[Tagging] coastline v. water
osm at imagico.de
Tue Nov 24 14:20:29 UTC 2020
There seems to be quite a lot of anger and animosity in here - paired and in parts probably caused by a very selective and in parts flat out wrong perception of history so i will try to sketch quickly how the development of mapping of names of parts of waterbodies (that is mostly bays and straits) in OSM developed historically.
For a long time - until a few years back - these features were overwhelmingly mapped with nodes. This was consensus, not because of technical constraints disallowing something else, but because of the realization that in the vast majority of cases this is perfectly sufficient to document all verifiable information available about the feature in question. Practically in 2016 there were about 5 percent of all bay features mapped with polygons:
which - generously estimated - probably matches about the percentage of cases where you could argue that with a polygon you could record some verifiable information that cannot recorded with a node or a linear way (which still does not mean the polygon is a good data model for such features, just that it has in those cases - besides all disadvantages - also some advantages over a node or a linear way).
This situation was relatively stable - there were some attempts to suggest universally mapping bays with polygons rather than nodes previously:
which however never reached consensus because of the weighty arguments against this idea and because it was always clear that this would be a non-sustainable strategy for OSM in the long term.
Until early 2018 when OSM-Carto (where merging changes was at that time possible without consensus) added rendering of labels for bay polygons with label size and starting zoom level being determined by the size of the polygon but otherwise with no visual feedback or consideration for the geometry of the polygon:
- dismissing warnings about the counterproductive incentives this creates:
This lead to a massive change in mapping activities with some mappers engaging in systematic endeavors of removing bay nodes and drawing labeling polygons instead. You can probably say this was by far the most successful attempt at steering mappers into a certain direction ever undertaken by OSM-Carto. While the relative number of bay polygons compared to nodes only increased from about 5 to 15 percent while very few bays were actually newly mapped the total surface area of bay polygons probably increased by a factor of 100-1000 - many of them evidently pure labeling geometries. See
for some examples. This has lead to some mappers removing such label geometry drawings as non-verifiable and pointless (like the mentioned Gulf of Bothnia) - though practically none of these attempts could make a dent against the massive labeling polygon drawing trends.
What does this have to do with technical limitations or constraints? Very little. Technical limitations and performance constraints in rendering have never been a factor speaking against drawing large and non-verifiable labeling polygons. OSM-Carto and countless other map styles have for many years labeled huge administrative boundary relations without issues and this is not any more difficult for bay polygons. And if it was an issue the solution would be rather simple: Precalculating ST_PointOnSurface() on import in osm2pgsql.
The argument against drawing bay and strait polygons is one of practical verifiability and maintainability for the mapper. This is not a technical issue, this is a social issue.
Now i completely get the frustration of both mappers and map producers here. Mappers want their mapping to be shown in good quality in maps and if the only way to achieve that is to draw non-verifiable labeling geometries they are willing to invest significant time and energy into that and rationalize that in various ways.
And for map producers with a rudimentary GIS data analyst background and experience mostly in more or less atomic processing of point, linestring and polygon geometries and their spatial relationships but no deeper background in cartographic data processing specifically, the task of producing high quality labeling from bay nodes and a flat set of coastline ways or the osmcoastline output is a steep hurdle. And in conventional digital map production from dedicated cartographic databases (in contrast to OSM with its generic geodatabase scope) labeling polygons is the state of the art to manage labeling of course.
The problem we have here is that of a widening gap between the goals and aspirations of the mapper community - which naturally grow as OSM grows in ambitions - and the abilities and engagement in the non-mapping part of the community to develop and satisfy similar ambitions in cartographic quality without outsourcing the hard part of that work to the mappers. Too many people have followed the illusion for too long that the large corporate OSM data users will provide the necessary support in that field while it turns out (non-surprisingly in my eyes) that they have neither an interest in above average cartographic quality nor in substantially sharing methods and competency in the little work they do in that domain.
More information about the Tagging