Public SourceMage Discussion List

[Eric Schabell <eric AT schabell.com>]

On Tue, Apr 05, 2005 at 01:17:53PM -0500, Jeremy Blosser (emrys) wrote:
> On Apr 05, Seth Alan Woolley [seth AT positivism.org] wrote:
> > So far the main arguments against our proposal are:
> > 
> > 1) difficulty searching by category in filesystem
> ...
> > For point 1, I'm philosophically opposed to the very "idea" that a 
> > heirarchical classification system can be made close to any general 
> > ideal.  They can only be made to specific ideals, and there's no way 
> > we'll be able to all agree on our ideals.  So I think it's better to 
> > break us from the restriction of attemtpting to make a coherent 
> > directory classification system intended for general audiences and 
> > instead focus on one that allows gurus to be assigned to one or more 
> > directories, which is the specific use with which we actually use the 
> > directories for.
> 
> The point is not to have a completely perfect ideal taxonomy.  As you've
> noted, taxonimists and information architects gave up on that kind of thing
> some time ago.  The point is to have something that is both consistent and
> meaningful, supplanted where it isn't perfect by documentation and (if
> absolutely necessary), other tools.  Anything that is presented to users
> needs some kind of taxonomy so they at least have a place to start.  This
> applies to websites and is why the unix world has (and sourcemage generally
> supports) the LFS, desktop file spec, etc.
> 
This is rather short cutting the world of information architects (as a
member of such a research department I couldn't in good faith let this
pass). Ontologies live on today, are in full development and an active
part of Aritficial Intelligence research everywhere... long from 'given
up some time ago.' The key is to pick one and stick with it, which is
what smgl has mainly done. 

But enough from the cheap seats out here! ;-)
erics

Attachment: pgponAl9gEzkg.pgp
Description: PGP signature