After getting some feedback on recent changesets, The changes below are possibly not fixing blatant mistakes, as I assumed. I assumed these were errors and did not need discussion. Before doing any more edits related to golf=* features, I’d like some feedback. There are a few topics closely related.I will revert or continue editing in an automated/semi-automated way based on discussion here. Thanks for your help!
Features within a golf course
To keep edits cleaner, each of these changes would be done separately to address specific tags.
I noticed a few things tagged as finely_mown=yes and found it was added around 2012-2014 with a slow decline in use. It appears to be mainly added by an automated edit. I assumed it was accidentally imported since I could not find documentation or discussion about it. Also, it is used very sporadically. But someone pointed out the wiki article Any tags you like. So maybe this was intentional?
Would it make more sense to use a schema similar to other features? Something like grass=finely_mown. But that is also in decline. Or should the finely_mown=* tag be removed with no replacement? I will wait to see what thoughts are before doing anything to the features in the changeset (178942807) where the tag was removed.
Defining a surface with landuse and natural
These tags are the most inconsistent and would need to be a semi-manual edit. Very odd ones (there are a LOT) would need to be viewed from satellite imagery in addition to getting context from existing tags. Mostly, it seems natural=grass,landuse=grass (with variations of grass(.*)?) and natural=sand are used to where surface=* should be. At least that is how I understand it from the wiki
Avoid using landuse=* or natural=* tags to indicate its surface, use surface=* instead.
Thoughts on replacing landuse=”grass.*” with surface=grass and natural=sand with surface=sand?
first and in general please stop carrying out undiscussed automatic edits. If you think something is odd on a larger scale, please first start a discussion here or in the local category.
To this specific matter. It’s not acceptable you mass-deleted finely_mown=yes. Regardless it’s documented or not, it is/was valid information. From my point, please restore this information.
If you are interested in the quality of golf features, I rather suggest to take an actual, manually look on those and actually improve them (and part of this can be a replacement of “obscure” tags). Your mass-deletion of “obscure” tags doesn’t improve quality for me. At least I can see no benefit and you didn’t mentioned one either.
Can you give some examples of where you consider landuse=grass to be a problem, and why you think a surface tag would be better. It’s a very common tag with more than 7 million uses as described on its wiki page. Do you feel it is used in golf courses in a way that is inconsistent with its general use as documented there?
I also believe landuse=grass is a problem, because it is not used consistently. It may have millions of uses, but they do not follow a common pattern, at least they do not seem to express a more specific meaning than “an unpaved surface with something growing there, but probably not many trees”. For example you cannot tell whether it is a natural surface or created/maintained by humans. Obviously I didn’t look at millions of instances, but from the observations in years of mapping, I can tell you that most of the times these seemed better to be retagged for clarity. The tag has departed a long way from its originally intended meaning (grass patches around / between roads), also thanks to one of the worst choices of wording in the OpenStreetMap tagging universe.
Converting landuse=grass to surface=unpaved would make it more evident that the feature tag is still missing.
Can you give some examples of where you consider landuse=grass to be a problem, and why you think a surface tag would be better.
In response to @alan_gr and @dieterdreist, landuse= is one I noticed along the way that is used inconsistently and could easily be changed to match what the wiki suggests. Sometimes it is surface=grass sometimes it is landuse=grass sometimes a combination that needs a closer look.
Land use is a weird one since it doesn’t seem to always be used for the original intent. Using landuse=grass seems to suggest it is grass that has no use. On a golf course landuse=commercial may make more sense… but I not going down that road here .
There is nothing glaringly wrong other than most uses of it are intending to show the surface is grass. Often surface=grass is not present. For example any of the golf features around this fairway.
With golf=* it seems clear in all of the instances I have seen that landuse=grass(.*) could easily be converted to surface=grass. The same goes for natural=grass(.*) which is a common tagging mistake.
Have you tried contacting people who added it in changeset comments? And looking at changeset that added it?
@Mateusz_Konieczny, after noticing the tag in a few places, I thought about doing that. At that point, all of the edits I found for finely_mown were very old and I assume the person would not remember much about the changeset.
Yes. But any of my larger edits are semi-automated. Now that I understand it may not be a blatant mistake. I am waiting for feedback before continuing or undoing.
Tag:leisure=golf_course - OpenStreetMap Wiki suggests how to tag all those different “areas” on a golf pitch. Just changing landuse=grass to surface=grass doesn’t make anything better, except it’s most likely not rendered anymore. Btw. if so, landcover=grass might be proper main tag for grassland which is not “used”.
If you actually want to improve the situation, I would suggest to go down that road. Though I would think you can’t auto-fix this.
It still renders. I have made these edits manually before. Are there instances where the wiki suggests doing the opposite of what would be rendered? Or the opposite of what is generally accepted? If so, I will not rely on the wiki as much.
Is having things tagged consistently as described in the wiki a benefit?
Yes, I don’t see much of a benefit. You exchange from one “draw it green” tag to another. No information gain, no validation against OTG or aerials, whether there is still grass.
I would see an improvement, when you end up with golf=* + surface=*.
Yes, we mainly disagree about the method, not the the result in the data. I don’t see much benefit in an automatic edit like: Way History: 251669309 | OpenStreetMap
Neither did your edit indicated that green still exists (aka you actually validated it against current aerials or OTG) nor did you added/improved any information. Maybe we disagree on the improved part.
Adding value to OSM (for me) would be to detect such faulty golf courses, validate all the features and get them up to your/OSM standards. My experience shows, if there are such “basic” issues present usually there is more that requires fixing.
because it prevents mapping the actual landuse on the same object, or because it can be simply overlaid with other landuses because it isn’t a landuse anyway?
.
Good to know. That is what I am suggesting.