« Best Buy moves into online video with Mydeo | Main | Microsoft Enterprise Search - Freemium for the enterprise »

November 05, 2007


Daniel Feygin

For each friend in the friends list you can expose only an ID with no private information. It would then be up to the friend him/herself to control what private details are divulged depending on who dereferences the ID.

Don Dodge

Daniel, That is the way it should be, but I'm not sure that is the way the OpenSocial API works today. In fact the demos I saw at the Google campfire were importing friends AND activities. I didn't see anything in the demo about EACH friend deciding what to expose.

Could you point me to call in the OpenSocial API that does that?


Unless I'm missing something, this seems like a non-issue...for this use case, I don't see a difference between importing from my list of e-mail contacts and importing from one of my social networking apps. When I import contacts from my e-mail address into Facebook, it looks at the e-mail addresses and finds existing Facebook members. If they're not in there, I can invite them to join. If they don't join, they don't show up as a friend. This is, in fact, the same approach that basically every social networking app takes to this problem.


Tim's quote is dead on, but it conflates 2 issues - "who we are," and "what we care about" - to the detriment of forward progress on both.

We need to de-couple enabling users to take control of their own online preferences from much bigger and hairer problem of online identity management.

See http://www.matchmine.com for at least one example of how.

Dan Lester

So far the OpenSocial initiative really only helps developers.

It is about portability of applications rather than data.

Facebook have previously enabled third-party developers to build applications to be hosted on their site. Now many others are creating similar platforms, all at the same time. OpenSocial just means they're all implementing the same developer standard so that the same code can be used on all of these sites.

It's a bit like writing a desktop application in Java. If Linux, Windows, and Apple machines all have a Java interpreter, then they can all run your application. But they do so completely separately.


I wrote two related posts recently:

1. First was pre the MSFT-FB deal and was about how a Yahoo-FB partnership might look, with in-FB Yahoo search leveraging the so called social graph. My idea was similar to Mark's but I went further to suggest that users could easily convert unresolved search queries to questions for their friends, friends' friends and other subscribed members.

2. The second was about what Open Social should have been in my opinion. Rather than portability of friend lists without permission, one of the things I suggested was the ability to quickly see which of your friends from other networks are on a new website you are joining and connect to them. Or invite those that are not there to join. I also suggested Open ID of course and the idea of a central set of profile data, a subset of which a user can choose to expose to a particular site.

Daniel Feygin

Don, OpenSocial API documentation does not specify privacy constraints of the host container. The data returned for each of the calls could be incomplete or obfuscated in container-specific ways. Consider that the OpenSocial container always knows which app is accessing its data and which user authorized it to do so on his behalf. It is revealing that the examples in http://code.google.com/apis/opensocial/docs/gdata/people/developers_guide_protocol.html reveal no contact information, only user IDs and names, which, for all practical purposes, is the same as just having IDs.

Even if someone explicitly releases all of his data to some app, that app's ability to use the API to retrieve that person's friends will likely be subject to the friends' permissions with respect to the app, which, by default, are likely to be ID only (though complete invisibility is potentially also an option).

Because of these privacy controls I believe social network friends are fundamentally different from address book or RSS feed entries.

Daniel Feygin

Dan, that's how I started thinking about the analogy to Java, but it goes much further: http://dondodge.typepad.com/the_next_big_thing/2007/11/50m-facebook-us.html#comment-88726902

The comments to this entry are closed.

My Photo

Enter your email address:

Delivered by FeedBurner

Twitter Updates

    follow me on Twitter