Thursday, April 27, 2006

Social Networks - Books You Must Read

Pretty frequently I get asked (usually by my wife) - "where do you come up with this stuff...can't you watch Deal-No Deal like a normal person?"

While there's a lot of great material on Social Networks and Social Network Analysis available there are a few books that seem to be referenced by others over and over again. Three that I have found pretty insightful are shown: Social Network Analysis is a great nuts-and-bolts introduction to SNA graphs; how they are constructed and what they mean. The Wisdom of Crowds is an excellent piece of writing (I wish I could write like that!) that paints a picture of how people behave in groups. Finally there's Rob Cross's The Hidden Power of Social Networks which has recently done more than any other to bring SNA into public conciousness.

For desert there's Nexus - another well written page-turner with a great introduction to the "small world" effect.

I'll appreciate any and all feedback on this selection.


Tuesday, April 25, 2006

Social Networks Analysis in the Enterprise

The weekend - so some longish blather... if you'd like to hear more screaming than the night the orphanage burned down start with Nick Carr's most recent post

To be honest I'm not sure what all the screaming is about - although I appreciate Carr's tough questions! A body of information (e.g Wikipedia) is not homogenous - a great proportion of the contributions will be commonplace. Similarly, so called numbskulls (ref Carr's choice of language) can contribute much of the mundane - and even some of the arcane. At the edges, without question, there is a need for highly qualified experts to define the most complex subjects. Basically the world looks like this (diag.)-

At last count Wikipedia had 50,000 contributors responsible for 2.9 million entries, 890,000 in English - however, only 2,081 had contributed 100+ articles. This is being interpreted as maybe bad news if the Wikipedia model is what is to be expected in corporations. But, guess what, this is not much different than you might expect. Let's say the average "habitual" contributor is responsible for about 100 articles - a total of 208,100 articles. This is approximately 23% of the 890,000 total - inline with the above guess of about 25%. A small number of contributors are doing a lot of work but apparently the system overall depends on numbskulls for 70%+ of contributions.

It goes without much saying that we all look forward to a time when online communities will be more like our networks in the real world. This vision may never be fully realized but it seems clear that the next stage, the next proof point, in the development of social networks will be in the context of corporations -so called Web 2.0 for the enterprise. This begs some tough questions: can Social Networks really produce tangible improvements in resource conservation, productivity, or competitive advantage? How will this work and, for a generation imbued with MySpace, Yelp, Digg,, Wikipedia etc., what will it look like?

Predictions -

1. Starting Now. We are currently navigating the “Trough of Disillusionment” of the hype cycle for social networks, corporate wikis and blogs, and we are beginning to see real corporate traction – the next 12 months will be both interesting and exciting. We all still have a lot of work to do but there is light at the end of the tunnel! Corporations are already intrigued by the possibilities of Social Networks and related tools in the context of disruptive events – the introduction of new products, for example, or bringing new facilities online, mergers and acquisitions, and disaster preparedness. Using Social Networks for mentoring and expertise sharing for complex products is available only in a few places although this seems to an extraordinarily valuable application area.

2. Solutions not Components. Companies need solutions that integrate with their existing business processes and that carefully take into consideration any new risks that they may create. New ways of capturing content will be valuable only in so far as they tangibly enable downstream processes. Almost certainly the Legal department will want to be involved – if blogs are to be shared with external readers there must be publication approval processes and clear rules of engagement; wikis must be devoid of personal references to other employees etc. In short, by building features that offer security and conform to well established business standards it becomes more probably that social networking tools will be accepted.

3. A little top-down will go a long way. For god’s sake never forego an opportunity to convince someone in the executive suite to participate in a wiki or start a blog! It’s true that a great many executives are unimaginative, clueless, or risk averse but this just means you have to step up your game. Today only about 4% of the Fortune 500 support a corporate blog but some of them are quite good (Ford Motors) and the field is growing. Social networks are inherently organic but, from my own direct experience, there is still a lot to be said for visionary leadership. My guess is that 10-15% of F500 with corporate blogs will be a critical turning point.

4. Education. Having great ideas that nobody knows about is not very useful. There is a huge open hole where educational books and papers on social networks in business should be. When we begin to see really useful text in this are – for the love of all that’s holy – written by non-academics, it will be time to fasten your spacesuit!

5. Show me the money. Business people are actually fairly easy to understand. Every strategic decision ultimately boils down to: will deploying this social networking tool cost me more (or less) than I can expect to get as a return; how risky is deploying this tool to my personal reputation and that of my company relative to the nominal reward? Social networking tools for the enterprise should focus on building communities within companies, making them more productive, agile, and competitive. When we can demonstrate how to get from here to there the warmth of good fortune will fill all our futures… more on this in a few.

Friday, April 21, 2006

Constraints on Group Size in Social Networks

How many people do you know? OK, not Brad and Angelina - but really KNOW - interact with directly and be influenced by? How many people can you know? After all we are all constrained by time - so there would seem to be limits to how many people we can interact with socially on a productive and regular basis.
Imagine living in a small village or being part of an autonomous tribe where each member has a role to play in achieving group objectives (the butcher, the baker, the candlestick maker; tinker, tailor, soldier, sailor). In this hypothetical village there are five sets of grandparents (10 people), each has given rise to three married adults, and these in turn have three children of their own. This modest group is 85 individuals. This five-family village seems intuitively about as small as a community can get and still be viable (I am not offering any evidence for this just yet - I hope you'll agree that it is at least plausible). If we expand the model to a core group of nine grandparents, with the same average number of married offspring and children, it will lead to a village with about 153 members, and so on as shown.

The relationship between the number of families in the community (with kinship to the core grandparents) and the amount of time available for social interaction on a daily basis is as follows.

What matters here is the overall shape of this curve - as the community size increases the time available for building bonds of cohesion, and social grooming between community members, decreases dramatically. Up to five families there is so much time available that it may seem like the community is truly just "one big family". But, at some point (around 9 families) there is a subtle transition where individuals will have to consciously/unconsciously decide to socially groom and bond-with members of a sub-group (their kin) rather than the community at large.

Surprisingly (to me it was VERY surprising) this line of thinking is supported by a wide selection of examples from the real world. A lot of this has been captured in a beautiful paper by Robin Dunbar from University College London (google him).

Reference - Robin Dunbar "Neocortex size as a constraint on group size in primates"

One quote from this paper -
"...the reason given by the Hutterites for limiting their communities to 150 is particularly illuminating. They explicitly state that when the number of individuals is much larger than this, it becomes difficult to control their behavior by peer pressure alone. Rather than create a police force they prefer to split the community. Forge (1972) came to a similar conclusion on the basis of an analysis of settlement size and structure among contemporary New Guinea "neolithic" cultivators. He argued that the figure of 150 was a key threshold in community size in these societies. When communities exceed this size basic relationships of kinship and affinity were insufficient to maintainsocial cohesion; stability could then be maintained only if formal structures developed which defined specific roles within the group. In other words, large communities are invariably hierarchically structured in some way, whereas small communities are not."

Pretty clearly this presents opportunities for modern Social Networking tools to intelligently identify, strengthen and support working teams and communities of practice.

Saturday, April 15, 2006

Measuring Social Networks

How to calculate the number of connections in a Social Network.

A social network is presented graphically as nodes (points) and connecting lines. The nodes represent people and the lines represent the existence (or absence) of a relationship between them. Let's assume for a moment that the relationships are of equal value in both directions - that is, node A interacts with node B at the same level of intensity as B interacts with A (obviously this will not always be the case, but let's accept this simplification for now). In this case the total number of possible connections in a fully saturated network is given by the formula shown.

For example: with five nodes there are 5x4 = 20 possible interactions, therefore 20/2 = 10 connections.

Rather obviously, and in agreement with common experience, the possible number of connections rises much more rapidly than the rate of increase in the number of nodes. As the community you live in gets larger the more difficult it becomes to objectively evaluate the relationships between othere members. The graph gives a sense of the scale of the problem - for a small company of 100 people there are almost 5,000 possible social connections. It seems clearly impossible for an individual employee to have any objective sense of how information is REALLY flowing through this social network (without some appropriate analysis tools).

How to calculate Network Density.

An important metric for social networks in the real world is their density - how well connected are the nodes in a real network relative to the theoretical number of connections possible. This measurement is intended to give a sense of how well communication pathways in the network are capable of getting information out to the network's participants. The calculation is straightforward - known connections divided by maximum possible connections. (an ideal, fully connected network would have a density of 1.00)

For example: the network in the graphic has 13 nodes and 17 known social connections. But with 13 nodes there are 78 possible connections. This means the density of the network is 17/78 = 0.22 (in a real-world network this calculation is usually completed after discounting any unconnected nodes).

Common experience confirms that the density of networks goes down as the size of the group increases. Research suggests that above a few hundred nodes the maximum density will never be greater than 0.5. Network density is intended only as a rough guide to connectedness it is not a hard-and-fast indicator of group performance - it needs to be interpreted intelligently.

However, these initially simple ideas begin to open doorways to interesting questions - what inferences can be drawn from the structural shape of social networks and what would be an optimal size for a functional team/community?

My next post will discuss metrics associated with optimal community sizes.