Re: Classification of ISSUE-57 change proposals

Hi Noah,

On Sat, 2012-03-31 at 11:24 -0400, Noah Mendelsohn wrote:
> On 3/30/2012 9:18 PM, David Booth wrote:
> > In other words, a resource-vs-description distinction will matter to
> > some applications and not others,
> 
> Right, but part of the power of the Web is that resources initially 
> deployed for some particular purpose may later be useful in other ways. The 
> closer you come to following good architectural practice from the start, 
> the better the chances that these opportunities will be there for you later.

Agreed.  But still, not matter how well one follows good practice, a
given URI will be unambiguous to some applications but ambiguous to
others that need a finer distinction along some previously unanticipated
axis.  

One might think that more precision in a URI definition is always
better, but in fact it is not, for two reasons:

 (a) Adding precision means adding constraints, and that limits the
  range of applications in which that definition can be used.

For example, one common mistake in ontology design is to specify (or
over-specify) the domain and range of each property, thus inadvertently
limiting the variety of applications in which that property can be used.

  (b) Adding precision means adding complexity, and that adds cost.

For example, an ontology that models the world as flat would obviously
be "wrong" in a scientific sense, and it would be useless for airplane
navigation.  But it may be fine for ground vehicle navigation, and even
preferable to a 3D ontology that is costlier to develop and use.

> 
> On the other hand, I don't think there's any problem with >indirect< 
> identification. Saying "get me the description of the resource, where the 
> resource in question is identified by URI_A" is not the same thing as 
> identifying both resource and description with the same URI. 

Agreed.

> It's clear in 
> this example that URI_A identifies the resource, not the description. 
> Indeed, it's possible that the description returned will self-identify 
> (using whatever means) as having URI_D_OF_A. When you don't care to have 
> two URIs, I think that's usually a better approach than just fudging the 
> distinction between resource and description of resource.

Indirection may help in some cases, but indirection also has a down
side, as TimBL pointed out a while back:
http://lists.w3.org/Archives/Public/www-tag/2007Jun/0012.html 



-- 
David Booth, Ph.D.
http://dbooth.org/

Opinions expressed herein are those of the author and do not necessarily
reflect those of his employer.

Received on Sunday, 1 April 2012 18:37:39 UTC