[OSM-talk] this has to stop: iD user mistakes all over the place

SomeoneElse lists at atownsend.org.uk
Wed Feb 11 03:14:49 UTC 2015


On 10/02/2015 23:38, colliar wrote:
> ... I am fed up with ...

... at this point it's probably worth mentioning that we've been here 
before:

https://lists.openstreetmap.org/pipermail/talk/2013-August/thread.html#67854

Unfortunately, experience suggests that there's relatively little that a 
discussion on on the "talk" mailing list is going to be able to do 
here.  There are essentially two sides to the argument - at one extreme 
new mappers "should never be able to break data" (and if they can't edit 
at all because they can't understand what they need to do, tough) and at 
the other new mappers "must have everything complicated hidden from 
them" (and if some complicated OSM structure breaks, tough).  Obviously 
you're not at one extreme and the iD developers aren't at the other, but 
there _is_ a difference of opinion here that it's not easy to 
reconcile.  If you want new mappers, you have to actually allow them to map.

If you've got specific examples of things that new users get wrong 
consistently (and even better if you can understand what they've done 
wrong and why) then I suspect that it would really help would be to 
raise an issue on Github about it, or add to an existing one if one 
already exists.

> * iD making it way to easy to delete objects but not offering an option
> to undelete them (is there any history information at all ?)

Whilst I'm in no way a fan of the iD user interface, even I had no 
problems finding the "undo" button.  I don't think that new mappers tend 
not to find it either, since an answer to the common question "what do I 
do if I get a conflict" is "undo back past the problem", and new mappers 
haven't said (on the help site or on IRC) "how do I undo"?

> * simply combining ways and merge nodes without any validation or
> warning about conflicts in tags or problems with relations

What might help here is to get details from the new mapper concerned of 
how they felt that they needed to merge nodes or ways.  The "merge" 
operation is fairly visually obvious when it happens; what's not so 
obvious is that the resulting merged node with semicolon-separated tag 
values isn't particularly useful in OSM.

There are a couple of "merge" Github issues; it may be that they already 
describe the problem that you are referring to here.

> * not telling the user about the importance of all tags, even unknown to
> the software and allowing user to communicate with user of the last
> change of the object

I suspect that this comes down to the "two sides to the argument" 
mentioned above - the idea is that new mappers shouldn't have to worry 
about "all tags" (or indeed, where possible, tags at all).

>
> Any plans of supporting lanes-tagging-system ? Otherwise there will be
> even more complains in the future.

This sounds like https://github.com/openstreetmap/iD/issues/387 to me.  
That's probably the best place to explain what you'd want the end result 
of a "new mapper knowing about turn lanes" would be.

>
> Is there anyone taking care of mistake made by iD users and documenting
> the most common ones to either better explain how to avoid them and/or
> fix the software ?

Back in 2013 I did have a look, and came up with this:

https://lists.openstreetmap.org/pipermail/talk/2013-August/068018.html

Since then the "Thing X changed to thing Y" problem has been much 
diminished by the fix for iD issue 542.  "POI added without a main tag" 
is still pretty common, and "unexpected deletions" are rarer than the 
were (perhaps also because of the iD 542 fix).

The initial "who made what sort of error" analysis was in 
https://lists.openstreetmap.org/pipermail/talk/2013-August/067936.html , 
and note that in there iD new users made statistically fewer serious 
errors than P2 ones (or, on a very low sample size, JOSM users).  I 
don't have the numbers, but based on a gut feel since 2013 I'd say that 
currently the editor for which the highest proportion of new users are 
going to cause _widespread_ problems is probably JOSM.

> ... So far, I try to keep calm and rather save my changes and upload them
> later after solving conflicts instead of starting an edit war by
> reverting or uploading older versions but I spend more time with
> communication and investigating problems than actually mapping and
> resolving notes and I still have quite some gpx tracks and photos from
> over a year ago to map.
>

Supportive communication with new users is really important, so thanks 
for taking the time to do this.

I don't believe that OSM has an "iD users" problem; it has a "new 
mappers" one -  or more accurately, a "data far more complicated than it 
needs to be" problem which means even experienced mappers can have 
problems.  For example, have a look at this help question and the ones 
that it links to:

https://help.openstreetmap.org/questions/40792/editing-large-multipolygons-in-josm

Those were asked by an experienced OSM mapper who usually edits in JOSM 
- how's someone without an in-depth knowledge of how OSM data is 
organised or any of the OSM editors supposed to manage? Similarly, how 
are new mappers supposed to manage here 
https://www.openstreetmap.org/#map=17/52.14720/5.09810 ?

Cheers,

Andy

(neither a developer nor even a regular user of iD, but someone who does 
care about helping new OSM mappers get started by whatever means)




More information about the talk mailing list