On Sep 20, 11:04 pm, David Recordon <drecor...@sixapart.com> wrote:
> Hey all,
> I realize I've been a bit quite since I started this group a few
> weeks ago, but now have a bunch of cool things to share that we've
> been working on and thinking about. Check out the post (http://www.sixapart.com/about/news/2007/09/were_opening_th.html) and I'm
> happy to discuss it here. :)
Enjoyed reading your post. The new prototype looks excellent - really
looking forward to seeing what you guys come up with next. The thing
that interested me the most was your comment:
> It is good to be able to find out what is already public about you on the Internet
What do you think the current approaches to this might be?
One other comment:
> blog your thoughts with the tag or category "socialgraph"
I would advise caution against encouraging this term. I read Brad
Fitzpatrick's (excellent) essay on the subject, but if you look at the
term social graph, from his wikipedia link, it actually points to
"Social Network". Anyone that has studied Graph Theory to a
sufficiently high level will be able to tell you that a graph is
actually a special type of network, where the links are always bi
directional. We should be cautious about redefining Graph Theory
Anyway, your screencast was fascinating, what interests me is your
thoughts on discovery of information for a given digital identity
(URI?). Do you think XFN will become the main standard?
Obviously there are the traditional search engines as well as more
specific sites such as Spock or Rapleaf. I think what becomes really
important is being able to have tools which put the user in control,
versus just aggregating all of this data and making it world viewable
by default -- especially when working with email addresses. I
instead envision a tool where I user can use OpenID to login (thus
proving ownership over a node) and then being able to walk their
graph and explore it found via open markup technologies like XFN and
>> blog your thoughts with the tag or category "socialgraph"
> I would advise caution against encouraging this term. I read Brad
> Fitzpatrick's (excellent) essay on the subject, but if you look at the
> term social graph, from his wikipedia link, it actually points to
> "Social Network". Anyone that has studied Graph Theory to a
> sufficiently high level will be able to tell you that a graph is
> actually a special type of network, where the links are always bi
> directional. We should be cautious about redefining Graph Theory
Any thoughts on a better tag?
> Anyway, your screencast was fascinating, what interests me is your
> thoughts on discovery of information for a given digital identity
> (URI?). Do you think XFN will become the main standard?
I don't think there will really be one main standard. There are a
lot of different types of data to represent, varying security
settings, etc. I do think that both Microformats and FOAF solve
important parts of these problems though would also like to see some
sort of hybrid FOAF, XFN, hCard format which is easy to represent in
JSON, YAML, and XML.
Take note of the coinage of "SNAP" as an acronym for "Social Network
Aggregation Protocol" here:
J...@Germuska.com * http://blog.germuska.com
"I felt so good I told the leader how to follow."
-- Sly Stone
On Sep 20, 11:07 pm, "Joe Germuska" <joegermu...@gmail.com> wrote:
> On 9/20/07, Plasticmind <je...@plasticmind.com> wrote:
> > Actually, maybe OpenSNAP makes more sense...
> Take note of the coinage of "SNAP" as an acronym for "Social Network
> Aggregation Protocol" here:http://openfriendformat.com/
> Joe Germuska
> J...@Germuska.com *http://blog.germuska.com
blog your thoughts with the tag or category "socialgraph"I would advise caution against encouraging this term. I read Brad Fitzpatrick's (excellent) essay on the subject, but if you look at the term social graph, from his wikipedia link, it actually points to "Social Network". Anyone that has studied Graph Theory to a sufficiently high level will be able to tell you that a graph is actually a special type of network, where the links are always bi directional. We should be cautious about redefining Graph Theory definitions.
Very good point. When the adjacency matrix is non symmetric, we can
have an undirected graph. Im used to the term graph to mean
bidirectional as this is where most of the theory gets interesting
(clique finding, eigenvectors etc.). But technically you're
absolutely right. I guess what I wanted to highlight is that network
and graph are techical terms with specific meaning, and rich theory,
that could be pretty valuable to this kind of conversation. I think
the concept of network (a network being a graph with weighted edges)
will become more interesting as the question goes from "Is person A a
friend of person B" (yes/no = graph) to "What is the relationship
between person A and person B" (score = network).
So I take by my ealier comment, socialgraph seems a reasonable tag,
however I like the emphasis that people have placed on an open system,
as it strikes me that this will be the long term solution.
- isn't specific to a particular model, the way 'graph' is, but still
- allows for a variety of protocols, as will probably be the case
- suggests that there's not necessarily one big MOASN (Mother of All
- includes the 'open'
- leaves room for OSNP (OSN protocols), OSNS (OSN Software) etc
While I'm writing, this seems like a good time to suggest revisiting the
'Semantic Social Network' (SSN)
Which gives us OSSN (Open Semantic Social Network), which is where we
want to get in the end.
And I look forward to an end user tool (as opposed to an academic proof
of concept) that does exactly that.
Almost every social network now imports contacts from gmail, web mail
services and CSV as a way of finding existing members. Who's going to be
first to allow import of FOAF to do the same thing?
Julian Bond E&MSN: julian_bond at voidstar.com M: +44 (0)77 5907 2173
Webmaster: http://www.ecademy.com/ T: +44 (0)192 0412 433
Personal WebLog: http://www.voidstar.com/ skype:julian.bond?chat
*** Just Say No To DRM ***
I only wish that "Almost every social network now imports" from CSV!
I've found all too few that do.
I find it very annoying that there are all these sites who will
happily ask for my gmail password to scrap my address book, but don't
have a way for me to upload a CSV, a file of vCards, give it a vCard
On Sep 22, 1:13 pm, Mark Atwood <fallenpega...@gmail.com> wrote:
I think the OAuth idea is right on. I think it can live in the social
graph ecology, because it doesn't have much effect that I can see on
the way that social graph would map social networks that are published
under open standards.
On Sep 22, 3:03 pm, Nate Westheimer <Nate.Westhei...@gmail.com> wrote:
> How does the announcement of OAuth play into this debate/discussion?
Couple of things worthy of note that post-date that article:
Firstly, re the concern that RDFa "only works with XHTML 2". While it
came out of the XHTML2 efforts, RDFa is now decoupled from XHTML2. You
can deploy it on Web pages together without using any markup idioms
that'll freak out contemporary browsers. For example, my colleagues at
Joost have put a little RDFa in pages such as
http://www.joost.com/044000l ... and these can be validated using the
updated W3C validator, eg. see
Secondly, Evan's article doesn't mention GRRDL, which recently got
finalised as a Web standard. GRRDL is a lightweight bridging
technology that allows you to associate a transformation (typically
XSLT-based) from HTML-based markups into RDF's "classes, instances and
properties" graph data model. This addresses one of his key concerns,
the need to be able to map from microformats into more decentralised
and extensible models without creating unnecessary duplication of
effort and data.
That's one effort I know of already underway to create a decentralized
social network map. Not only do users own their own data, either on
their own site or SocialGrid -- it also works using any web search
engine (Google in this case).
Cool! Thanks very much for that enlightenment!
I've also been looking at some ideas about how to aggregate "ratings"
from many different sources, and your pointers might be helpful there,
On Sep 22, 4:26 pm, "Dan Brickley" <danbrick...@gmail.com> wrote:
> On 22/09/2007, Sam Rose <samuel.r...@gmail.com> wrote:
> > Check this out about some of the issues that are emerging in the co-
> > evolution of RDF-related (FOAF) and XHMTL-embedded
> > "microformat" (XFN) open standards:
> Couple of things worthy of note that post-date that article:
> Firstly, re the concern that RDFa "only works with XHTML 2". While it
> came out of the XHTML2 efforts, RDFa is now decoupled from XHTML2. You
> can deploy it on Web pages together without using any markup idioms
> that'll freak out contemporary browsers. For example, my colleagues at
> Joost have put a little RDFa in pages such ashttp://www.joost.com/044000l... and these can be validated using the
> updated W3C validator, eg. seehttp://validator.w3.org/check?uri=http%3A%2F%2Fwww.joost.com%2F044000...
If all sites did allow CSV import/export then transferring data (with
some cleaning & purging along the way) to a new site would be easy.
Lysander Meath Baker
On Sep 22, 1:13 pm, Mark Atwood <fallenpega...@gmail.com> wrote:
FWIW CSV import is pretty much perfect for FOAF import too, since the
SPARQL query language, being like a funny cut down webby version of
SQL, returns tabular result sets which can easily be formatted into
CSV (see also JSON output format,
http://www.w3.org/TR/rdf-sparql-json-res/ ). Of course it depends what
column names are required in the CSV etc., and what's available in the
FOAF. Our mailbox hash trick seems to have some traction even outside
of direct FOAF usage - eg Brad's 'Social Graph' slides show use of
hashed mailboxes, and last.fm seem also to use it, see
Hello? Facebook already does CSV import.
Friends, Find friends, Email Application Outlook, Apple Mail, etc.
Upload a contact file and we will tell you which of your contacts are on
This is pretty common since it's the only easy way to get data out of
Outlook and into the YASN. Note also that LinkedIn has CSV export.
I don't use Gmail for mail. but I've taken the decision to use GMail as
my Contact Aggregator. I load everything I can into that. It then makes
it easy to find the same people on a new YASN because they almost all
now support GMail import.
I had always looked at their request for my Gmail password and quickly
moved on - the idea that I might reveal my Gmail password to any other
organisation is absurd. Thank you for prodding me to read further down
It does seem as if this might work. No time to fight with it now, but
FaceBook tells me that I have some kind of problem to do with field
The field name mapping problem has also caused difficulties with my
attempts to sync my address file with Outlook and the application on
my "smart phone". I have never yet found an address book application
that has the field names I want or will allow me to add the ones I
need. Or has different rules regarding field length or character set
used. Any attempts to allow transfer of data between applications must
deal with these issues. For example, US addresses have a very set
format and require very few fields. UK addresses can be up to 6 (yes
that is six) lines long and so most US based contacts software can't
> >I join my vote in favour of CSV import/export. I keep all my data in
> >an Excel file (easily allows insertion of extra columns - eg who did I
> >send holiday cards to in 2006) and would dearly love to import a
> >selected section of it into FaceBook. Gmail does not hold my address
> Hello? Facebook already does CSV import.
> Friends, Find friends, Email Application Outlook, Apple Mail, etc.
> Upload a contact file and we will tell you which of your contacts are on
Another perspective.. What's the motivation for importing and
exporting stuff? Why couldn't all services just use an existing social
network and decorate it with their own stuff? The source would of
course be OpenID+XFN&FoaF. The services wouldn't need to track changes
in the social network as there would be only a single, distributed
source for it; the services would fetch the graph on the fly and cache
appropriately. Not a big deal. Naturally, all the services (the social
network*s*) would stay up-to-date and I didn't need to invite my
friends to all the different networks separately. And when I logged
into a new service, it would already know my social surroundings. I
can't imagine a better user experience.
Please, quit making redundant copies of the same stuff. :)
A fine sentiment. As a social network developer, I'm stuck with the need
to hold local copies or a local cache of quite substantial account
records. These records contain both common data about the user and data
that is unique to my system. There are lots and lots of functions that
simply would not be possible without having the data in a local
So now rather than querying a common store on the fly, I have to think
about synchronising my local store with some common store. Which is
where we came in with Brad's white paper.
So where do you think this common source would be and who would run it?
There's a number of options here.
- Google (or similar) scrape all the XFN, FOAF and whatever machine
readable data on the web into one huge store with an API. Yes, please.
- Plaxo (or similar) encourage people to lodge their data and network
with them. They then provide an API. This is already starting. If
there's a real need then services like this will appear.
- Wordpress (or similar) build modules that allow people to create and
maintain their own store with a common standard that everyone else can
work to. At this point see OpenID AX, oAuth, SNAP and any other people
here building standards.
It goes a little time consuming to describe this further, but I'll
give some pointers.
The social network should be distributed, just like OpenID. (It's
pretty fundamental for Internet-based systems that there are no single
sources.) For example, I could have links to different parts of my
social network at my openid-url http://janne.savukoski.name/.
(Although, there's no need to have this kind of a 'starting point',
either; there are also several services that let you compose that list
very easily.) That resource (at the url) should then contain links to
all my other openid/xfn-supported accounts as 'xfn:me'-links.
(Something like "<a href='http://jsav.livejournal.com/' rel='me' />".)
Each of those services (livejournal there) would then publish their
part of my social network as further xfn-links (friend, contact)
pointing to the openid-urls of my friends and such. Very simple, and
robust. Much overlap, but everyone would still retain full control
over their own network.
The publishing part here is of course already being implemented
(apparently that's what the original thread was about, 'Six Apart is
Opening the Social Graph'), but there's probably no service yet that
can use this distributed social graph as their primary network
> So now rather than querying a common store on the fly, I have to think
> about synchronising my local store with some common store.
I'd suggest you to try design a system that would "loosely bind" your
decorations over the "common network" (as opposed to a common service)
so that your own system would then reflect changes in the cloud; if a
connection is removed somewhere else in the cloud, so would your
system remove data related to that connection. The less all this would
require effort from the user, the better.
Btw., one potentially important component of this could be a semantic
web search engine, which you could query about the relationships. Such
a service would save the trouble for each service constantly
traversing the semantic networks. http://pingthesemanticweb.com/ was
introduced as such, but currently it seems to be reduced to something
For a related issue, see also the Sam Ruby's post on mapping email-
addresses to openid-urls at http://intertwingly.net/blog/2007/09/28/Email-addresses-your-OpenID-via-DNS.