[foaf-protocols] First WebID Teleconference minutes (July 27th 2010)

[Nathan]

Melvin Carvalho wrote:
> On 2 August 2010 13:54, Henry Story <henry.story at gmail.com> wrote:
>>
>> In the end what matters is that we can all interoperate, and that we can
>> build cool apps.
>>
> 
> +1 to interoperate and apps ...

The interoperability issue transcends WebID though, and WebID only 
solves part of a much bigger problem.

Efforts like JSON-LD and XSPARQL and indeed all others aim to address 
the other issues in the overall RWW.

IMHO 'WebID' is not one thing, what is commonly referred to now as WebID 
and formally as FOAF+SSL is one case for a group of different things.

What is a WebID?
- a WebID is a URI which Identifies an Agent, where upon dereferencing 
of that URI you receive machine readable data.
  Also being Human Readable is in some cases beneficial but not required.

What is this protocol?
- A party in a +TLS connection presents a Public Key together with a 
WebID, ownership of the Public Key is first established, then ownership 
of the WebID is then established.

To me, that's 2 distinct things, WebID needing specified first, then the 
protocol specified afterwards.

Anything out with the above comes in to mapping the protocol to specific 
set(s) of technologies (like RSAv3 Cert with subjectAltName, like HTTP+TLS)

Further, any stipulation of what classifies as Machine Readable Data (ie 
only this or that serialization of RDF can be provided when a 'WebID' is 
dereferenced) will further limit and inhibit the protocol (even though 
it doesn't feel like it). Again, this can be addressed by providing 1 or 
more mappings to common serializations.

I've said this many times, but we're trying to build a stable 
interoperable protocol on an unstable foundation. The SemWeb and Linked 
Data serialization and format issues are inherited, and that leaves only 
two approaches for 'WebID'/this protocol:

1: define most of it as abstract then provide mappings to techs.
2: fix the protocol to a single format of mrd.

2 kinda defeats the purpose unless you tie to RDFa. Anything in between 
will pretty much ensure the death of this protocol and certainly ensure 
the interoperability is not globally possible.

Best,

Nathan