[OSM-talk] The completeness of OpenStreetMap

[Edward Johnson]

----- Original Message -----
From: "Iván Sánchez Ortega" <ivan at sanchezortega.es>
To: talk at openstreetmap.org
Cc: "Edward Johnson" <ed at cloudmade.com>
Sent: Wednesday, 25 June, 2008 6:47:53 PM GMT +00:00 GMT Britain, Ireland, Portugal
Subject: Re: [OSM-talk] The completeness of OpenStreetMap

El Miércoles, 25 de Junio de 2008, Edward Johnson escribió:
> One of the most interesting parts is how can we quote completeness of the
> map?

IMHO, the problem with any metric is that you cannot get an accurate number 
for "all". 

However, you can get a metric from a datasource considered "complete", and 
compare the numbers. You can suppose OSM is "complete" when the OSM metrics 
are equal or better than the other datasource's metrics.

Case in hand: Here in Spain, the OSM community has a very good relationship 
with the national mapping agency (IGN), so we can ask them for some numbers. 
Currently we've got some numbers regarding the total lenght of the road 
network. Also, the gazeteer for the electoral census (containing *all* street 
names) is near-public domain, so we could match how many street names OSM has 
got, compared to the official number of street names.

I'll get you the lies, the damn lies, and the statistics in the State of the 
Map, though. ;-)


Cheers,
-- 
----------------------------------
Iván Sánchez Ortega <ivan at sanchezortega.es>



Sounds like Spain would be a pretty good place to carry out some research then, if the stats are good and easily available. I'm hoping to some how model/predict/know from statistics the number of roads in any given area and then compare that to figures from OSM data. take a look at my blog (http://edwardmjohnson.wordpress.com/) and tell me what you think of my hypotheses on completeness. This is a big project and I would appreciate as many ideas as possible.

Thanks

Ed