[p2p-research] Henry Story on Open Distributed Social Networks

[Dan Brickley]

On Mon, Nov 22, 2010 at 8:08 PM, Samuel Rose <samuel.rose at gmail.com> wrote:
> FOAF is amazing. It's been around for a long time (since at least 2003
> if not earlier), and works really well. Yet, it's been largely under
> the radar

Thanks Sam! Yeah, we've kept things a bit 'relaxed' the last few
years, partly to cool down some unrealistic expectations from ~2004/5,
but also while other pieces of the technology landscape came together.
Without the likes of OpenID, OAuth, FOAF+SSL/WebID, and some machinery
for handling trust/provenance, there's only so much you can do with
FOAF on its own.

> It's implemented by ostatus (identi.ca and GNUSocial) and I think
> there is talk now of using it in diaspora. ELGG also supported it at
> least for a while. Semantic Media Wiki also supports, and so does
> Drupal, WordPress.

Yep, it's also supported in http://code.google.com/apis/socialgraph/
from Google (their social search machinery); in the Russian search
engine Yandex (http://company.yandex.com/press_center/press_releases/2008/2008-08-15.xml),
in Google Rich Snippets
http://googlewebmastercentral.blogspot.com/2009/10/help-us-make-web-better-update-on-rich.html
... though as far as I know not directly in Wordpress. I think they
emit XFN, which is a similar system from the Microformats community,
and that's great too.

I'm most excited lately to see FOAF and RDF playing a role in open
datasets. The Social Web API scene is very crowded and I think the
vCard-based Portable Contacts spec is doing fine work there. FOAF
tends more towards extensibility and wide-scale data merging, which is
getting much more interesting now we're seeing huge releases of data
from goverments, activists, libraries, ...

And yeah we started back in 2000 when I turned my homepage into RDF
and Libby Miller did the same to hers. When big sites like LiveJournal
started publishing FOAF a while later, lots of our cute early demos
simply fell over due to the scale needed. So it's great to live in the
modern day Web now with tools like Hadoop around to take the strain.
Lately I've been pushing user data (from FOAF and misc open datasets)
into Apache's (hadoop-friendly) machine learning tools,
https://cwiki.apache.org/confluence/display/MAHOUT/Recommender+Documentation
.... .... it's amazing the tooling you can get for free :)

Anyhow, I've been quietly lurking here on the p2p research list,
something somewhere gave me the impression that many of the (largely
unarticulated) thoughts behind FOAF fit with the mindset here, so I'm
happy if FOAF is of any interest in these parts...

cheers,

Dan