[OSM-talk] Change: How mature is OpenStreetMap?

[SomeoneElse]

On 06/01/2015 01:25, Michał Brzozowski wrote:
> * Software for monitoring OSM changes is still very rudimentary. I
> wanna be the f**king NSA. It's incredibly hard to check newbies' work
> quickly (eg. you have to load every changeset separately into OSMHV).

I'm not sure that it is "incredibly hard" - I rarely need to throw new 
users' changesets at osmhv - that usually gets saved for the wide 
changesets of people making things match JOSM's presets.  It's usually 
pretty easy to categorise new users into "adding new things; no 
problems", "adding things OK but haven't quite grasped $some_concept 
(like joining roads at nodes)" or "Oh dear they're really struggling".


> * Why do these newbies make so many mistakes?

Because it's difficult, dammit!  When I started mapping there was a 
large area of white space for several miles around my house - not even 
the roads were mapped.  It took a long time to get the hang of things, 
but while I was doing it there were no local mappers breathing down my 
neck saying that I was "tagging for the renderer" or similar.

We have to give new mappers the time to get the hang of things, and 
offer help when required, but constructively and not just saying 
"your're doing it wrong".  One of the sad things about OSM is that many 
people are willing to fix the _data_ but not to fix the _people_ - if 
you look at the changset history anywhere you'll often see quite wide 
changesets with descriptions such as "fix typo" - but rarely are the 
people making these changes going back to the original mappers 
explaining the best way to map a certain feature.

> The documentation is a
> mess, editor presets are incomplete (whereas they should include all
> approved and other widely used features)
>

Sometimes we forget that real life is complicated.  It's not a simple 
case of "tag X or tag Y" - something might be a pub, or a restaurant, or 
somewhere in between, and sometimes what might be the best category can 
change.

We saw it recently where well-meaning people tried to mechanically 
change "wood=deciduous" to "leaf_type=broadleaved" (most deciduous trees 
in the UK are broad_leaved, though some aren't - for example 
http://www.openstreetmap.org/way/40614704 ).  At the weekend I went and 
had a look at this area:

http://overpass-turbo.eu/s/6SY

and it turns out that things are _much_ more complicated than how it is 
currently mapped (by me!) suggests.  There are at least four groups of 
"planting type" there (old-growth broadleaved deciduous on the SSSI, 
planted-for-forestry pine in neat rows, some "odds and sods" mixed 
deciduous between the pine plantings, and some areas that are virtually 
heathland).  No amount of "remotely changing tag X to tag Y" will 
capture that detail - you need to go there and have a look.

However, if a new mapper arrives at an area like this part of Clipstone 
Forest but blank and maps it all just as "some sort of woodland", 
perhaps even very roughly to start with, they've still made the map 
better than it was before.

Sometimes we forget that we were all new mappers once.

Cheers,

Andy