OpenStreetMap

Label painting guide continued

Posted by imagico on 9 January 2019 in English.

This is an update to my previous piece on techniques how mappers who reject the concept of verifiability in OpenStreetMap (i have written more elaborately on that) can paint labels in the standard map style on openstreetmap.org.

Meanwhile some mappers have taken my suggestions to heart (irony is a tricky thing…) and started drawing labels more or less exactly the way i demonstrated. The most prominent label in the map, showing at z2 and above, is now based on a completely non-verifiable polygon geometry.

natural=bay extreme so far the pinnacle of abstract non-verifiable polygon drawing to paint labels - but there is still a lot more potential, no one aimed for membership in the zoom level zero club yet

But there are other good news for label painters.

There is a new alternative to natural=bay for blue label painting in the form of natural=strait (introduced here). Works exactly like natural=bay so you can add some excitement to label painting by contemplating if you want to call it a bay or a strait.

natural=strait for some just facepalm-worthy, for others the best invention since sliced bread…

Also helps with all those nasty complaints you might receive from other mappers about your label drawing efforts. If you get a complaint about using natural=bay just re-tag it natural=strait and the other way round.

For black labels there are also exciting news. You can use natural=cape as alternative to place=island (see here). Works the same way, depending on size of the polygon labels start appearing as early as z4. And while place=island is easy to falsify even for non-locals natural=cape has much wider applications without being too suspicious for mappers overly concerned with semantics.

place=island works also with natural=cape instead of place=island now

Have fun!

Discussion

Comment from kocio on 10 January 2019 at 23:09

Here’s a proposition how to verify shape of the Baffin Bay for example (item number 14, page 10):

http://www.iho.int/iho_pubs/standard/S-23/S-23_Ed3_1953_EN.pdf

Comment from imagico on 11 January 2019 at 00:08

Nice to see someone is falling for my trap (sorry).

Please check if the following applies:

☐ you have understood that verifiability in OSM means independent verifiability based on the observable geographic reality.

☐ you are aware that the polygon painting in OSM is not anywhere near the IHO declaration.

Comment from kocio on 11 January 2019 at 10:50

Ouch! I was not expecting this trap, my leg…

2) is pretty easy:

At the core, “verifiability” is that everything you do can be demonstrated to be true or false

This is what just happened - you demonstrated that it is false and needs to be improved. I can compare them and say the same.

1) is more complicated. How do you observe the name of the named area outside the land? If you can’t, does it mean it does not belong to the OSM database at all?

Comment from imagico on 11 January 2019 at 11:17

This is what just happened - you demonstrated that it is false and needs to be improved.

No - and you selectively quoting what you like and leaving out the rest does not make it right. If i demonstrated something that is the self referential cultivation of completely made up ideas is swashing over to OSM from Wikipedia.

As linked to i have written at length about the idea of verifiability in OpenStreetMap. I don’t really feel like explaining the fundamentals again here.

Everyone is free to map stuff the way shown. What i try to do here is give readers a bit of an idea why this is a bad idea and a bad development for the project. Not everyone will understand that - as said irony is a tricky thing. And some people just want to paint labels in the map (which i perfectly understand).

Comment from kocio on 11 January 2019 at 11:23

OK, you criticized my take on 2, but did not answer 1, which is more interesting for me. I do not understand your position on that.

Comment from imagico on 11 January 2019 at 11:57

I have not expressed any position on name tagging here. I have no issue with the verifiability of the name Baffin Bay.

Comment from kocio on 11 January 2019 at 12:02

That’s exactly why I ask you. How do you verify “Baffin Bay” name, for a start, if not using 3rd party sources you are so much against it?

Comment from imagico on 11 January 2019 at 12:30

I am not a local of the area nor have i ever visited it so i have no way to reliably verify the name from here.

I would suggest not to concentrate on name tag verifiability. Completely different topic which has nothing to do with the subject of this diary entry. If you want to reject verifiability as a concept because you have no way to verify names at the distance i have no issue with that.

Comment from kocio on 11 January 2019 at 13:56

So, if a local will draw a geometry, you will accept it?

Comment from Claudius Henrichs on 21 January 2019 at 12:37

I think what Christoph is aiming at is the the lack of verifiability of large natural geographical objects, particulalry in a maritime context, but also named objects of topograhical relief. Particularly how do you verify the outer boundary of a OSM polygon for a: * maritime bay * maritime strait * valley * ridgeline * mountain range * peninsula * cape * cliff * mountain (not it’s highest point, but the geographical feature all the way down to its feet) It’s pretty clear that humans generally have a consensus on where roughly a geographical feature is located by pointing at some central point or area. But how do you translate that fuzziness of a centroid label into exact outlines as required for representation as an area object?

Comment from kocio on 21 January 2019 at 13:02

It’s not that easy.

The other day Christoph asked about how one verify location of a (bay) node told that it’s a middle of that bay. Which suggests me that he believes it’s possible to estimate the shape, since otherwise it would be not possible to calculate the middle. So node position is less verifiable for me, since not only you have to know the shape, but also choose some centroid method, which can give different results on non-trivial shapes.

But if you believe the local can say the name, why shouldn’t we believe she knows the area, only the node? (there are different problems to say who could be “local” in remote places, but that’s not the core).

It’s pretty clear that humans generally have a consensus

But which humans? If not locals, then 3rd party sources for verifying are OK - which Christoph does not like to accept.

Log in to leave a comment