Re: Should tags be URNs? (was Re: Proposal: 'tag' URIs)

At 10:25 AM 5/2/2001 -0400, Al Gilman wrote:
>I would say that the way one would use a 'tag' has been inadequately developed
>so far.   The activities that handle tags should be identified, in particular
>where they yield benefits, and where one is investing effort in creating
>them.

'The activities that handle tags' are, in my case, 'resolution': getting a 
web resource from reading an identifier on, in or near a physical object or 
in some content. I have a paper about this at 
http://www.champignon.net/TimKindberg/IDResolution.pdf.

>\
>If you review the current popular uses of real tags, a common function of tags
>is to enable relative strangers to associate an article with its proper
>context.

I can't understand what 'proper context' could mean. Which is 'proper' out 
of the (barcode of the) can of beans being resolved by (a) a Safeway 
employee to get a stock reading and (b) Frank Smith, at home, who wants to 
know whether the beans have been organically modified?

>If your encoding is not supported by a lookup service, it may not be
>used.  Competing symbolizations that _are_ supported by a lookup service
>may be
>used enough more so that your encoding will be buried and forgotten.
>
>The choice between LDAP and URN is debatable.  But at least LDAP is a viable
>option.  LDAP DNs already embody precisely the same "<who> calls <the item in
>question> <what>" logic as the 'tag' proposal.

Some reasons why DNs don't cut it for me:
(1) What are the distinguished names of objects within Al's cafe or some 
other small entity whose place in a hierarchy is not obvious or is 
ambiguous? I suppose Al would have to put his street address in the DNs to 
make sure no-one confuses his with the Al's cafe across town. Or maybe he 
could put his home address and his full name in ('Al Smith of 196 Morton 
St, .., trading as Al's cafe'). The point is: we'd have to define practices 
for descriptions that always produced a globally unique name for any type 
of entity (and not just relatively large organisations). I think that that 
would be error-prone and difficult, compared to the tag proposal.
(2) Do DNs provide, as a matter of course, uniqueness guarantees over time? 
Not the last time I looked.
(3) DNs are verbose: hard to put in cheap id technologies such as linear 
barcode symbologies, and hard for humans to copy and say to one another.

Cheers,

Tim.

Tim Kindberg

internet & mobile systems lab  hewlett-packard laboratories
1501 page mill road, ms 1u-17
palo alto
ca 94304-1126
usa

www.champignon.net/TimKindberg/
timothy@hpl.hp.com
voice +1 650 857 5609
fax +1 650 857 2358

Received on Friday, 4 May 2001 13:54:07 UTC