Re: Don't Make Me Think (About Linked Data)

On 7/2/13 9:34 AM, Manu Sporny wrote:
> For those of you that continue to be unconvinced that some very educated
> folks think that RDF is too complicated, this blog post may be enlightening:
>
> http://berjon.com/blog/2013/06/linked-data.html
>
> My response:
> http://berjon.com/blog/2013/06/linked-data.html#comment-943596712
>
> Robin's response:
> http://berjon.com/blog/2013/06/linked-data.html#comment-944074845
>
> Robin is very smart, he builds a ton of very useful tools for the Web
> (like ReSpec, which is what we use for writing W3C specs), and is one of
> the W3C HTML5 editors. He truly understands the Web stack (more than
> most everyone on this mailing list, I'd imagine) and is very critical of
> RDF. He's not alone. It is this sort of discussion is the driving reason
> that we wanted to push RDF to the back of the spec, as something that
> JSON-LD is compatible with, for the people that care about that sort of
> thing (most don't).
>
> I'll start posting more of these sorts of discussions to this mailing
> list since it seems that there are a number of RDF WG members that are
> unaware of the amount of push-back to RDF from seasoned Web developers.
>
> This discussion also underscores my reluctance to talk about RDF at
> the beginning of the JSON-LD spec. We're damaging the simple and concise
> message that was JSON-LD a year ago. The spec is moving toward the
> current consensus of the RDF WG and JSON-LD CG, but speaking as one of
> the primary creators of JSON-LD, I'm pretty unhappy with the changes -
> they make the spec worse, not better.
>
> -- manu
>
Manu,

As I see it, here's the problem:

We have a concept and an implementation of said concept. These two 
things are distinct. Unfortunately, RDF is always afflicted by the 
problems that arise from conflation of concept and technical 
implementation.

Note: RDF 1.1 is a big improvement on prior efforts re., separation of 
concept and implementation details.

Breakdown of problematic narratives over the years:

1. it used to be you weren't RDF compliant without embracing RDF/XML
2. now its about not being Linked Data compliant without embracing RDF* 
(whatever that really means) and SPARQL (however that makes sense) .

Problem:
At this point in time (even as I type) you can't easily map the RDF 
specs to Linked Data since URI de-reference isn't part of the RDF specs 
(which isn't a minus re. RDF the concept).

The challenge (been so for years): how to communicate concepts that 
underlie RDF without drifting into implementation details. Basically, 
how do we somehow take the top-down as opposed to bottom-up approach etc..


Will this be issue ever be resolved? I don't have the faintest idea, 
especially as even broaching this matter remains a potent permathread 
vector, of the worst kind.

Little suggestion:

You can make a conceptual guide (even a white paper) about JSON-LD 
that's that's distinct from a technical spec for Web or RDF developers 
endorsed by the W3C. Only technical specs need to be approved by the 
W3C's RDF WG (since you seek their endorsement).

Once people grasp the concepts, RDF becomes less of a distracting problem.

-- 

Regards,

Kingsley Idehen	
Founder & CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca handle: @kidehen
Google+ Profile: https://plus.google.com/112399767740508618350/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen