Social graphs: How Dense is too dense?


Image courtesy : flickr

According to McAfee, in his book : “Enterprise 2.0”, ties for any knowledge worker can be represented in what he calls: “The Enterprise 2.0 Bull’s Eye”  

Consider the prototypical knowledge worker…She has a relatively small group of close collaborators; these are people with whom she has strong professional ties. Beyond this group, there’s also … ‘professional acquaintance.’ In Granovetter’s language, she has weak ties to these people.

Beyond this group there’s a still-larger set of fellow employees who could be valuable to our prototypical knowledge worker if only she knew about them. [these are potential ties]…

[The] fourth ring…labeled “none”…encompasses people who… are not necessarily ever going to form valuable ties…with our focal knowledge worker…however [They] can generate valuable information in the form of prices.

Every Enterprise 2.0 platform is meant to solve issues relatively to a specific ring. Strong ties, for example, can use wikis to ensure better collaboration and agility. While network bridging (what social researchers like to call spanning structural holes), possible thanks to social Networking softwares, focuses on Weak ties.

Let's model an organization using a collaboration graph where vertices represent the knowledge workers while two distinct collaborators will be joined by an edge whenever they collaborate on a certain project. It will then be obvious how E20 platforms help increase the graph’s density and span structural holes mainly by adding new edges. The question is: How dense is too dense? Is there a point where adding edges wouldn't do the organization any good? Can it even backfire on the collaborators' productivity…?

Fred Brooks, in his famous book “the mythical Man-Month”, introduced the concept of over-communication’s cost. Effective communication is in fact about channeling the right information only to those who need it to complete a certain task or achieve a goal. Effective communication would also be about getting non-redundant information from distinct collaborators. So, to decrease communications’ cost to a minimum, we need to compress a social Network to an extent where only useful edges remain.

While this could be feasible in theory, real-world networks have been proved too complex to be predicted. I have been researching this point these last few days and still can’t find a satisfying answer (I'm still searching by the way). Imagine the possibilities of applying such a theory: A structure with Just-enough links. An organization with no structural holes, no redundancy and where every possible information is within reach and is available throughout the shortest path. Wouldn’t that be the ultimate collaboration Eldorado?

Got any research findings to point me out to? Would be glad to hear your thoughts.

Lamia Ben