Re: The tone of the "JSON-LD vs. RDF" debate (was re: Sub-issue on the re-definition of Linked Data)

On 12 Jun 2013, at 05:49, Arnaud Le Hors <lehors@us.ibm.com> wrote:

> Hi all, 
> 
> Although I haven't been able to get involved in the discussions on this (because I can't keep up with the volume), I'd like to say a couple of things. 
> 
> First, as chair of a WG that sometimes suffers from the same problem I can only support Markus's request not to get personal. This is just not helpful. 
> 
> Second, I think the JSON-LD group had the right idea in trying to provide people (so called "web/JSON developers") with a framework that lets them access RDF stores without knowing anything about RDF. I personally got feedback from someone at SemTech last week that made that very clear. His personal experience in being able to deliver RDF content to his customers using JSON-LD even prompted him to ask me why the LDP WG doesn't make JSON-LD the defaut/mandatory format for LDP. 
> 
> This being said, I do feel like the group went a bit too far in not even having RDF as a normative reference. Like Pat, I don't really care that this be done in the intro but I find it odd that it goes as far as it does in trying to distance itself from RDF. So, I have to ask: do you guys really think the target audience - "web/JSON developers" - is going to learn about JSON-LD by reading the spec? My guess is it won't. 
> 
> I believe most web/JSON developers program based on documentation they read specific to the service they want to use. So, as long as that documentation doesn't bother them with RDFy types of consideration and as long as the programing model seems natural to them you'll achieve your goal. 
> 
> So, as Pat said, I'd rather we don't go too far in trying to lure that crowd. The deliverable still ought to be a spec that live up to the expectations one has of a W3C standard. 

+1

Being able to rely on the huge work put together by the RDF group, such as the RDF Semantics specs, allows 
one at least to have discussions that can be informed by some of the best work in the field. There are difficult
issues that would be nearly impossible to discuss without these tools and it would be all to easy to get them
wrong without. 

So just as most Turtle devs don't read the RDF semantics specifications, having the spec explicitly tied to the
semantics means that in the rare cases when difficult issues do arise, one can refer to them, study them, 
and so raise the level of conversation considerably. 

As a Java developer I never closely read the JVM bytecode specs, but that does not mean that I am not 
happy they exist and that forums exist where these can be  debated clearly. It is that which guarantees
interoperability and why in the beginning I used Java.

In the same way I'd like to make sure that if I use JSON-LD I can have the same assurances as when using
Turtle.

> 
> Best regards.
> --
> Arnaud  Le Hors - Software Standards Architect - IBM Software Group

Social Web Architect
http://bblfish.net/