Announcement of Withdrawal

Posted by Shernott on 30 August 2011 in English (English)

Due to a strong disagreement with St├ęphane Brunner, who seems to feel an irresistible urge to "correct" and "improve" my contributions, I will stop mapping in the larger Lausanne area.

Location: Penau, Manloup, Le Mont-sur-Lausanne, District de Lausanne, Vaud, 1052, Switzerland

Comment from Firefishy on 30 August 2011 at 13:46

Can you link any example changesets?

Hide this comment

Comment from Shernott on 30 August 2011 at 14:24

Our disagreement is basically about whether areas with landuse=* can and should go across a way with highway=* and whether you can and should use a way with highway=* as border element of a multipolygon with landuse=* or if a road needs to have some physical space to its left and its right.

I think the first way of doing it is simpler for moving and correcting nodes of the road (you don't have to worry about keeping two other lines to the left and right in parallel) and allows for a cleaner rendering without white spaces. Now I can also accept the validity of Mr. Brunners perspective of representing the reality of physical space to the left and right side of the road. Here it needs to be said that Mr. Brunner has great merits in tracing satellite and aerial images, and he has created himself a large amount of areas and multipolygons that do not correspond to his own ideal of physical space to the left and right side of the road.

If he would start "improving" his own contributions first, I would be much more inclined to accept his "improvements" on my contributions. But the way it is, it feels more like having some know-it-all breathing down my neck, and it's just no fun to work this way.

It's not the first time that we had this disagreement, but this time, things went a bit too far and got out of hand.

Here are his first two change sets:

Here is my reversion:

Here is his rereversion:

Hide this comment

Comment from Pieren on 30 August 2011 at 14:36

This is more a question about the granularity of OSM data. We always have in OSM highly detailed map details coexisting with less accurate data. The "edits" should go in one way only, the way of 'improvements' : contributors are allowed to replace existing data if their contribution is more detailed, not in the other way.

Hide this comment

Comment from TomH on 30 August 2011 at 14:44

Yes and no - there is a second side to the issue here in that the "more detailed" data is harder to maintain and hence increases the maintenance overhead which may not fall on the person that contributed the more detailed data.

Personally once an area gets details (ie has buildings) I tend to delete the landuse altogether as it is very annoying to maintain, and makes maintenance of roads etc much harder, and is of very little benefit.

If it is present then I like to share it with the road so that the edges move properly as the road is refined over time.

Hide this comment

Comment from Shernott on 30 August 2011 at 14:48

I agree with your idea of increasing the level of detail. But given the offsets of various maps and the precision of GPX traces, there are limits to the absolute detail that is possible. At some point, it comes down to aesthetic preferences.

And in a situation, where there is plenty of opportunity to increase the detail all over the place, it feels a bit weird, if a user chooses to stalk you and "improve" your contributions. Under such circumstances, I prefer to map in areas where my paths don't cross Mr. Brunner's. That's all.

Hide this comment

Comment from Sundance on 30 August 2011 at 18:25

Do people build houses on streets? Not normally, do you have parks on streets, again not normally. Streets should rarely be in a landuse area in my view. One of the basic concepts of OSM is "MAP EVERYTHING", so eventually 100% of the world may have a landuse tag.

To me I try to never delete somebody else's work but try to add to it.

Hide this comment

Comment from Shernott on 30 August 2011 at 18:51

@Sundance: Well, depending how you see it, roads do run through forests (as do rivers) and tracks are often at the border of forests, fields or meadows and separate two of those areas.

In OSM, roads are usually represented as a line, which is of course a rough, but quite useful approximation. So if you have a field on one side of the road, and a forest on the other, it is way more convenient for subsequent editing to use only one line instead of juggling with three lines that you have to keep parallel.

Hide this comment

Comment from dcp on 31 August 2011 at 11:16

Basically either the land is used or it is natural.
Roads, tracks and paths go through residential, farmland,forests and natural land.
If I walk, ride or drive on any one of the above then I am on the relevant landuse/natural area; e.g. a residential road going through a residential area is on landuse=residential.
For an extreme example of excessive mapping around roads goto:
using JOSM.
I contacted the contributor about this time consuming redundant work but I was unable to convince him. As I said to him/her: It not wrong so do it your way.
As far as my own work is concerned, it is being continually corrected by others and that is good so. What I don't understand, though, is why my contributions are no longer in the relevant changesets: They are lost in history, but that does not worry me at all.
As anyone who uses OSM-Inspector knows, there are hundreds of thousand of geometrical errors in the OSM Databank. Where areas have mutual boarders then it only makes sense to link these boarders together thereby reducing the risk that new users will once again create new geometrical errors.
My tribute goes to the renderers who somehow make usable maps out of the OSM data. It is the chaos theory working here!
As far as buildings are concerned, where is the benefit within a defined residential area. Are we heading for door to door routing? Nice to have but not needed. Or is it 3D animation? Nice to have for game players I suppose.
We OSMers just do the bits we like to do and leave the rest for others and that is as it should be.

Hide this comment

Comment from jutezak on 31 August 2011 at 19:28

I also feel that a road is symbolic. It may or may not match its actual width. If landuse were to accurately follow the road, it would have to overlap anyway or maybe keep a distance.

I have a feeling that getting landuse not overapping with roads makes some sense. But mostly after sidewalks have been added to those streets. And especially then, it will make more sense to use the tooling to "make things right": a park could be a square with 'landuse park' and the roads just overlay on that, the understanding being that there is no greenery when there is road. When the roads have attributes about width and layout, thay would automatically carve that out of the landuse (they do, they render over it normally).

Hide this comment

Comment from nmixter on 1 September 2011 at 05:11

I try to not cross major roads with landuse tags when possible but will overlap smaller roads. Either way is fine though. If you look at official munipality zoning codes, they will have many different codes like r1, r2, r3, c1, c2, c3 etc. osm doesn't have any way to diferentiate them. An area could potentially be divided into dozens of smaller areas. Things like flood plain aren't even defined in osm. Zones are highly subjective. Rather than worrying about the location of zones we should add what we can as time permits. There is way too much ground to cover to be worrying about minute details of preference.

Hide this comment

Comment from Shernott on 1 September 2011 at 06:57

Of course everybody should map the way they feel it's appropriate. But that's not my point. My point is that I took the time to go out with my GPS and my mountain bike. I wanted to complete, correct and refine some tracks, paths and landcover in a certain area that was only roughly traced with satellite maps and had lots of unconnected and incomplete ways.

And in this situation, I find it quite frustrating, when another guy in front of his computer screen decides based on outdated satellite maps (which do have an offset and show the tree cover in winter -> decidous trees are barely visible) how things should look like. So I have withdrawn my GPX traces from that area and I'll leave him to his wisdom. Apparently an armchair and an old satellite map is all that Mr. Brunner needs to do "his work".

Hide this comment

Comment from dcp on 1 September 2011 at 15:53

Shernott, I agree with you. But, it's not worth getting angry: Just stay annoyed and move on: we are doing this for fun, aren't we?

Hide this comment

Comment from Genscher on 3 September 2011 at 21:34

Yeah but we should kick people who are worsening the map, at least warn them. Time is too precious to be wasted.

Hide this comment

Comment from Shernott on 5 September 2011 at 09:10

@Genscher: My personal experiences with Mr. Brunner aside, to say that he "is worsening" the map would not do justice to his positive contributions.

As far as I am aware, Mr. Brunner has contributed a copious amount of (new) information by tracing aerial images in and around Lausanne and has probably already contributed more to OSM than I ever will. My ambitions are far more modest. I only map areas and places that I know and care about. And most of all, I am doing this for fun. So with regard to Mr. Brunner's undisputed merits in mapping the area around Lausanne, I decided to withdraw from that area to avoid further conflict and to allow Mr. Brunner to have things looking whichever way it pleases him and based on whatever data sources seem trustworthy to him.

Hide this comment

Leave a comment

Parsed with Markdown

  • Headings

    # Heading
    ## Subheading

  • Unordered list

    * First item
    * Second item

  • Ordered list

    1. First item
    2. Second item

  • Link

  • Image

    ![Alt text](URL)

Login to leave a comment