[foaf-protocols] WebID - mandated syntax or market solution? was WebID Incubator Charter draft

peter williams home_pw at msn.com
Fri Dec 17 18:24:43 CET 2010


Kingsley shows strong tendencies towards realism, while having a nice turn of phrase that points to a more beautiful tomorrow – one that’s a bit more logical than that of today.

 

Henry on the other hands tend to be absolutist. But, it’s an absolutism of assumptions, from which one reasons. Its rigor, vs religion.

 

We might as well get the RDF issue on the table, too, seeing as that Dan’s main domain.

 

So, after 3 years here on the topic, I finally get it enough to like it. But, I also like ASN.1 macros and modules (which probably sabotages my hopes of ever publishing an RFC since ASN.1 is essentially deemed unAmerican). It was Henry, channeling berners-lee apparently, who insisted one studies the model, not the notations and syntaxes. Kingsley insistence that it should all be reducible to what one might play with in a spread-sheet … is EXACTLY the right tone to take. Between them, one seemed hard is actually something any 2nd year CS student should pick up with no problems.

 

Now that I get it for what matters to me (and there ARE now books that give one the skills now, that there were not when I first came on board 3ish years ago), I could not care less about its previous incompetent presentation (and all the silicon valley wars). I can even see some beauty in its abstractisms, and ability to describe both formal and intuitive aspects of a thing at the same time. The trouble is, I think you have to have the thinking power of Dan to then exploit it naturally, much like only a natural composer notates in blobs and stalks the Opera she hears in her head.

 

In webid land, what I care about is FOAF cards (in RDFa, so far) making semi-trustworthy propositions, hosted on agents that are only partially under the governance/control of cloud providers (ISP caches, DNS naming, SAN storage, amazon web services, Azure virtual machines…). The cloud issue is crucial, as it changes the hosting assumptions (and thus the control assumptions).

 

If this turns out to be is a progenitor of the full-power semantic web so be it; but that’s a side-effect. First, I want to stop carrying around 25 pieces of plastic in my wallet, and formulating 50 passwords…

 

 

From: foaf-protocols-bounces at lists.foaf-project.org [mailto:foaf-protocols-bounces at lists.foaf-project.org] On Behalf Of Kingsley Idehen
Sent: Friday, December 17, 2010 4:51 AM
To: foaf-protocols at lists.foaf-project.org
Subject: Re: [foaf-protocols] WebID - mandated syntax or market solution? was WebID Incubator Charter draft

 

On 12/17/10 3:46 AM, Jiří Procházka wrote: 

I outline a repeating pattern above re. our industry. RDF's futile 
> attempt to break this time tested pattern that has led to its negative 
> (typically) and enigmatic status (at best) over the years. Claiming to 
> be everything (markup - RDF, model - RDF, and logic - RDF) only makes 
> matters worse.

To be a bit blunt, it means you disregarded my original concern and
failed to understand the people this part of WebID should cater for.


What? I don't understand the profile of person WebID should cater for? 

I've spent 90% of my professional career in the middleware space (mixing and matching data and protocols). 

I've been integrating data (as described above) since the late '80s.




What you suggest, is that WebID should be a heterogenous system of
sometimes interoperable, sometimes not applications for years to come,
like RDF app land is now (I disagree that RDF tried to enforce one
mandatory syntax, that it failed and it failed because of it) and web
was pre-HTTP. 


So you are saying to me that the Web is 100% interoperable at all levels?

To be more precise, that HTTP is used consistently across the Web?




Why and how HTTP came out as a "winner"?


1. Data Representation is decoupled from Data Access Protocols
2. Data Representation is distinct from Data Presentation
3. Platform agnostic.

Again, HTTP isn't used consistently on the Web, most developers don't even understand or exploit content negotiation. It is a major success because of the points above which enable it to deliver untethered client-server computing without platform lock-in.

The emergence of Linked Data showcase how we can up the ante by combining URIs and HTTP as based for Distributed Data Objects. Network oriented Objects with de-referencable Identifiers at Network scale (intranet, extranet, Internet). 

There is a graveyard full of initiatives that preceded HTTP -- in the distributed computing realm -- that failed due to any combination of items 1-3 above.




 I would argue it
is because it seemed for most people the syntax which could by
understood by most clients, not because of its superiority, which is
exactly what I've been saying we should create and call WebID, like now
Web is understood as using HTTP.


WebID leverages HTTP, URIs, and Linked Data. That's inherently potent and doesn't need any syntax distraction re. RDF.

HTTP based Linked Data took off because HTML (no RDFa in it at the time) was used as an alternative representation to RDF for pages read by humans. A nuanced move from the LOD community exemplified by the DBpedia project. Prior to this effort RDF existed in its own litter vacuum. 

History is always the greatest teacher.




 
You think implementors can live in such syntax-reality-show system, well
yes, some can, but I say a lot of them refuse to.


So let's move them to RDF covertly?

Implementers ultimately seek the most productive path for implementing a clearly understood concept. They don't start with Syntax. 




 They want to trust to
rely on WebID like they rely on TCP/IP stack, which is impossible
without at least one mandatory syntax. You say choosing such mandatory
syntax would deter many people, I say a lot more will be deterred by
lack of it and the assurances it provides.


Well I disagree, and time will prove one of us wrong.




 
If you knew me a lot more, you would learn that I would let almost
anything for market to sort out, but for machine languages is simply not
practical nor efficient. There is a large crowd of people who would
stick out the flag "Hey we don't want to participate in this popularity
contest, lets choose something basic, simple which just works and can be
relied on, and call it WebID".


I don't really get your point. Primarily because I have no sense of how you've arrived at these conclusions.

My views are based on hard core experience based on many years of dealing with interoperability, where data access and integration is the prime focus. Middleware has been my focus forever, pre Web.






More information about the foaf-protocols mailing list