Would someone kindly check the borders in Lake Constance? Has this always been that way? It appears to be pure fiction.
There is a German contributor who edited countless Swiss municipality borders (in most of the country) and apparently this was done indirectly to edit borders within the lake (that doesn’t have an official one).
Changeset is Changeset: 177500898 | OpenStreetMap and a few within the lake.
See DACH: Landesgrenzen im Bodensee (National borders in Lake Constance) for previous discussions.
Salü zäme und Bonsoir !
I think you’re mixing something up. My changeset updating the national boundaries in the Obersee of Lake Constance is this particular one:
It should be noted that there is no border defined under international law in this section of the lake. Nevertheless, the neighbouring countries have agreed on a “technical border” (e.g. for geodata, bc boundaries have to be closed areas…). My sources regarding the technical border can be found in the changeset.
Whatever the solution may be, the border segment in OSM would ideally be a separate way (or a relation linking to separate ways) with tags providing an explanation and references about what it represents.
Your description comes very close to the current mapped scheme. We have way segments like w122991127 who are members of all the adjacent boundary relations and additional tags describing the actual status and source of the way segment.
The problem with this edit wasn’t the contents per se (it’s a fictitious border after all) but making a larger edit on a country border that effects three communities without prior discussion, not the least making alarm bells go off all over the place.
I would note that Changeset: 177500898 | OpenStreetMap has much the same issues. While the individual changes might be OK it is practically impossible to determine what was changed and why in detail, because the changeset is simply far too large and no rationale is given for the individual changes.
What is this this new relation Bodensee(TG) that is mapped as an administrative boundary 8? Are you saying there is an official Gemeinde in Switzerland that consists of the water of the Bodensee? That sounds very odd.
It’s complicated.
Historically lakes were mostly not considered part of the municipalities, in the mean time most of the cantons however have switched to including the water areas.
See [Heads up] Municipality boundaries in lakes - #9 by SimonPoole for the work I did on fixing that last year (note: not only did I give the Swiss community a heads up at the time but both the French and Italian communities while working on the lakes with international borders).
The BfS allocated/allocates municipality ids and names for the water areas in the remaining cantons that do not include the lake areas in the municipality. These are not “real” administrative entities for any reasonable meaning of the word they are just for filling up the otherwise empty space (statisticians …).
If you go back to the thread I mentioned above, you will note that I didn’t do any work on the Bodensee/Lake Constance as, as it has been pointed out many times, there are no borders there, and definitely the water area is not part of Swiss territory, and even less part of any cantons territory.
With other words the fake municipalities make even less sense than on the inland lakes, and while it can be reasonably argued that we need the polygon closing border bits there for technical reasons, there is no such reason for fake municipalities. But iirc one was included at the time of the original import (I believe the SG one) and it has stayed that way since then.
bfs_nummer=9329 exists in swissBOUNDARIES3D with name=Bodensee (TG)
Here’s how it looks like when loading https://data.geo.admin.ch/ch.swisstopo.swissboundaries3d/swissboundaries3d_2026-01/swissboundaries3d_2026-01_2056_5728.gpkg.zip in JOSM.
So, what you are both saying is, it is fine to just delete both these void-filler relations? Happy to do so.
The fake municipality isn’t really helping.
Also, ideally, the mapping of Lake Constance would differentiate between the main part of the lake (Upper Lake) and the lower part (which does have an agreed border).
For the upper part, I would have drawn something like way/416771494 (a bit further from the shore and excluding the north part) and left the center of the lake outside any country. This is also closer to what’s helpful for transportation and navigation.
I believe it already does that. Naturally there is an argument to be made that only the two (actually three) polygons in the Obersee should be removed.
Your description suggests that new changesets should automatically assumed to be malicious, without the slightest review. That is not how I understand the OSM project. It also becomes clear why some contributors may seek to avoid forum discussions, given the way this is being handled here. I dislike how this whole thread is set up. Seems more like “naming, blaming and shaming” than a constructive discussion with competent input. But we can still fix this.
That’s not how state boundary mapping works! Edits of national boundaries should be based on solid evidence that can be verified via reliable and widely recognized sources. My changeset is based on three official sources provided by the authorities of all neighboring countries. Your proposal to simply draw an arbitrary line may explain why you chose to label this thread with the keyword “fakes”; this comes across as a not unbiased way to start a discussion about my changesets. I find the title of this thread somewhat misleading btw.
All neighboring OSM communities have been comfortable for years with the way the lake area is mapped using shared boundaries, and this status quo is firmly established. Introducing overlapping boundaries or even no-man’s-land areas now would suggest an ongoing border dispute that simply does not exist. The neighboring countries cooperate very well in all matters concerning the lake. It should also be noted that the line geometry of the status quo (prior to my edit) was not based on any comprehensible data. The previous boundary—presumably originating from outdated imports made before the “technical border” was established—could not be confirmed by any verifiable source.
I personally would also have no issue with deleting these void municipality relations. My intention was simply to fill this one last missing gap inside the whole Swiss boundary that was not tagged as admin_level=8. The neighboring St. Gallen lake area relation has existed for years. My suggestion is to either delete both or neither, and to also review the other Swiss lakes. NE and BE have similar fake municipality relations as well.
Bielersee (BE)
Bielersee (NE) ➔ should be named French btw
Lac de Neuchâtel (BE) ➔ “Verwaltungskreis Seeland” is monolingual German-speaking imho
Lac de Neuchâtel (NE)
Thunersee
Brienzersee
As a moderator of the Swiss category I removed the unnecessary keyword from this thread.
As there seems to be general agreement (or at least no disagreement), I’ve now deleted the void-filler municipalities in the Bodensee: Changeset: 177637925 | OpenStreetMap
I’d be in favour of deleting the other void-fillers on the lakes as listed by @welzheimerwald as well but I’ll leave that discussion open for a bit.
@habi: Can you add some other indicator instead? Maybe “fictional_border” instead. The contributor clearly confuses actual agreed upon borders with what he copied here.
Edit: completed it with fiction_municipality and fictional_border
An alternative could be to give the fake municipality some other admin_level that doesn’t imply it’s a municipality. Similiarly, admin_boundary could be replaced with something else.
In general, making a changeset unreviewable by even the most experienced members of the community, may not be malicious by intent, but might just show a complete disregard for others.
If a user knows what they are doing, they could easily redo the edits in smaller steps, once reverted.
The big difference is that they actually belong to the territory of Switzerland/the resp. cantons contrary to those in Lake Constance. As I pointed out Border in Lake Constance/Bodensee - #8 by SimonPoole these have been around since the original import and that is when they got their current names.
As this is a matter of cantonal legislation it will probably just be a matter of time till Berne and Neuchâtel catch up. Except if they are actually causing issues, as a tendency I would just leave them. Not that there aren’t other oddities like the Staatswald Galm, or Comunanza Capriasca/Lugano with iffy level 8 admin boundaries.
Just a word of caution about references:
to be a reliable reference on a Swiss municipality, it’s better to look for a cantonal source. To put it simply, anything with “Bund” in its name needs to be cross-checkable by other references.
It may well be that OFAS/BFS is reliable, but they are generally thought to be reliable for statistics (at least until they withdraw them).
So something may be a reliable reference for a technicality, but that doesn’t make it one for everything else, be it cantonal matters or national borders.
For some reason, I couldn’t find Way: 123004322 | OpenStreetMap earlier, but I do think it’s good to detail it that way. The question still remains if it qualifies as “boundary”=“administrative” with “admin_level”=2.
What you want to map—agreed national borders defined by international law—simply does not exist in this part of the lake (Except for the Konstanzer Trichter, where the border is defined by the intersection geometry of several shoreline points and the straight line of sight to the tower of Constance railway station. Here btw, the way segment could be split at n1372919438; the “technical border” tags could then be removed, and the Baden–Swiss Border Treaty of 1878 tagged as the source for the remaining western way segment).
Consequently, if mapped correctly under international law, your proposal would result in an unclosed boundary polygon.
This is suboptimal for spatial queries and similar tasks…
