[Tagging] Stop the large feature madness (was: Tag for a plateau or tableland?)

[Kevin Kenny]

On Thu, Apr 18, 2019 at 5:49 AM Christoph Hormann <osm at imagico.de> wrote:
> You apparently misunderstood what i said.  My 'surveyable in a single
> day by a single mapper' rule of thumb refers to mapping something as a
> single feature.  A river several thousand kilometers long for example.
> The river is locally still a verifiable element of the geography and
> can be mapped - piece by piece as it is generally established practice
> in OSM.  But if you create a feature for the whole river extending over
> thousands of kilometers that is not something you do based on local
> knowledge, that is based on social conventions you have read up in a
> book, on wikipedia or elsewhere.

And therefore the Amazon, the Nile, or the Mississippi ought not to be
named in such a way that a large-scale map can show the names?
Essentially, you're making the statement here that if local mappers
pool their knowledge to realize that the river in Alexandria is the
same river in Aswan, that's a mere social convention and has no place
on the map. Yesterday, Farouk maps one piece of its shoreline, and
today Karim maps an adjacent piece. According to your argument, those
ways ought never to be merged because ... why, exactly? Because
there's no provable single mapper with local knowledge of both of
them?  If a few dozen mappers do that, you can wind up with a feature
that's a thousand km long. They have each contributed the local name -
and it's the Nile for the entire extent, as the boatman who plies its
length could have told them.

> Everything else in physical geography is typically mapped locally piece
> by piece like the rivers and creating large features - while done by
> some mappers for the purpose of label painting - is generally disliked
> by most mappers because it is very hard to work with these and
> represents no additional meaningful information.

That's where we disagree. The additional information is that the
multiple features represent the same physical object.  Without that
information, dealing with them at large scale becomes an exercise in
ambiguity - trying to build up larger features from tiny ones, by
guessing that adjacent ones with similar attributes are parts of the
same whole. One misplaced node on a shared boundary, and the whole
thing falls apart.  That's one of several things that relations are
for - identifying that some number of small features are parts of a
larger one.

Please avoid the term "label painting." What you call "label painting"
is the entirely reasonable desire to have recognized, named objects
appear on the map with their names.  Calling it "label painting" and
saying that it provides no useful information is stating, "you
shouldn't want to work with that identified, named object, only its
parts or only some arbitrarily chosen part within it." It belittles
the user, and cuts off the conversation. There are much more
productive approaches: "I understand your want, and you can't have
that right now because we haven't worked out how to do it. Can you
help us define the needs?" is one. "That doesn't fit exactly with the
way we structure the data right now, would it be an acceptable
alternative if you were to ..." is another.

The "hard to work with" argument is what I said is a technological
limitation. If the Gulf of Aqaba - a well-defined object except for an
imaginary line across its narrow mouth - cannot be mapped because the
resulting relation would be too large, or because the data model
cannot cope with the indefiniteness of having two adjacent waterbodies
(waterbodies ordinarily do not have a well-defined bright line
separating them), that's a limitation of the data model or the tools
that work with it. Those are technological limitations, pure and
simple.

I return to an example that I used a few months ago, which I still
haven't attempted to map because largely of your and Frederik's
extremely vociferous objections: Jamaica Bay.  Let me point you to
maps that include the bay, created by skilled human cartographers:
https://caltopo.com/l/1UL6

What you see at that link includes the boundaries of several map
sheets. Note that Jamaica Bay is lettered as a prominent feature on
all of them.  The smaller features - Island Channel, Pumpkiin Patch
Channel, Grassy Bay and so on are parts of it, and are typically known
only to those who go out on the water - many land-dwellers around the
bay typically do not know their names. But the bay is known to all of
them. When I lived down that way, near https://caltopo.com/l/TPG3, my
neighbours might not know the individual parts were named Mott's
Basin, N****r Bar Channel (N****r bar is now Johnson Bar, and not a
moment too soon!) or the Head of the Bay - but everyone knew Jamaica
Bay!

If you simply left it as 'coastline' they'd look at you as if you had
two heads, and even more so if you called it the Atlantic Ocean.  The
ocean is a short distance to the south. It has pounding surf; the bay
has slack water. The ocean is salt, the bay is brackish. The bay has a
considerably wider tidal range at spring tide, and so on.

But in the world you describe, there is absolutely no way to make an
object called 'Jamaica Bay' and define its boundaries. Because there
is a small ambiguity in precisely where the bay meets the ocean at
Rockaway Inlet, the bay cannot have a legitimate existence in OSM, and
someone who lives along the shore of Mott's Basin cannot see it on a
small-scale map - because the only way to assign it a name is to
"label paint" a point somewhere in it.  (And yes, I do despise that
particular form of label painting. But I don't say "don't do it" -
because painting the label, as long as the label is reasonably close
to representative of the object, is a starting point that another
mapper can improve.)

To me, the idea that I cannot contribute to OSM the information that
would allow me to produce a small-scale map of, say, Inwood or Far
Rockaway with the bay indicated is frustrating.  You relegate the bay
to being a 'social construct'. It is toward the lower bound of sizes
that you consider unacceptable for a single object - but it surely
falls within that range, and I could just as well have made the
argument with the Long Island Sound, which is an
equally-well-determined, named waterway, with all the same objections,
plus being much larger.

Plateaux, mesas and tablelands are not among my local landforms, but I
can well imagine that their inhabitants feel the same sort of
frustration.

> My argument is not a technological one, it is a social one.  Mapping
> only things verifiable based on local knowledge in OSM is essential for
> the social cohesion of the project across many different cultures world
> wide without creating an imperialistic dominance of some cultures over
> others.

So, we have, say, the Mississippi River, or Jamaica Bay, or Cape Cod.
They are well-defined physiographic features, which vary in size by a
couple of orders of magnitude but share the concept that for their
entire extents, the locals consider them the same river, the same bay,
the same peninsula (with tiny bits of ambiguity at their margins.
Wanting some representation of them as unified wholes is now cultural
imperialism?

Now, I could imagine that if the world were other than as it is,
another culture might insist that the main stem of the river was the
Missouri, rather than the upper MIssissippi, leading to disagreement
about the boundaries. That disagreement could be very ugly if the
cultures were, say, continually embroiled in political conflict about
other matters. In that case, making a single decision about the
boundaries might conceivably be imperialistic. Even there, we have
flexibility - we can create relations that represent the facts that
both sides agree on, and we can create competing relations
representing both sides' view of the conflict. Relations are that
flexible. That's not the ordinary situation in any case. For most of
the objects in question, there's no dispute.

I cannot see where calling Jamaica Bay by its name is cultural
imperialism! (Ignoring, I suppose, the fact that the First Peoples
surely had a name for it, which name is lost to history.) Even if it's
too big to map solo (which might be debatable), surely the local
mappers around it agree with their neighbours that it exists, it has a
name, it's a bay, it's the same bay whether you're looking at it from
Canarsie or Neponsit. If arriving at a consensus with your neighbours
is imperialistic, then I fully support empire.