On Mon, Dec 7, 2009 at 12:59 AM, Anthony <o...@inbox.org> wrote: > On Sun, Dec 6, 2009 at 11:37 AM, Eugene Alvin Villar <sea...@gmail.com>wrote: > >> Now whether one set of 20 nodes or a different set of 20 nodes better >> represent the shape of a road is a matter of creative subjectivity. Neither >> set is more mistaken nor more inaccurate than the other. >> > > What set of nodes constitutes a "best fit" to a given shape with a given > number of points, is fairly objective. You may "creatively" choose > something other than the best solution, but again, I don't think that > constitutes copyrightability. Not within context. (If you intentionally > chose something other than the "best fit", for something sort of stylistic > purpose, fine, but I really don't see how that's applicable to road > mapping.) >
Sure, there is only one set of N nodes that best represents a particular shape, but our problem is determining what exactly is that shape in the first place. Our GPS-based methodolody is only accurate to so much that the particular shape can't be defined if you want to be objective (in an Ayn Rand way) about it. Hence, we OSM as a project cannot determine the "best fit" per your definition. Well, unless you specify an accuracy tolerance level AND the number of nodes for each geographical feature. But then, the selection of both metrics for each geographical feature can still be considered a creative selection since they will be both arbitrary. Moreover, unless there is a ridiculously easy and systematic way of determining the "best fit", mappers would not bother doing it and will always use a subjective and personal criteria to determine the "good enough fit". This good enough fit is "fit" for OSM purposes and just because it's not mathematically proven to be the best fit doesn't make the data useless. Thus, the good enough fit (for increasing levels of good enough as time passes by) is still a product of a creative process deserving of copyright. > For practical purposes, we can't add an infinite number of nodes or should >> even add 100 nodes to represent a perfectly circular roundabout, >> > > We could, however, introduce a "arc" tag. And if I was better at making > proposals (and/or the OSM processes were better at accepting proposals), it > would probably already be introduced. To represent an arc, you only need > three points (start, end, and any third point on the arc uniquely defines a > triangle which is circumscribed by exactly one circle). This could even be > made backward compatible. Just split the way at the beginning and end of > the arc and put "arc=yes". Renderers that don't know about arcs would use > three points (or four, or five, or whatever). Renderers that do know about > them would use as many as is necessary for the resolution of the image. (In > the case of an arc=yes tag with more than three points > > Of course, I can't copyright this idea... So you're free to use it if > you'd like with or without attribution to me. > Maybe a perfectly circular roundabout is not the best example. How about an S-shaped road? Sure, we can add Bezier curves but we go back to the argument just above.
_______________________________________________ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
