On 02/10/09 10:10, Andy Allan wrote: > The alternative to forcing arbitrary rules of consistency on our > volunteers is to acknowledge that OSM is in fact inconsistently > tagged, and chill out about the whole thing. Different people can then > experiment with different approaches to produce "consistent" datasets > tailored to their own needs. They don't need to be proprietary - in > fact, given the number of people around here talking about it I'd have > hoped someone would have stepped up and produced a tailored dataset by > now.
There's a giant assumption behind this, and that is that the information necessary to produce a tailored and consistent dataset has been preserved. Say there was an Andy Allan scheme of tagging which rated highways from 1 (biggest) to 10 (smallest). There's also a Gervase Markham scheme of tagging which rates them from 1 (smallest) to 10 (biggest). How does one produce a consistent data set out of that, without knowing the preferences of every OSM member as to whether they use Andy Allan tagging or Gervase Markham tagging? This example is obviously extreme to make the point, which is this: unless there is agreement on what the values mean for particular keys, information is lost which cannot be retrieved. If one person's highway=tertiary is another person's highway=unclassified, then what do you do? Gerv _______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
