Is Linked Data useless?

The recent exchange between Kingsley and Dave Longley triggered an 
important question:

Given the current definition of Linked Data, which we seem to have 
consensus on, would each of us be interested in a specification that 
allowed one to do /only/ Linked Data? That is, if JSON-LD did not 
support unlabeled nodes, would we use it?

The clear answer in my mind, for our company, is "No". Linked Data is 
useless for our purposes if we can't do something as simple as 
expressing Linked Data and Non-Linked Data in the same document.

Linked Data makes sense if your data is inherently meant for global 
consumption. It does not make sense if your data is only meant to be 
consumed locally (within a document or graph, for example). The Web is 
filled with both types of data. If we make a spec that only deals with 
the first case and not the second, we will have created a useless spec.

So, why are we creating a definition for something that many of us feel 
is useless on its own? To provide an analogy, we could attempt to define 
"Green Computing" as the creation of no greenhouse gasses when operating 
a software program for a year. However, we know that doing so would have 
massively negative short-term and long-term consequences on the 
productivity of our world - or, "Green Computing" is just not practical 
for most software developers to employ.

Kingsley said the following:

"Structured data is important, its actually more important than Linked 
Data at this point in time re. Web evolution."

I agree with that statement. We've been spending a great deal of time 
wrestling with the definition of Linked Data, only to find that we have 
found consensus on a definition that isn't practical for the vast 
majority of us. It's a non-starter for Web developers.

What many of the folks in the group seem to find more interesting is 
this thing that we've temporarily labeled "Structured Data / 
Contextualized Data / JLO / JSON-LSD / etc.". That should be a clear 
indicator that something is wrong with how we're defining Linked Data.

That said, I'm starting to care less and less about Linked Data as we've 
defined it. I thought it was something that would be useful to our 
company. I'm starting to have doubts that, as defined, we would find 
much use for this "pure Linked Data" spec in our products. I don't 
necessarily care about theoretically pure definitions that do not result 
in useful specifications. The definition we have for Linked Data reeks 
of theoretical purity.

Does anyone on this list feel that their organization would utilize a 
"pure" version of JSON-LD (one that didn't support unlabeled/blank 
nodes) over a version of JSON-LD that did support unlabeled/blank nodes? 
If the answer is yes, please explain why that "pure Linked Data" 
specification would be more useful to you than the "mixed Linked Data 
and Non-Linked Data" specification.

-- manu

-- 
Manu Sporny (skype: msporny, twitter: manusporny)
Founder/CEO - Digital Bazaar, Inc.
blog: Uber Comparison of RDFa, Microformats and Microdata
http://manu.sporny.org/2011/uber-comparison-rdfa-md-uf/

Received on Thursday, 28 July 2011 03:52:08 UTC