Social Web Incubator Group Teleconference

07 Oct 2009

See also: IRC log


Adam AlexPassant CaptSolo Carine DKA FabGandon MacTed OpenLink_Software alex bblfish caribou cperey danbri evanpro hhalpin joined melvster mischat pchampin petef rreck tinkster trackbot yuk




<trackbot> Date: 07 October 2009

<hhalpin> Convene SWXG WG meeting of 2009-10-07T15:00-17:00Z

<hhalpin> scribe: peterf

<hhalpin> PROPOSED: to approve SWXG WG Weekly -- 30th September 2009 as a true record

<hhalpin> RESOLVED: approved SWXG WG Weekly -- 30th September 2009 as a true record

<tinkster> evanpro, when you call in, we're on extension 26632 instead of 7994.

<hhalpin> PROPOSED: to meet again Wed. October 14th

<hhalpin> 2. General Organization

<hhalpin> TPAC, IIW

<hhalpin> http://www.w3.org/2005/Incubator/socialweb/wiki/Planning_for_TPAC_Open_Day

<hhalpin> TPAC?

<hhalpin> I imagine go a few...

<hhalpin> Hoping to have an "Open Day"

<hhalpin> but it conflicts with OAuth/OpenID Summit, which is also on Monday 2nd.

hhalpin: you are scribing?

<hhalpin> This is at Yahoo.

<bblfish> do you have a URL for the openid oauth summit?

I'm happy if you are.

<hhalpin> http://openid.eventbrite.com/

<hhalpin> Petef - please scribe!


<tinkster> petef, hhalpin likes to type up everything he says to prevent himself from being misquoted.

<hhalpin> I try to type dates and stuff...

dka my initial view was we should have the open day anyway

<evanpro> Seems like there'd be great synergy to have them complementary events

[please tell me if I am missing scribing protocol]

dka but now think maybe we should cancel the openday

and work with the openid thing

thanks guys

dka: should we just forget about it?

hhalpin: how about Tue at the Marriot?

<bblfish> Well I can get Sun on Tuesday too

<cperey> what about later in the week?

hhalpin: not too many people registered

dka: we could invite people to observe our working group
... unless we could get externally parties in to present, like on the calls

hhalpin: we could invite people Michelle Adamic (sp?) has a list of names we could invite

<cperey> many apologies but what about having Open Day on another day?

<cperey> later in week?

<evanpro> two days close together please

<evanpro> I would attend if both events were close together

<evanpro> A week apart makes things hard for out-of-towners

<cperey> Henry


<bblfish> So the question is how many people would be able to come on Monday?

henry: it's only an openid meeting, we aren't expecting any new announcements

<bblfish> there are three confilcting events

<hhalpin> My concern was two-fold: first, conflicting events might be misinterpreted by OpenID/OAuth folks

<bblfish> there is Apache conf

<bblfish> there is IIW

<cperey> New proposal: what about having an Open Day when there are not conflicting events?

<bblfish> there is W3C conf

<hhalpin> Any ideas?

<cperey> early Dec?

<cperey> I mean completely separate

cperey: I could take an action item to propose other dates at anoher location, not TPAC

<bblfish> ok, then the problem is that a lot of people won't be out there

<bblfish> anymore

dka: I won't be out in the Bay area again for a while
... part of the idea of doing it at TPAC was to leverage people's attendance
... could we do a hybrid event with the openid folk?

<bblfish> sounds good too

dka: half a day on each
... is that a middle way that would satisfy everybody?
... if so we could start drafting a page around that and get it out to people

<hhalpin> My question is can we just get these people to get to TPAC - i.e. TPAC into an "open day"

<hhalpin> Assuming very few people are coming

<bblfish> I think I agree with that

<bblfish> how many people can come to OpenDay?

dka: concerned that we may not be able to network a largish event at short notice

<bblfish> the IIW will be all week?

hhalpin?: won't people also be attending IIW?

<bblfish> sorry I mean they will be all week

dka: we want to make sure we aren't stepping on any toes
... seen as trying to put on a competing event
... but if we could get some CTOs and mktg consultants into TPAc that would be good

hhalpin: we need to move on

<cperey> yep, meeting after

<cperey> henry you are breaking up

<evanpro> petef, 514-554-3826 == me

<cperey> haha!

<tinkster> danbri: we're not in 2994.

dka: evan is on the line, let's make the most of his time

evanpro: I'm enjoying listening

<tinkster> we're in 26632

<Adam> 26632 is the new code dan

dka: we'll go straight to... the invited guests

<hhalpin> 3. Invited Guest - Evan Prodromou from OpenMicroBlogging and Alex Passant on Semantic Microblogging

<hhalpin> Evan?

<evanpro> I'm here

<hhalpin> Evan Prodromou of OpenMicroblogging will give a talk on his work on OpenMicroblogging (see the specification) and its connection to the wider Social Web ecoystem

dka: hopefully evanpro can give us an interesting introduction to OMB

evanpro: a conjunction of openid, oauth and swxg would be enough to get me out to CA
... I am the inventor and shepherd of OMB
... bg on myself - working on open source and content for 15 yrs

esp wikitravel

around issues of identity

<danbri> http://evan.prodromou.name/

briefly on JSF Council

<danbri> http://identi.ca/evan

with the launch of Twitter in 2007

there was an expolsion of interest in OMB

defined as exchange of small text message with your social network

twitter, jaiku, pownce etc.

contrast to blogging , tumblelogs and social bookmarking

people seem to enjoy OMB and the responsiveness of it

as the growth of witter got ahead of their ability to support it

interest grew in the evolution of an open standard for microblogging (OMB)

seeing as noone else was doing it I thought I would set up an easy hack


started laconica in May 2008, hosting identi.ca

includes features important to the future of the open social web

openid, foaf, import and export of data

you can take the identi.ca source and install it in silo

so decided to create a protocol

my design criteria were

http based system

tend to go for the low end on servers

so that you could install a node on commodity shared hosting

wanted it to work with openid and oauth

and to be realtively easy to implement with existing tools and libraries

flirted with openid attrib exchange

<danbri> cf http://openid.net/specs/openid-attribute-exchange-1_0.html

but the mechanics don't allow pushing and pulling to other openid servers

so wouldn't really work

instead used oauth

subscribing to somebody's notices means that you authorise another server to push to your inbox

four players:

publisher, subscriber, publiher-server, subscriber-server

profiles within OMB are http urls

subscriber provides their url to the publisher server

which uses oauth discovery

still very fluid

OMB uses XRD simple files

draft since deprecated for LRDD

will share links later

use oauth discovery to find endpoints

two additional endpoints are defined

the user experience is to go through the typically oauth dance

token exchange

unlike some other pubsub mechanisms

there's an expectation that I should know who subscribes to my feed

closer relationship

pass along profile info in oauth dance

avatar, full name, location, homepage, short bio

via http post params

once sub complete

the two additional endpoints are used

push based (unlike atom or rss)

no polling

good responsiveness, better performance

post notice endpoint

used by pulisher server

profile update endpoint

used if publisher profile chganges

any kind of client to server api is out of scope

laconica (now called status net) uses a clone of the twitter api

also working on an atompub version

any kind of syntax def'n is also out of scope

<Adam> the pub/sub explanation sounds very similar to pubsubhubhub

just pushes plain text

hashtags #foo and directed messages @bar are not defined in OMB

0.1 protocol has been fairly successful

thousands of nodes on the network

important weaknesses

no system for unsolicited notices

if you aren't subscribed to me no way for me to message you

<tinkster1> Adam, it sort of is, but there's no hub per se. Updates are pushed from publisher server to subscriber server.

parsing syntax left out which means the hashtags and replies have to be guessed at

threading questions across multiple servers

no threading bult in

no direct private messages eiother

low level thing, in oauth two kinds of keys - one to identify a particular conversation and one for th sending server -

OMB actually uses the root URL of the sending server

hard to verify in the client

I'd like to talk about where we are going with OMB 0.2

four other implementations:

openmicroblogger, jaiku (at 0.2), etc.

for 0.2 biggest change is move away from plain text to atom entries for pushing notices

richer data format

<DKA> NB: In Vodafone we are already using laconi.ca for an internal microblog system and we are looking at using it as a platform for building mobile social network prototype apps.

we'll also e able to leverage work around activitystreams

which is also atom based

<danbri> http://activitystrea.ms/

<danbri> (aside: they're moving this to be equally atom and json)

second important thing is we will include parsed html

some security issues obviously

will require sanitizing

but good step for usability

adding in-reply-to info

will improve threading

@xyz will resolve to an identity url for the recipient

that will travel with the packet

replacing oauth discovery with LRDD

support for unsolicited messages

in protocol, up to clients what they do with it

proposal to provision verifiable user keys

[tinkster I'll get it right eventually]

we want something interoperable by the end of the year

relation to openid

OMB identity is an URL


want to make sure we mesh nicely with openid

two new feed based systems announced

<danbri> bradfitz

<danbri> fitzpatrick


atom based

<danbri> http://en.wikipedia.org/wiki/PubSubHubbub

and rsscloud

embedded into rss 2.0 but never really implemented

now being revived

neither supports feeding back subscriber relation to publisher

we are implementing both of these and trying to work with the standards to add in subscriber publisher feedback

activity streams - the hard part will be the unification between AS, atom and opensocial

always wanted http based system

but would also like an upshift to xmpp based if both servers support it

well known protocol, very nice pubsub system

some work done to adapt it for OMB

looking into that

stubs for that in 0.2

discovery process over the next few months


<Zakim> danbri, you wanted to ask if anyone has done usability testing on OMB remote subscription scenarios

<danbri> peterf first

petef: have you looked at wave-protocol in relation to OM


evanpro: v. intersting but ow it will fit into OMB is still open

<danbri> ok i fixed my audio, i'm next on queue please :)

for low end target systems wave won't be available for quite a while

whether it picks up as ubiquitous will be interesting

my hope is that it does see some success

don't think the current generation fits well with OMB

<tinkster1> yes

interested in wave and how it develops

<tinkster1> no longer audible

<rreck> you were audiable

<DKA> losing part of danbri's speech

<danbri> q:

<danbri> have you done any usability testing

<danbri> esp on remote subscribe aspects of OMB

<danbri> do users understand what's going on, etc

evanpro: usability testing - pretty important usability problems

v. similar to usability problems for openid

mechanism is very similar

I need to remember my identity url

<danbri> cf http://lists.openid.net/pipermail/openid-user-experience/

and provide it to remote servers which use it to start a conversation on the back end

we haven't done much work on that

hopwe we can borrow from openid world

<danbri> i'm interested to see more work figuring out how much implies protocol design changes, and how much can be done with improved web design, addons etc

widgets and keeping an eye on webfinger

<danbri> http://code.google.com/p/webfinger/

<danbri> http://hueniverse.com/2009/09/implementing-webfinger/

users associate people with email type addresses

and urls with docs

<danbri> thanks evanpro, ... i think webfinger pretty relevant here

<danbri> http://hueniverse.com/webfinger/

webfinger also nice simple way to include addresses into OMB msgs

<tinkster1> http://buzzword.org.uk/2009/fingerpoint/

passive subscription

<danbri> """What Problem is WebFinger Trying to Solve? ... WebFinger is trying to address the lack of adoption and usability of current HTTP-URI-based user identifier solutions, most notably OpenID. """

would speed things up and improve usability

could also happen with a global directory, although there are concerns about recentralisation

<danbri> (so maybe some role for http://code.google.com/apis/socialgraph/ etc ?)

<Zakim> AlexPassant, you wanted to discuss synergies with SMOB / SIOC and other vocabs

<danbri> skypish sounds

alex(?): very brief overview of SMOB (audio not clear)

wondering how it relates to OMB

<tinkster1> Alex, StatusNet already published FOAF, RSS and SIOC data.

<AlexPassant> SMOB: http://smob.sioc-project.org/

evanpro: we support sioc(?)

<tinkster1> s/shark/SIOC/

<danbri> SIOC is pronounced like Shark, yes

<danbri> (irish word)

doesn't address server to server mechanism

<CaptSolo> danbri: commonly pronounced as "shock"

similar to smtp

SMOB does a good job of keeping your published info and pushing it out

but not really server to server interaction

<danbri> (shark is commonly pronounced 'shock' in some countries too ;)

wouldn't be a big leap to implement OMB in SMOB

would be very welcome

<CaptSolo> danbri: the proper irish pronounciation is similar to "shuck" and means frost (imagine social networks connecting together like snowflakes or frost on a window)

OMB doesn't (yet) cover sharing the social graph

you can use other mechanisms, through FOAF say,

<tinkster1> http://chatlogs.planetrdf.com/swig/2009-10-07.html#T11-56-51

might be nice to promote it or provide an url

as part of profile info passed through

SMOB is cool, would love to see it brought together with OMB

alex: more than happy to work together

evanpro: one of our devs took the OMB stuff out of staus net and turned it into a WP plugin and a PHP library

<tinkster1> you're breaking up

<danbri> did we have DKA's question yet?

hhalpin: how are you using FOAF?

<DKA> nope

<DKA> :)

<bblfish> w: are you making use of foaf to do something

<tinkster1> http://identi.ca/tobyink/foaf

evanpro: we provided FOAF info and use FOAF ids in graph

toby has been great in helping us implement that and the SIOC stuff better

<danbri> try putting http://identi.ca/danbri into http://socialgraph-resources.googlecode.com/svn/trunk/samples/exploreapi.html

<danbri> google sgapi indexes identi.ca foaf

Could use FOAF for profile exchange in OMB

makes more sense for discovery and so usability

who should I follow?

who might I kalready know?

<danbri> poss foaf use cases 'what kind of people should i be subscribing to, who sohuld i follow, who is there that i know, ...'

<danbri> (i think also around groups, ...)

groups is another great case

there's going to be a close interaction between OMB and FOAF

<tinkster1> SELECT ?person WHERE { ?me subscribesTo ?x , ?y . ?x subscribesTo ?person . ?y subscribesTo ?person . FILTER (?x != ?y) }

<danbri> evanpro, if there are tweaks you need to foaf to support any of these use cases, ... that's not so hard to arrange!

dka: we are limited on time

<danbri> i've already had a question so i'll take it to email!

<danbri> a-

dka: are you thinking about making lcation part of the platform

<danbri> (my q was basically http://lists.status.net/pipermail/statusnet-dev/2009-August/002117.html again)

one reason many of us use Jaiku is the sophisticated support for location

evanpro: fascinating that we are reintroducing location to the web

in OMB 0.1 we only have a points tring associated with a profile

if I update it, that can change (but would be lost)

making changes to staus net to support a richer more machine friendly location format

will be out in 0.9 later this month

<tinkster1> geonames IIRC?

lat long and geo place name vocab

where on earth, geonames, OSM

we will also include location info for each notice

<AlexPassant> sounds along the lines of the linkage we're doing in SMOB with geonames and DBpedia, great synergies

would make sense to get it into OMB 0.2

optional but important

moving to atom - should be easily included in atom entries

<danbri> (http://status.net/trac/report/6 seems interesting to poke around...)

evanpro: would love to attend open day, thanks for having me on the call

<hhalpin> evanpro - will stay in touch via e-mail, and thanks so much for being on call!

dka: thank you

<danbri> thanks evanpro!

dka: we have to close the call now
... we need tocome up with a plan for the openday offline by tomorrow
... how to mesh with SWXG workshop, openid and oauth

<hhalpin> Oh, can we have an ACTION by someone

<hhalpin> to summarize that call?

<pchampin> bye

<AlexPassant> thanks, bye

<tinkster> I can summarise

<DKA> good point harry -

<hhalpin> thanks tinkster!

<rreck> way to go tinkster

<DKA> :)

<hhalpin> ACTION: tinkster to summarize Evan's talk [recorded in http://www.w3.org/2009/10/07-swxg-minutes.html#action02]

<trackbot> Created ACTION-96 - Summarize Evan's talk [on Toby Inkster - due 2009-10-14].

<DKA> :)

<hhalpin> Dan - do you really want to do it, or tinkster?

<DKA> If tinkster can do it that would be great.

<hhalpin> OK, will delete that from the minutes then...

<tinkster> I'm happy to.

<DKA> thanks toby!

Summary of Action Items

[NEW] ACTION: tinkster to summarize Evan's talk [recorded in http://www.w3.org/2009/10/07-swxg-minutes.html#action02]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2009/10/08 15:51:23 $