RE: combination of location and identity

On Fri, 13 Jul 2001, James M. Polk wrote:

> >> I read your draft, very interesting. What we should also be aware is that if
> >> I document is cached, the cache should update the geo tags.
> >
> >No. The geo tags refer to the content - if the page describes a lake, the
> >tags give the location of the lake. Where the document is hosted or cached
> >does not affect this.
>
> And if that lake is Lake Michigan..... what coordinate does one give?

Whatever the author chooses. This (draft-daviel-html..) is for normal
folks to use, and half the time I expect they'll have significant errors
anyway - locations taken from gazetteers or zipcode polygons, etc.
Professional users would probably use something like GML, with a polygon
of the lake shore.

I'd recommend that a private individual give the position of their
residence to a few km accuracy on their homepage (i.e. rounded to the
nearest 0.1 degrees) assuming they didn't want to reveal their address,
while a retail store give the position of their front door to 10m.
Lakes, who knows? Accurate enough to differentiate it from a different
lake, I guess. This was an imaginary example in my mind "here's a picture
of Lake Woebegone" - in which case the position might be taken from GPS
when the picture was taken, or looked up in a gazetteer a few weeks later.

>
> There was discussion last year in the predecessor group to this of how accurate
> the reply is - and a concern came up with regards to where on a plane or ship
> does the coordinate (or Dataset value) specify.... with no consensus. The
> exactness of what the requester is doing with that reply is of great concern,
> agreed?

If someone is that concerned, I'm not sure that they are not too
specialized for this WG - at least, in my interpretation.

Normally, the coordinate on a plane or ship would perforce be the location
of the GPS set. If the requirements on precision are such that this makes
a difference, then one would probably have two or three sets together with
a geometric model of the ship that would allow the calculation of
position of any feature (fore anchor, aft anchor etc.) I guess that GPS,
or DGPS, is now accurate enough to do this, though I think that e.g. in
aircraft landing systems that relative coordinates are used with a
transmitter at the airport, which avoids problems with absolute accuracy,
datum conversion etc. (Many map datums in current use are significantly
different from WGS84 - hundreds of metres, or kilometers - so that there
are significant problems in trying to navigate to metre accuracy using
maps and charts rather than differential methods)

-- 
Andrew Daviel
http://huzizit.com - transferable ID for your stuff