is
also strongly deprecated. (Wikidata references appear to be tolerated,
but the hard-liners are against them as well.)

By 'hard-liner' I do not mean disparagement. (I disagree with them,
but try to respect them.) In fact, I mean 'hard-liner' in two senses:
holding a hard line on the admission of indefinite features, and
demanding that area features be bounded with hard lines. One phrase
does double duty here.

Everyone appears to tolerate the presence of feature names in OSM. It
is unclear whether any of the hard-liners would argue that a
natural=peak, for instance, ought to have its name omitted if there is
no sign giving it. 'Look the name up on an existing
(license-compatible) map,' is clearly an unacceptable practice in the
hard-line world, but it's not clear whether 'ask a few locals what the
name is' would be depending on unverifiable external information,
since it's not the mapper's personal local knowledge, and it's even
more unclear how personal local knowledge of the name of a feature
could ever be acquired other than by transmission of someone else's
knowledge. How do you learn the name of something without someone
telling you? The only exception would be if the mapper were the person
who named it, and I think there's a broad consensus that we don't want
mappers naming things arbitrarily! Most of us would call that

More information about the Tagging mailing list