[OSM-talk] Streetcomplete and Geographically Large Changesets

Mateusz Konieczny matkoniecz at tutanota.com
Fri Jul 23 21:32:46 UTC 2021

Jul 23, 2021, 22:40 by miketho16 at gmail.com:

> I recently started using OSMCHA to monitor changesets in my local area.  There have been a number of cases where a changeset from Streetcomplete has shown up in my feed, not because an edit was actually made in my local area, but because the changeset spanned a large area (.e.g. half the US).  When I have commented on these changesets and I get a reply, the mapper has informed me that this is an issue with Streetcomplete and they (the mapper) cannot do anything about it.  
> 1)  Is this really a characteristic of Streetcomplete, or are people just misusing it? 
StreetComplete is supposed to be used while surveying - not for guessing.

The only case where valid editing would produce such situation are cases where
- someone disabled automatic uploads, switched to manual trigger
- and solved tasks while traveling across continent
- uploaded solutions from entire trip

solution: switch to automatic uploads (default) or trigger upload at reasonable intervals
of distance at long travels

- someone manually prefetched quests across continent
- had no internet connection while traveling
- once connected answers were uploaded in one go

solution: connect to wifi at reasonable intervals (or mobile internet)

> 2) Is there a work around?  
see previous section

> 3) What can we do to convince the developers of Streetcomplete to address this (if this is really an error in Streetcomplete)?  
> 4) Does anyone really believe the mappers using Streetcomplete are "surveying" the changes as the changesets state?  That would mean that (in the examples I saw today) they made some edits in one part of the country, drove for ~13 hours to another location (ok, they could have taken a flight, but we are still talking many hours if one includes the drive to the airport, security, the flight, and getting some sort of ground transportation on the other end), made some more edits, and only then did Streetcomplete decide to upload their changes (actually making several changesets but grouping by type of change rather than the time they were made or the geographic area in which they were made)?    
Sadly, some mappers ignore clear requests to actually survey. 

I recommend contacting user for explanation and to consider reverting their edits.

I reverted some StreetComplete edits where user was clearly guessing.

Making easier to spot such invalid edits, very low probability of such giant
bounding boxing happening in real editing is reason why autosplitting changesets
is not present.

It would mostly hide edits that are invalid and should be reverted anyway.

In general https://github.com/streetcomplete/StreetComplete/issues is the best place
to report StreetComplete issues or problems.

This specific one was discussed before and kept as is to make easier to spot people
misusing StreetComplete and damaging data by guessing rather than surveying.

> Does anyone really believe the mappers using Streetcomplete are "surveying" the changes as the changesets state?

Vast majority is actually mapping.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20210723/38a043c4/attachment-0001.htm>

More information about the talk mailing list