Re: Literals as subjects in Turtle (but not in the RDF model) [was: Inverses of RDF and RDFS predicates]

Hi Richard,

On Fri, 2012-05-04 at 15:35 +0100, Richard Cyganiak wrote:
> Hi David,
> 
> Since you're presenting this as some sort of conspiracy between a
> backward WG and lazy tool makers, 

My comments and suggestions certainly were not meant to disparage the WG
or tool makers, so I sincerely apologize if it sounded like they were.  

> it's worth re-reading the responses to the community survey that W3C
> did to determine this WG's priorities, and that received 127
> responses. Among all the proposed work item, this one got by far the
> most negative response:
> http://www.w3.org/2002/09/wbs/1/rdf-2010/results#xg13
> 
> On 4 May 2012, at 14:13, David Booth wrote:
> > BTW, this restriction is not just "silly" in an
> > aesthetically-displeasing-but-harmless kind of way.  It is *harmful*
> 
> Out of 127 respondents, 14 thought that removing the restriction would
> benefit them or their organization, while 59 thought it would harm
> them or their organization.

Hold on, I think you may be misreading those results.  Remember, the 1-5
scale in that survey was *not* "1=very harmful ... 5=very helpful".  The
scale was "1=not at all important ... 5=very important":
http://www.w3.org/2002/09/wbs/1/rdf-2010/results#xq13
Thus, presumably even an answer of 2 indicates that the respondent sees
*some* importance in the benefit that this feature will bring to their
organization.  By that count, 45 saw some benefit, while 44 saw no
benefit.  (Spreadsheet attached.)

But I don't want to quibble about the exact numbers.  AFAICT:

 - *Everyone* agrees that there would be work involved in making this
change, even if the change would be beneficial.  

 - A *few* believe that this change would be harmful even if there were
no cost involved in making the change.  

 - *Most* believe that this change would be beneficial in some way, but
the cost of the change must be weighed against the benefits.  

As of two years ago, significantly more people thought that the cost of
change would outweigh the benefits than vice versa, and that's why it is
not in the charter.  But these numbers could change over time --
especially as more tools support this change anyway, such as enabling
all SPARQL CONSTRUCT results, or INSERT statements to be inserted.  I
think it would be interesting to repeat this survey periodically.

> 
> > In fact, I think it would be helpful for the WG to put tool makers on
> > notice that this restriction is likely to be removed at some point in
> > the future, and they should plan accordingly.  Otherwise, we would
> > forever be stuck in a circular situation similar to what Nathan pointed
> > out, that we can't remove this restriction from the tools because it's
> > in the RDF standard, and we can't remove it from the RDF standard
> > because it's in the tools.
> 
> If this was true, then we wouldn't have any tools with Turtle support
> or named graphs support or with support for aggregates in queries or
> with JSON results in the SPARQL protocol. All of these started out as
> independent proposals or vendor extensions, and multiple vendors
> picked them up without “being put on notice” by some WG. W3C
> eventually followed the herd and commenced work towards making them
> W3C RECs.

True, I guess I was being overly pessimistic.  Fortunately (IMO) at
least some of the tools have gone in this direction already.  Maybe I
just have to wait longer before pushing for formal endorsement.

In the meantime, without this change, I *cannot* always treat properties
as invertible.  This *forces* me into special casing my code or adding
redundant, inverse properties.  :(


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

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