Sunday, May 28, 2006

Social Networks and Memorial Day

If you have a the day off work and you'd like to catch up on some reading you should check out these materials very generously posted by the University of Melbourne. here

Tuesday, May 16, 2006

NSA and Social Networks

I swore to myself I wouldn't say anything about this...

There's been a lot of flap about NSA and the phone companies recently (USA Today) and a couple of counterarguments have emerged. These are mainly along the lines that either the whole technique is just plain goofy and produces unreliable results or there are clearly much better approaches to uncovering clandestine networks (based on message contents). See Jonathan David Farley in NY Times, and Jeff Jonas here

Still, the initial revelations that some sort of program even existed got so many people upset that you'd have to wonder if there's a non-obvious nugget we are missing. The small number of folks I've met from NSA impressed me as being high on the sharpest-knife-in-the-drawer scale and not given irrational analysis.

While links between phone numbers alone may not be of much use there are other pieces of information within phone records that could be useful. As a minimum, phone logs have directionality built into them, there's also the duration of the call and the timeframe of related calls. For example, if a call from A to B habitually causes B to immediately make a series of short calls to C/C'/C" that might be a pretty interesting pattern to be aware of. (I'm suggesting that B is acting as a "cutout"). After looking at my own phone bill for a few hours I cannot see anything of significance in this context - folks call me about as often as I call them and the calls are of widely different durations.

All I'm getting at is that with sufficient processing power it should be possible to tease quite useful intelligence out of a seemingly very large and random haystack.

Added note - The New Yorker 05.22 here

05.28 Freedom of Information hah! here

06.02 Federal judge allows lawsuit against NSA here

06.06 Can Data Mining catch Terrorists? here

06.09 Pentagon sets its sites on social networking websites here

Friday, May 12, 2006

The Value of Enterprise Social Networks

How do we calculate the value of a social network? How will we know if we are increasing (decreasing!) the value of a network? Answers to these questions are important because they ultimately determine if companies will be willing to make investments in social network tools and applications. Here’s a condensed version of where we are today – references below – fasten your seatbelts and trays in the upright position...!

There are four types of social network: loosely connected, tightly internally connected but without strong external ties, strong external connections but with loose internal ties, and finally a balanced network with strong internal and external ties. It is easy to think of these in terms of archetypes – the real estate office comprised of independent agents, an Amish or Mennonite cooperative, a screen writer’s guild, and a contract manufacturer respectively. Numerous studies conclude that groups that have a well developed internal structure as well as ties to other external communities will, on average, be more competitive, successful, creative, and have higher incomes, and adapt better to unexpected challenges.

The primary purpose of an enterprise social network is to save its participants time. The advantages of networks are (at least) twofold: they enable internal standards of compliance and trust to be quickly established for members of the group, and to rapidly connect with external sources of critical information whenever needed. Indeed, “There is an impressive diversity of empirical evidence showing that social capital is more a function of brokerage across structural holes (connections to other networks) than closure (cooperation) within a network.”

One example of this that very much sticks in my mind was the development of the astronauts’ gloves for the Apollo program. On one hand you had a team of engineers that knew a lot about the specifications that would need to be met but without much real experience making gloves. On the other hand you had a team of women who had worked all their lives sewing baseball and ice-hockey mitts. Clearly a great deal of new value was created in the ties that linked these two very disparate networks.

Therefore, in terms of a business value calculation there are three parameters that need to be estimated: the inherent value of information resources within an organization (and how optimally this information is made available for use), the relative number of connections the organization has to external agents (customers, suppliers, professional organizations, investors, even competitors), and the nominal value of project time (how much would it be worth to shave a month from a new product introduction).

I have worked up some interactive models to capture and validate these parameters – contact me for more information.

The Network Structure of Social Capital, © Ronald S. Burt, May 2000
So Many Ties, So Little Time, © Hansen, Podolny, Pfeffer
Cool Vendors in Social Network Analysis, ref Gartner
Email as Spectroscopy, Tyler, Wilkinson, Huberman, HP Labs