[Talk-GB] Adding a further 250, 000 UK roads quickly using a Bot?

Ed Avis eda at waniasset.com
Thu Feb 3 17:34:50 GMT 2011


Richard Fairhurst <richard at ...> writes:

>>The OSM model has worked well for densely populated, prosperous areas.  It's
>>not at all clear that it is working well for remoter ones.

>I spend half my week in Charlbury (which ain't densely populated) and the
>other half in Burton-on-Trent (which ain't prosperous). They're both
>complete, and certainly not entirely through my own efforts.

Oh, I don't mean to suggest that no well-mapped rural areas exist.  Only that
coverage is patchy, and that there are many where nothing has been done despite
several years of sitting with a blank map waiting for somebody to turn up.

How important this is depends on the use you envisage for the map.  If OSM is
to be used by individuals navigating round their local area, then a map with
50% coverage is half as good as one with 100% coverage.  Half as many people
can use it, but for those lucky enough to live in a covered area, the map serves
their needs.

If the example use of OSM is address lookup, for example by businesses, then
a map with only 50% coverage scattered across the country is useless.  For that
particular use it is far better to cover every part of the country, even if at
only a basic level of roads and street names, than to have a patchwork of
excellent areas and missing areas.

Personally, I tend towards the 'world domination' school of thought and would
like OSM to replace proprietary addressing products such as the PAF as soon as
possible, at least for some applications.  So I tend to think that what matters
is to get 100% coverage.  I enjoy micro-mapping and putting in footpaths as
much as anyone else, but I don't want the best to become the enemy of the good.

>Bing and OS OpenData are great ways of assisting survey, and in
>many ways remove the need for a GPS unit. But they don't remove the need for
>a survey.

I agree with this, but given the choice between a map that has the basic details
complete but needs survey, versus a map that has nothing at all and needs
survey, I would take the best we can do right now with what we have.

>I personally like to print out the OS StreetView maps for an unmapped area
>before visiting it, and to annotate it - it saves a bit of scribbling and
>helps me plan my survey. If I'm cycling along a residential distributor
>road, there might be a cul-de-sac branching off it where I can both see the
>street sign (to check against the OS) and see all the way to the end (to
>check there aren't any footpaths I'm missing). If so, that's just saved me a
>minute cycling up and down there - and, crucially, staved off the boredom.
>That's great. But I still need to eyeball it to check there's no footpath
>there.

This is the one thing that perturbs me too about using the OS data.  Back in the
days when we only had Yahoo, I would not tag the name on a way until I had
walked all the way down it checking for footpaths.  If I'd only explored part
of the way, I would tag the name on that part only.  Then the noname map was
a guide to unsurveyed areas.  But now there isn't such a clear indication.

So I did hesitate about adding name=Newton Road to a street which I had not
visited.  But then I considered the folk living on that street typing its name
into Nominatim and getting no results.  There is no reason to make them wait
for a thorough ground survey of their street before it appears in OSM at all.
We must make the best map we can now, and improve it later.

Also it has to be said that the quality of the OS data, at least for street
layout and street names, is pretty good and arguably superior to the typical
OSM ground survey (at least as things stood in London a year ago).

-- 
Ed Avis <eda at waniasset.com>

> 
> > In many countries the Yahoo imagery (and now Bing) has turbo-charged 
> > the project by allowing large areas to be approximately mapped using 
> > your excellent Potlatch tool
> 
> :)
> 
> There is a _lot_ which the tools can do to help and we're still only
> scratching the surface. Aware that I don't do as much mapping as I like, I
> try and console myself by thinking that an hour spent on Potlatch will
> hopefully lead to 100 more hours of mapping productivity for others... well,
> that's the theory!
> 
> cheers
> Richard
> 







More information about the Talk-GB mailing list