Google accessing other parts of the social graph

By December 19, 2007 One Comment

I’m getting increasingly interested in the notion of social graphs and how the concept extends beyond our friends list on Facebook.  The basic idea is that the list of who you communicate with and how frequently is an information set that can be mined to create valuable services.  Put that way your list of Facebook friends is actually pretty simple.  Potentially richer data sets include who you email and who you telephone.

On the email side check out Fred Wilson’s post on Outlook plugin Xobni and a month or two back he wrote about how he used an Outlook plugin which showed who he emailed most frequently to generate an invite list for a party (which I can’t find now).

On the telephone side, here in Europe, Zyb allows you to use your mobile phone call records to social network.

Then yesterday I used social graph based feedreader Blogfriends to navigate to a post about Google Profiles – a new service from Google that uses social graph data from Google services like Gmail, Google IM and Google Talk to try and offer you better services.  For example they have integrated with Google Reader to tell you which posts your friends have read.

This a good idea but the execution is really tricky.

There is valuable information in these communication tools that is simply not getting used at the moment – so anything that tries to rectify that situation is a good idea in my book.

But, once again, the execution is really tricky.  There are a couple of reasons for that – firstly it bumps up against privacy issues so you are in an area where it is really easy to screw up and put people off your service before they have even engaged, and secondly social graph information is really complicated.  There is no straightforward correlation between the volume and medium of my communications with someone and my relationship with them.  If you need any convincing on these points check out the negative reaction to Google’s profiles on Venturebeat – it is full of these sorts of concerns.  E.g. if you have frequent Gmail communication with a competitor this new service would infer that you are friends and tell her what posts you have read – information you might not want her to know.

To me this is all a question of service design and getting the defaults right.  Get it right and you will have a valuable and popular service.
That is easy to say, but difficult to do, particularly when designers will always be tempted to have automatic opt-in with voluntary opt-out as a way to quickly get traction on the service.  In an ideal world the default should be opt-out, but you have got to recognise that it might take forever to build traction that way.

It looks like Google has got the defaults wrong so far with Profiles, and getting the design and defaults wrong is why Facebook is struggling with Beacon.