Re: centralized vs decentralized extensibility

On 8 June 2016 at 16:59, Sandro Hawke <sandro@w3.org> wrote:

>
>
> On June 8, 2016 7:47:28 AM PDT, Melvin Carvalho <melvincarvalho@gmail.com>
> wrote:
> >On 7 June 2016 at 17:03, Kevin Marks <kevinmarks@gmail.com> wrote:
> >
> >> If you are pointing to centralised proprietary silos such as
> >> schema.org and facebook OGP as examples of decentralisation I really
> >> don't understand what you mean by the term.
> >>
> >> Just because they are using markup that if you don't look too closely
> >> you can claim as RDF does not make them decentralised. This is
> >> duckspeak.
> >>
> >
> >So it's been a while since I looked at OGP itself but I seem to recall
> >linked data being used.  The point I am trying to make is that facebook
> >return RDF / Linked data in their profiles and a ton of other data.
> >
> >What I have looked at lately is the Facebook graph API.
> >
> >http://semantic-web-journal.net/sites/default/files/swj282_0.pdf
> >
> >Schema.org plays nicely with other vocabs, just in the same way that
> >AS2
> >does, or am I missing something?
> >
> >What I am trying to get across is that bottom up design scales to the
> >whole
> >social web even if you have a few centralized curators.  Can anyone
> >make a
> >serious argument against this, either theoretical or practical?
> >
>
> Straw argument.


I am trying to demonstrate why the bottom up decentralized approach works.
Why do you think that is a straw [man?] argument?


> Even if it were true it wouldn't affect Webmention because it's deployed
> differently.
>
> The best argument for webmention to use RDF IMHO would be so an LDP
> container automatically works as part of a webmention endpoint, but LDP
> hasn't caught on and there are too many other bits that would be missing
> (e.g. Validation) if you did it that way.   Plus, if you have
> authentication, as you probably would with LDP, webmention is entirely the
> wrong protocol.
>
>     - Sandro
>
> >
> >>
> >> On Tue, Jun 7, 2016 at 3:06 AM, Melvin Carvalho
> >> <melvincarvalho@gmail.com> wrote:
> >> > Sandro recently made a point in favor of centralized extensibility.
> >> >
> >> > I would like to argue that this kind of centralization does not
> >scale on
> >> the
> >> > social web.
> >> >
> >> >> Personally, I feel like decentralized extensibility is a moral and
> >> >> psychological issue, but I'm well aware that the case for
> >decentralized
> >> >> extensibility is weak.
> >> >
> >> > I strongly disagree that the case for decentralized extensibility
> >is
> >> weak.
> >> >
> >> > Centralized standards have been tried a number or times on the
> >social web
> >> > and they have almost all failed.  My reasoning is that the whole
> >social
> >> web
> >> > is too vast for some central authority to please everyone.  You can
> >only
> >> > please a small group, and that leads to balkanization, which is
> >what we
> >> see
> >> > today.
> >> >
> >> > I do agree it is moral and psychological because it is top down
> >decision
> >> > making vs bottom up grass roots self organization.
> >> >
> >> >
> >> >> The vision is of a wonderfully free and open yet interoperable
> >> ecosystem,
> >> >> but in practice that doesn't seem to happen.
> >> >
> >> > Why do you think this?  The facebook open graph has proliferated.
> >> > Schema.org has proliferated.  JSON LD has proliferated.  These are
> >all
> >> > largely interoperable standards all along the same lines.
> >> >
> >> >> By far the greatest adoption of RDF happened when it was coupled
> >with
> >> >> schema.org, with only centralized extensibility.
> >> >
> >> > I think it's worth pointing out that facebook open graph is a
> >significant
> >> > adopter of linked data / RDF.  Indeed last time I checked all of
> >their
> >> user
> >> > profiles and graph are available as RDF.  There have also been
> >other
> >> > adopters of RDF orders of magnitude bigger than the usage I see of
> >> > webmention.  I'd welcome numbers, but from what I can tell
> >webmention
> >> > adoption numbers are statistically insignificant when compared with
> >even
> >> > minor RDF deployments
> >> >
> >> > I think the arguments that are being made for centralization are
> >just not
> >> > accurate.  We've been down this path before in the social web.
> >Hint: it
> >> > doesnt work.
> >>
>
>