FriendsListPortability

Presented by jsmarr


 * oauth, microformats, openid
 * privacy issues... portioning friends connections


 * the problem: too hard to stay on top of what the people i know are doing online
 * siloed data, social apps don't know who i know automatically, little control over who i can share data with, have to recreate friends on every site with social features
 * "massively underdeployed social graphs"
 * if friction is too high for super geeks, imagine the friction for regular folks! there's no flow...


 * we lack a technical and user control model to make this happen


 * the fallout: missing content from friends... (what about the overload?)... social apps desperate to get even PART of your social graph (re-implementing gmail scrapers, building facebook-only apps)... sometimes i'm too easy to find... especially by email address! (yikes!).... sometimes i'm too hard to find on sites (can't look up friends by other attributes besides email)


 * the vision: "facebook-like" platform for the Open Web.
 * friends list == people you know from any of the sites you use
 * user ids == email/urls or other identifiers by which you know people
 * social apps == running anywhere with the same richness (apps running in iDPs?)
 * activity streams and profile badges exist in other social networks
 * managing relationships across sites... fine grained control?
 * try new services, meet someone new, social app developers outsource "who you know"
 * allows for niche site development... hard to sustain without portable social graph


 * the building blocks
 * who am i? openid, rel-me
 * identifiers: mailto link, url, account url, IM link, xri
 * role: managing my set of identifiers (which identifiers to reveal to whom?; which identifiers can i be found at per app?)
 * who are my friends? XFN, FOAF
 * friends list == set of identifiers that I know; need portable list of these identifiers, often via authorized requests
 * who do I know? gating with OAuth, 6A RUS
 * how can i use my data? opensocial, foaf, xfn, vcard
 * proposal: social sites should provide a persistent url to your friends list; url can contain OAuth token for private data... lists all identifiers you know... allows export via raw identifiers or hashed identifiers (what about writing to the list)
 * allow sites to match known identifiers against the "people i know"
 * find all people i already know on new site
 * add new site as a source of friends list data (be able to subscribe to friends list so when new people sign up, you'll be notified)


 * clarity on roles and responsibilities
 * user: manage your identifiers
 * social apps: provide access to shareable friends list, let users control how they're findable, provide lookup for findable identifiers
 * shouldn't leak any new information in this approach (i.e. foaf)
 * built on existing tech: openid, oauth, xfn, microid, URIs
 * bridges lookup by email address vs URL --> transitioning


 * room for everyone to win
 * social networks become more relevant as they extend their reach
 * social apps are easier to build -- outsourcing the social graph (better friends list == more compelling app)
 * users can find each other better, more efficiently share content


 * next steps
 * clarify/propose basic steps for interop
 * get early adopters to implement it
 * watch for early results (usage, privacy)

Demo of Online Identity Consolidator


 * what's the business model? what's the economic incentive?