[OSM-dev] mapping quality program version 2.0 published

[Steve Hosgood]

I'm new here (new to the dev list, but over a year's time mapping for
OSM). So please be gentle with the flamethrowers!
>
> Gary68 wrote:
>   
>> hi,
>>
>> today I published a new version of mappingquality.pl here:
>> http://wiki.openstreetmap.org/wiki/User:Gary68 
>>
>> the program takes an osm file and gives you 
>> - statistical data of "mappedness" (mapping quality) in HTML and CSV
>> - list of unmapped or sparsely mapped places TXT
>> - street directory for each place the program found (nodes and ways) CSV
>> - simple map that displays the area and lists key figures right where
>> they come from.
>>
>>     

This looks like a fine attack on a problem I've been pondering recently
with the rise of routable Garmin (and other) maps from OSM data.
Basically, we don't (currently) seem to have a system in OSM that can
tell you if a given part of the map is any good or not.

There was some talk on the "todo" list of the wiki a while back that it
might be nice for mappers to be able to request alerts should anything
change in areas of the map that they might consider "local" to them. It
sounded like a good idea, but I've not noticed anything come of it yet.
Then we have the reverse problem (which Gary68 seems to be considering)
- where someone not familiar with an area might like to know how
complete the map is.

( BTW: There are great swathes of central Wales and the highlands of
Scotland where the absence of roads for many km in every direction might
still indicate that the map is 100% complete and accurate for that area!  )


So: it struck me that if the OSM dataset could be associated with a grid
of (say) 1km x 1km tiles, then all or most of my wishlist could be
accommodated. Properties associated with each tile of the grid could
indicate who wants alerts on changes, and how well the local mappers
think they've dealt with that tile. Not just that, but you could
consider splitting the 'quality' indicators so that you could have
categories for major roads, minor roads, railways, cycleways, footpaths
etc. If you're just driving through a place you might not mind using a
tile of only 20% quality on minor roads as long as it's close to 100%
for major roads and motorways.


A bit more thinking, and I realised that the 1km x 1km grid idea might
be re-inventing a wheel: there already is a concept of rendering-tiles
used by tiles at home (and maybe Mapnik). If those tiles physically 'exist'
in the layout of the databases, could they be used to implement a
reporting and Q.A. system?


Just a few cents worth.
Steve Hosgood