On Fri, 24 Jan 1997, Paul Miller wrote:
> At the moment, I'm not convinced that an MCF fly through REALLY shows
> you anything of much use (although demo's like Andy's for eLib, and
> bits and bobs we've done here are still useful, of course), but it
> possibly points to a way forward...
I just wonder whether the MCF chaps at Apple have been reading a bit too
much Gibson; I can quite easily imagine that with enough sites hosting
metadata (be it MCF or DC or whatever) it would be possible to generate a
view of the original cyberspace as in Gibson's Neuromancer novell.
Imagine; each site has a block of site information metadata that defines
not only the usual name, organisation it belongs to, lat/long, access T&C,
etc, but also the prefered colour and maybe shape to render it with in a
graphical navigator. The size of the object when rendered might be based
on (say) the number of pages that it holds or the quality of its resources
(better quality == bigger object). The objects relating to sites could be
positioned with some network topological significance and, hey, lets be
wild and say that we might be able to represent the bandwidth in use
between sites in someway (some of the SNMP management tools for LANs use
pulsating line widths for example). The last bit is immediately useful to
spot network hotspots that you might want to avoid (or go to depending on
what you're up to).
Once you select an object you can fly inside and then see some structure
to the available data provided by the metadata available. Quality
reviewed objects could be big, or shiny or pulse or something, whereas
lower quality resources could be small or dull or translucant. You're
viewer might offer several ways to view all the metadata, eg clustered by
author, by keyword, by LCSH, etc. Finally you can fly up to a single
metadata object and either retrieve all the metadata it holds (that you
have permission to see) or the object itself (again if you have
permission, or your name is Case).
Whether this is useful or not, I don't know. The HCI chaps tell us that
concrete representations are better than abstract ones (if I remember my
Ergonomics lectures correctly), so making the site objects look like
buildings and the resource metadata objects look like (ooh) say library
shelves might be good HCI by giving people a concrete metaphor that they
can understand and relate to. Then again, it all depends on the user
really. I can't stand filemanagers and even though I use X Windows alot,
its mainly so that I can have multiple xterms running with command line
shells and use lots of UNIX commands which I can easily remember and type.
I'm therefore a "power" user.
Similarly I can imagine that concrete representations of available
metadata and data might well be useful to Joe Public but not too useful
for Josephine Librarian, who'll be much more at home (and faster) with a
command line interface full of shortcuts. But who can say. Unless people
like Andy and Paul play with the technology and see what it can do, we'll
never have the opportunity to stumble over that obvious, world changing
application for it. To your VRML engines gentlemen... :-)
Tatty bye,
Jim'll
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Jon "Jim'll" Knight, Researcher, Sysop and General Dogsbody, Dept. Computer
Studies, Loughborough University of Technology, Leics., ENGLAND. LE11 3TU.
* I've found I now dream in Perl. More worryingly, I enjoy those dreams. *
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|