[Talk-GB] Update to OSM Analysis
peter.miller at itoworld.com
Sun Feb 6 22:27:29 GMT 2011
On 3 February 2011 17:32, David Earl <david at frankieandshadow.com> wrote:
> On 03/02/2011 14:30, Tom Hughes wrote:
>> I wonder how much of that jump was people blindly copying names from the
>> OS anywhere it showed a difference...
>> That's exactly what happened round me. I had started work on surveying
>> the differences to determine which was correct when another contributor
>> came along and just blindly copied in the OS names thus making it
>> impossible to tell what I still needed to go and check!
> Yes, I had that too. Except in my case they duplicated the roads the ones
> I'd already done - assuming there wouldn't be anything I guess - (nearly all
> of which were correct, and theirs in many cases wrong - because on a
> straight road you can't tell where one name ends and another begins on
> OSSV). Took me ages to sort out.
> But that's not what Peter is suggesting. I'm in two minds about that.
To my mind one the main benefits of the bot I have suggested is the reliable
attribution of content that comes with it. When a name is added from os
locator then there is always a 'source:name=os locator' added as well so
anyone coming along later knows. When geometry is added from os vector
district then there is always a 'source:geometry=os vector district' and
also a 'surveyed=no'. If left to keen tracers we are very unlikely to have
that level of detail.
When I have found discrepecies without that detail it has been hard to
determine if the name field in osm is a miss-typing of os locator or if it
is actually what the sign says or it is a mis-typing of what the sign says.
With good attribution we would know exactly what had come from the os and
there would not be additional typos introduced if that was the source.
Fyi, we are close to releasing a bunch of useful new slippery map rendering
views for debugging and for viewing 'invisible' content. This will include a
view that will be useful for identifying areas needing ground surveying, ie
with surverved=no etc.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Talk-GB