jasnell, tantek, tilgovi, bblfish, jasnell_, elf-pavlik, the_frey, the_frey_, shepazu_, wilkie_, wilkie and melvster joined the channel
#melvsterhi all, I sent a mail to the list ... regarding the creation of an architecture document for the social web ... anyone in this channel interested in helping? rhiaro, interested?
#melvsterelf-pavlik: an excellent start, but what I envision is a stand alone document for those familiar with the web, that builds on this, and talks about building people centric and relationship driven systems
#melvsteryou might consider it to be awww volume II
#melvsterelf-pavlik: so that's an excellent start, what I would suggest is to think about the importance of the social graph, when I was asking about boeing they have some really interesting use cases around profiles, relationships, discovery and search -- which are all read only operations -- then we can talk about what the ability to write to the social graph brings ... and how to do this using HTTP, URIs etc.
#melvsteri would like to start with saying why social is special and building on that
#rhiaro_hey melvster. Interesting discussion.. you've mentioned prerequisites to social web or user stories a couple of times, I've been meaning to ask, where are these prerequisties derived from?
#bblfishI am pretty happy with SoLiD as a base to move and explore
#melvsterrhiaro_: some of the things we talked about in paris, ie following, friending, two way relationships, does two way following equal a friend? how do co-workers fit in etc. ie relationship driven
#melvsternowhere is the social graph explained to people that may be new to the space etc.
#rhiaro_melvster: I know what you suggest the prereqs are, and a the things you suggest make sense, just wondering if you're getting these from somewhere specific
#rhiaro_or if it's your perpsective on what the prereqs should be
#melvsterrhiaro_: no, it's work that needs to be done, my experience of it was coming to the user stories and trying to implement them, then noticing in several cases that previous assumptions are stated, in a sense it would be good to collate previous assumptions and document so that everyone can start from the same point
#rhiaro_So for example, the section you added to API Reqs on the wiki seems kind of backwards. I mean, the reqs are derived *from* the user stories, not the other way around. You could probably get very similar results from working through the friendship graph user story to derive as requirements what you have called pre requisites
#melvsterat the same time I do feel there's general principles that apply to social in general, partly captured in the XG report and in other places
#bblfishmy fear melvster is that there are three groups that are part of the Social Web WG that come from different points of view, so there is not really much you can do to find a common starting point
#rhiaro_it's just calling them prerequisites without showing evidence they're based on makes them somewhat weaker
#bblfishyes, just saying. It is not easy to find common ground.
#bblfishEasier to say: these are the pre-requistis or vision of SoLiD
#rhiaro_So, extracting assumptions underlying the user stories would be useful
#bblfishthen its easier for people to disagree or to join
#rhiaro_That makes sense, since as you say, if people come with different assumptions they may interpret them in different ways
#rhiaro_And certainly we need to allow for many different experiences in implementing social stuff
#melvsterrhiaro_: agree, there's more work to be done, which I think is common as wiki pages evolve
#melvsterrhiaro_: yes it's really important to learn from experiences, it was very interesting yesterday to learn how boeing is using social for their 180,000 employees, people can find other people based on expertize and self defined interests, ask questions in a group, or globally, and other cool stuff
#melvster2) what do you think of the term "Social Graph"?
#melvsterelf-pavlik: ok let's try that to start with, I must admit I have found IRC to be unproductive, but we can try, I'll focus on the the gh issues
#elf-pavlikdo you feel comfortable yourself with using JSON-LD?
#melvsterelf-pavlik: because it's already a REC. I like JSON LD also, but my understanding was that this group has not yet decided whether to use JSON LD or JSON, so many of us are waiting on that to give concrete recommendations ...
#melvsterelf-pavlik: I would be comfortable using JSON LD, yes, but I think some are opposed to it, so that must be resolved somewhow
#melvsterelf-pavlik: imagine for some reason that activity streams 2 rejects json LD, then what do we do?
#melvsterturtle remains a solid choice at that point
#elf-pavlikdo you really think that if people in this wg don't agree on JSON-LD will agree on Turtle @|@
#elf-pavlik"his specification describes a JSON-based [RFC7159] serialization syntax for the Activity Vocabulary that follows the conventions defined by the [JSON-LD] specification. While serialization forms other than JSON-LD are possible, alternatives are not discussed by this document."
#elf-pavlikwhile I agree that jasnell_ seems to try in AS2.0 something that might not really work
#melvsterelf-pavlik: MIME media type is "application/activity+json".
#melvsterso for example .jsonld files in gh-pages would break etc. while it may have a common syntax, it would not be JSON LD, it would be a competing media type
#melvsterim unsure you could actually call it json ld
#elf-pavlik" .jsonld files in gh-pages would break" i don't find it a major issue ...
#melvsterthe idea is to make a document based on awww but socially oriented, be People oriented and Relationship driven (terms from the IBM slides)
#melvsterjust at a high-ish level explaining the desirable properties
#melvsterthat will give implementors a good basis to create interoperable and scalable solutions
#sandroThat probably makes sense, but it's not something that excites me right now. (Personally, I don't feel we know what we're doing well enough to document it for others yet. I know lots of folks disagree with me on this.)
#melvstersandro: thanks, yes that's true. Hence my suggestion to keep it pretty short and cover basics, such as the social graph and why it's valuable, why we need extensibility etc.
#melvsterhow to do profile, how to do relationships, as an extension of awww
#melvsterbblfish: Hi, I'm Jon - I work on OpenPGP support at Facebook. I thought you might be interested to hear that we now support fetching public keys via the Graph API.
#melvsterKeys can be fetched from someone's profile "public_key" field, e.g. you could fetch my public key with the query:
#melvsterbut then I think that's also true of all its rivals (except webid)
#melvsteri think you can get all confused with definitions like silo, monoculture etc. and end up becoming what you oppose ... a true test of a silo is whether or not it allows connections, and this can be quite easily observed
tilgovi, bblfish, elf-pavlik and the_frey joined the channel
#KevinMarksyou can link outside from facebook, you just can't see the links unless you have an api key
#pdurbinmelvster: sorry, what discussion? did slvrbckt post something about federated identity?
#melvsterKevinMarks: link from one profile to another? I think you're mistaken ...
#KevinMarksfacebook used to support xfn in the logged out profiles. it will show the links when logged out, but they're not curlable
#KevinMarksso you still get a link with rel="nofollow me" on
#melvsterKevinMarks: links outside of facebook, so you could link from facebook.com/kevin to tantek.com/#this ?
#aaronpkif you manage to actually get a curl request through (by setting the user agent to a browser) then you just get a blob of javascript crap and html comments
#melvsterKevinMarks: ah ok, I guess maybe they did it before but not sure they still do, that's good for a federated profile, but it's better if you can crawl it, rel="me" doesnt scale either, where facebook have a restrictive policy is person to person relationships that may exist outside of facebook ... which is a shame, but I understand the business case
#KevinMarkssocial graph api at google crawled it nicely for a while
#KevinMarksfb used to list a random subset fo friends on the logegd out page too
#melvsteraaronpk: curl request to public key works rather well, it even sets the correct mime type 10/10 to facebook on that front, it's more disappointing the way oauth tokens are now needed for stuff that wasnt before
#KevinMarksyou're clearly using "scale" in some odd personal way
#aaronpkwhat part of this makes facebook a "half silo"
#melvsterwhat's interesting to me about public key is that not only is it a pointer to another identity, but it's a token that can be used for authentication
#melvsteraaronpk: you can link *in* to facebook from another place, using web standards, if the *policy* allowed you to link out it would have two way relationships, but the policy is restrictive, so there are only one way relationships
#aaronpkyou mean you can link in to facebook because things have URLs?
#melvsternot quite, special kind of URLs that denote people
#melvsterso i can link in to facebook, but not out
#melvsterin the sense of a relationship driven design
#melvstermaybe we should try and capture these points in some documentation
#KevinMarkswe did standardise on xfn in 2005, except facebook broke it for business reasons
#KevinMarkstwitter to to a lesser extent - they made the follow graph api only
#melvsterI think that's going to be a good way to divide up the social architecture doc. instead of "data model" we can have 1. People 2. Relationships 3. How that creates a Social Graph <-- elf-pavlik ... some food for thought
bblfish joined the channel
#melvsterfor people we can also include things like profiles and accounts, perhaps businesses etc. in general, the concept is a node
#melvsterfor example facebook uses people traditionally, and google uses accounts
#melvsterit's really helpful to have that type in there as you scale, because then processors can branch and do different things with different types of data that they come across
bblfish joined the channel
#melvstergood discussion, ive summarized and sent to the ML
#KevinMarksnot really. it limts you to only have one kind of thing on a page
#KevinMarksembedding an h-card lets you mark up a person