August 17, 2007

More on social network portability

It’s very exciting to see Brad Fitzpatrick and David Recordon’s thoughts on the Social Graph make it into the public discourse. I’ve been talking off and on with them about the open social web for most of the year, and it’s impressive to see how quickly things are materializing. It also echoes the growing desire that both users and developers have for freeing their data from walled gardens and putting users back in control of who they know. This is something we’re actively working on at Plaxo, and there’s still a lot more to do!
First of all, we agree emphatically with several of the points Brad and David make:

  • Users are frustrated that they have to re-build their friends list on every new site they go to, and this also impedes the development of new socially-enabled applications.
  • The answer is NOT to have one company own the social graph and require all apps to be re-built on top of a proprietary platform in order to gain access to it.
  • We have to help users regain control now, and we can’t assume that all social sites will be cooperative (at least right away), nor can we wait for everyone to agree on a single interop spec.
  • Users care about enhanced functionality and convenience being delivered, not about any particular standards or data formats (per se).
  • Users won’t generally want all the same friends on all the sites they use, but they do want to know when anyone they know on a given site is also using other sites they also use. In other words, the goal is to aggregate who you know across all the sites you use and then let you choose who to connect to where in what capacity.

Looking at things through the lens of Plaxo, which helps 15+ million people around the world keep track of who they know by syncing their existing address books across the many tools they use, there are also a few additional points we think are important to make:

  • While having an open-source, non-profit entity collecting and serving the entire social graph may be better than a single proprietary company or a mess of disconnected companies, we think the ultimate solution has to be that each user owns and controls their own profile and list of friends. Different people will trust different companies to act on their behalf as stewards of their online identity and relationships, but no single entity should ever have to be the gatekeeper for the entire world.
  • Our users tell us that the contents of their address book are private and that preserving their privacy is very important. So while some users are happy to declare their list of friends in an open and public way, we feel that dealing with private data is essential, and certainly much more than “10% of the problem”. I think there are ways to separate the details of authentication from the exchange of info, but it’s worth noting that wanting your data to be portable doesn’t mean you want (or need) it to be public—it is sufficient that you can move your data between two trusted parties without those parties needing to agree explicitly a priori to inter-operate.
  • Linking and crawling the graph of URLs that describe your profile on different sites to aggregate your profile and friends lists is a great idea (and one we intend to support in Plaxo Pulse). But for many users, e-mail addresses in address books is still the de facto standard for representing their relationships, and we believe supporting linking by e-mail address in the open social web will still be important for the foreseeable future. These are complementary techniques of course, and they can be partially unified by using mbox_sha1sum and similar tricks, but the basic process that most social networks use today of “give me your e-mail address book, and I’ll tell you who you know on this site based on their e-mail address” should not be overlooked. Sites will differ on how much access they give you to other users’ e-mail addresses, but I think there are ways to make it work that may actually be simpler than assuming that profile URLs are the main identifiers to deal with.

These are exciting times and I’m thrilled to see so many people getting involved in these important discussions. You can count on Plaxo to stay involved, both in the discussion and as a service that will support emerging open standards and best practices for keeping users in control. It’s hard to say yet how this will all turn out, but it’s great to see the momentum building for the opening up of the social web.

Email Print

About the Author

View Plaxo Profile >>
Redgee Capili

General Manager,

Related Articles
  • James Tauber

    Great post! Especially the notion of separating aggregation from actual hosting. See for links to some of my thoughts in this area.

  • Johan Myrberger

    Just noticed two new features from you related to this area. I blogged about it here:
    However I have some concerns around the feature, care to comment on that in my blog post?

  • Johan Myrberger

    Joseph, thanks for the comments on my blog. I added some from my side as a new comment(
    (And if you would like to move follow up of the conversation to the Plaxo blog it is fine with me)

  • Nels

    The python script you use for the Open Social Grapher apparently only looks for rel=”me” links that have the quotes in them. The profiles on * don’t have the quotes, so instead of saying “is” for my Plaxo profile (starting from my blog), the scripts only “claims” it.
    I would think that would be a simple fix, but I’m also a web developer and I hate it when people say stuff should be easy.

  • Joseph Smarr

    @Nels-you’re absolutely right! I forgot our web-page optimizing code strips quotes for attribute values with no spaces (it worked fine on my test box, heh).
    I’ve patched the crawler to correctly handle bare rel=me attributes, and your site seems to work fine now. While I was at it, I added some stronger URL canonicalization (e.g. lowercase the protocol and domain), and I also now look for link-rel=me tags in the head of pages.
    Thanks for the feedback! js