OpenStreetMap

OSM: Why can't contributors check/correct their own work!

Posted by Little Brother on 17 July 2017 in English (English)

Just take a look at this:

http://tools.geofabrik.de/osmi/?view=geometry&lon=3.44678&lat=48.05899&zoom=6&overlays=self_intersection_ways,self_intersection_points

and ask yourself: Do I check my own work?

OSMI is just one of the many very good tools available: Why don't more contributors use these tools?

Here is another view:

http://tools.geofabrik.de/osmi/?view=geometry&lon=4.30371&lat=49.27784&zoom=6&overlays=single_node_in_way,duplicate_node_in_way

and this one shows unconnected ways which lead to routing errors:

http://tools.geofabrik.de/osmi/?view=routing&lon=4.30371&lat=49.27784&zoom=6

and there are plenty more views. Have a look and do some cleaning, please!

Comment from Marián Kyral on 17 July 2017 at 11:36

JOSM has many controls and data validations on "commit". I'm notified when something seems to be wrong and I can fix it immediately. iD has nothing like this.

Using of any external tool that is updated next day, is not much comfortable. You need to not forget and have a time to do it.

There are some possibilities: 1) implement basic validations to iD 2) implement basic validations to OSM API 3) implement a special layer on main osm.org page, so any error became visible almost immediate.

Hide this comment

Comment from Little Brother on 17 July 2017 at 14:36

To Marián Kyral.

You are exactly right of course. I use JOSM exclusively with its useful error detection capability. I have not checked it but I suspect that most errors are generated using Potlach and ID. JOSM is so easy to use. It makes me wonder why a lot of contributors don't graduate to it.

Hide this comment

Comment from Marián Kyral on 17 July 2017 at 16:05

I know several advanced mappers that tested JOSM and they don't like the way, how it works and think that JOSM is user unfriendly.

Hide this comment

Comment from Richard on 17 July 2017 at 22:10

most errors are generated using Potlach and ID

"Why can't contributors check/correct their own work!"

It's called "Potlatch". With two 'T's. You should use the validation tool in your browser (the spell-checker). ;)

Hide this comment

Comment from Warin61 on 18 July 2017 at 06:17

JOSM can only check what it has. If I don't download everything in an area I'm working on I can break relations and not get any warning from JOSMs validator.

OSMinspector is one of the QA tools .. there are others too. I do use OSMinspector (keeping track of my own errors!) .. and I fix some errors there (mine + other peoples).. sometimes I make changeset comments to try and inform others of the problem so they might be more aware. Some times contributors are not aware of these QA tools, beginners don't get to 'see' them. Osmose tends to have things that I cannot fix without a lot of on ground work. So many of these have to wait.

Hide this comment

Comment from Marián Kyral on 18 July 2017 at 07:20

JOSM can only check what it has. If I don't download everything in an area I'm working on I can break relations and not get any warning from JOSMs validator.

Really depends on way, how are you working. But in iD or PoTTlach ;-) you have no checks at all.

Hide this comment

Comment from Marián Kyral on 18 July 2017 at 09:20

Sorry, it is PoTlaTch…

Hide this comment

Comment from SomeoneElse on 18 July 2017 at 10:12

Actually, that editor that you don't know how to spell has the QA cues built into the UI, not as an afterthought at upload time - see for example http://www.openstreetmap.org/#map=18/56.51881/-5.75593 , where the duplicate nodes appear as red pustules saying "fix me now!".

More seriously, the reasons why contributors don't correct their own work include:

  • They're new editors, still finding their way. Everyone needs to be allowed to get it wrong a few times as they learn how to edit OSM (whatever editor they are using).
  • No-one has actually told them that something is a problem. If people don't know about QA tools they won't know that they exist. The best way to get in contact with new mappers is to comment (politely!) in a changeset discussion. Anything that just sounds like "you're doing it wrong!" is unhelpful, even if they really are doing it wrong.
  • The QA tool is wrong. JOSM's validator is good, but still "cries wolf" quite a lot of the time, because not everywhere in the world is like Germany.
  • What's been uploaded is only the "first draft". For example, last night I uploaded this. It was important to upload it because it's a major change to the area, but it's far from complete, and there will be many things in there that need refining based on further survey. JOSM's validator there finds 1 error and 186 warnings, most of which are indications that further survey is needed (e.g. highway crossing building is obviously wrong, but which is in the right place?). Some of what JOSM's validator says there are just completely invalid, and it's up to the mapper to work out which and why.
Hide this comment

Comment from Marián Kyral on 18 July 2017 at 10:55

Sorry. Initially I've spoken about iD only. This is the editor recommended to beginners and there are no validation checks. So new user does not know, that something is wrong. Editors should be more pedantic and help users with common data errors as much as possible.

The best way to get in contact with new mappers is to comment (politely!) in a changeset discussion.

Nice idea, but they mostly doesn't answers to these comments :-(

Hide this comment

Comment from SomeoneElse on 18 July 2017 at 11:03

Maybe those comments weren't always polite :)

More seriously, a comment such as "you have broken a multipolygon here" will not communicate anything to someone who does not know what a multipolygon is.

Of course, there are genuine issues with e.g. MAPS.ME users not replying to comments. Part of that is because MAPS.ME users sometimes don't even know what OSM is, so to find it "talking to them" is very unexpected. Also matching the language of the comment to the language of the mapper is important (not everyone speaks English, and web translation these days is good enough that they'll at least get the sense of the problem, so there's no excuse not to).

Hide this comment

Comment from Little Brother on 18 July 2017 at 18:01

All the above comments are correct (even the one about spelling, but that does not bother me at all). Using OSMI and Bing I do a lot of correcting and when I see a consistent error being made I do contact the pertinent contributor (with a link to OSMI): Only about 5% reply and in these cases most are thankful for the information.

A message of mine looked like this.

On 2016-02-13 15:53:19 UTC Little Brother wrote:

Link

http://tools.geofabrik.de/osmi/?view=geometry&lon=-102.39286&lat=31.85214&zoom=15&opacity=0.53&overlays=self_intersection_ways,self_intersection_points

This is not the correct method for entering a bus route. See:

http://wiki.openstreetmap.org/wiki/Relation:route#Road_routes

Have fun

Thank your for letting me know. I will go in and correct

In this case the contributor did reply but check the date (Feb.2016) and the link (today): Nothing has been done. Question: Should I remove the data?

Hide this comment

Comment from Alan Trick on 19 July 2017 at 18:44

iD has validation checks, they're just relatively simple (like unmarked ways, and points without any tags). I think its validation features should be enhanced, but probably not too much (we really don't want people to start faking data just to pass a validator).

In theory, it would be nice if everyone could JOSM, but it really sucks with high-DPI screens, and it's only user-friendly if you already have a fairly solid understanding of how OSM works, which is a total no-go for most people who are not armchair mappers.

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

    [Text](URL)
  • Image

    ![Alt text](URL)

Login to leave a comment