[Geowanking] ***Location Linked Information***

[Rob Kyle]

On Tuesday 18 November 2003 7:05 pm, Daniel Smith wrote:
> >  3.	A distributed network of databases manage the information
> >nugget pointers which are
> >  URLs to actual information. Information nuggets themselves are
> >position/time/url tuples that
> >  lead  the viewer to further sources of data.
>
> Hmm, sounds like fun.
>

I like this a lot too :)   - partly because it's sounds very similar to some 
of the location based systems I've been thinking of:

http://spacetime.gurtlush.com

It's getting a bit out of date - but it too is all about referencing data 
using space + time co-ordinates.   Major difference being that I envision 
this as a 'file'-system, and not a web-service.

It scares me that Microsoft seem to be edging towards this via WinFS.  I like 
the sound of some of the ideas behind WinFS/Longhorn, but they don't have the 
same space/time focus.

Anyhow, good ideas in Longhorn are likely to be crippled with a need for 
backwards compatibility.  The Longhorn Hype is interesting, but the actual 
product will probably be very dull.  All the interesting bits will get left 
behind in the research lab - there's no immediate profit in the cool stuff.


Rob Kyle
======================
www.manchesterwireless.net



> I may have not read the paper deeply enough in a 15 minute lookover,
> but I don't see enough emphasis on Time.
>
> Example: lets say I use some geo-tagging system to point at a restaurant
> I like.  Next year, the restaurant goes out of business and
> becomes a nightclub.  Do we update the data and lose the
> history of what existed in a given space before?  What about
> a friend that keeps moving, etc?  In a nutshell, I want
> to search on a location for last week, or for next month
> (future event).
>
> When I wrote up the scheme at http://ThereAndThen.org [1], I was
> thinking about the two indisputable axis relating to any
> given event - time and space.  Without attention paid
> to time, any halfway interesting location becomes
> a blob of data to sift through - static points on
> a map that could be clutter, or could be relevant
> to a given search.
>
> Great paper though (even if it is from an ivory-tower
> institution ;-)
>
> Daniel
>
> [1] don't get hung up on the pure URL aspect of it.  It could
> just as well be RSS, chunks of XML w/Jabber, and so on.  The
> URL aspect is a very portable, bookmarkable representation.