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

[Kevin Kenny]

On Sun, Apr 28, 2019 at 7:39 AM Tobias Knerr <osm at tobias-knerr.de> wrote:
> Yes, it's often not possible to agree on a precise border for these
> features. But nevertheless, there are typically areas that are
> definitely part of them, and other areas there are definitely not part
> of them.
>
> The above is verifiable geographic information, so it ought not be
> off-limits for OSM. I've always thought of mapping features with fuzzy
> boundaries as an unsolved challenge for our data model, not as something
> that should be categorically excluded from OSM.

Exactly this.

It's 100% verifiable that Port Sudan lies on the Red Sea. It's 100%
verifiable that the Red Sea is longer than it is wide, and that it lies in
a generally north-south direction. An enormous majority of its shoreline
can be mapped, and the locals would agree that the body of water beyond the
shore is the Red Sea. None of that information is non-verifiable, and none
of that portion of its margin is ambiguous, except for potential arguments
over which point in the twice-daily and monthly tidal cycles determines it.

That information is there, in the field, and we can't map it.

Essentially, our data model quests for mathematical perfection. Because the
boundary is indefinite at the strait at the sea's southern margin, some
argue, topology implies that the entire interior must be ambiguous, because
that boundary could be drawn anywhere - from excluding the entire sea by
bending it north to the entrance to the Suez Canal, to stretching it east
until it encompasses the entire Indian Ocean - or even all the
non-landlocked waters of the globe. That's absurd. Nobody would argue that
Mumbai or Dar-as-Salaam lie on the Red Sea; in fact, nobody would argue
that Djibouti lies on the Red Sea, and by the same token nobody would argue
that Port Sudan or Jeddah do not.

Precisely the same quest for topologic perfection is responsible for the
rule that fixes the mouth of a river at its tidal limit - which gives rise
to the absurd result that the mouth of the Hudson River is at
https://www.openstreetmap.org/way/90929525, or of the Connecticut River at
https://www.openstreetmap.org/way/22889153, while the reaches below are
part of the ocean. This absurdity arises directly from the concept that
what the locals would call the river's mouth, being perforce represented by
an imaginary line across otherwise unmarked water, is not verifiable.

I am NOT arguing here that we should open the door to utter chaos. The bad
examples that people like to trot out, such as the Drake Passage or the
Amazon rain forest, are just that - bad examples. Their margins are
verifiable nowhere, or only on a relatively small portion of the object,
rather than nearly the entirety of the margin.

Note that in this argument I've consistently used the word, 'indefinite'
rather than 'imprecise' or 'inaccurate'. 'Imprecise' means that either the
measurement technology or the quantity asserted have limits to how finely
they can determine a quantity. 'Inaccurate' means that owing to some
influence, the measurement differs from the true value, generally by an
amount that exceeds the precision. 'Indefinite,' by contrast, denotes that
the quantity in question is unknowable exactly - a better measurement will
not help.

If the quantum mechanicians are to be believed, every quantity that we can
measure is indeterminate at some level. Moreover, the things we measure are
referred to abstractions that are indeterminate, such as geoid models that
try to describe the Earth's gravitational equipotential while ignoring
'local gravitational anomalies'. Nothing that we map is a Platonic ideal,
because all of us inhabit Plato's cave.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20190428/e8bdd6f5/attachment-0001.html>