Network design

by eskokilpi

In a typical large organization everybody is a long way away from everybody else. As a result the individual perception of the world is narrow and confined to a small group of immediate acquaintances.

That did not matter in factory-type of settings because physical tasks could be broken up. Bigger tasks could be divided by assigning people to different, smaller, fairly independent parts of the whole. Hierarchies made sense as a way to modularize work. The worker did not need to communicate with many people. The downside was a lack of flexibility. Reconfiguring a hierarchy always created a mess for a long time. And if you had a lot of interaction going on in a hierarchical structure, with many steps going up and down, it was slow and prone to misunderstandings.

For intellectual tasks, it is much harder to find parts that make for an efficient division of labor. Intellectual tasks are by default linked and complex creating an increased need to interact. Knowledge workers are often put in a position where they have to negotiate some understanding of what they face. The same event means different things to different people. The cognitive opportunity lies in the fact that as we don’t all select the same things, we don’t all miss the same things. If we can pool our insights in a creative, enriching way we can thrive in the complex world we live in.

New technologies give an organization the ability to reconfigure its form any way it desires. We are not confined to any one structure any more. The smartphone revolution has changed the logic of the network. The Web is no longer about linked pages but about connected purposes. We want to do something – with the help of information and other people. Often this means wanting to learn and respond in a situation.

Most often we seek two things: information and interaction.

For information the best structure would be a random, contextual network. A random network has the shortest possible path lengths. An example of this is performing a search. The key measure here is path length. That indicates how far everybody is, on average. The path length measures how many steps a piece of information has to go through between people. To create short path lengths in a typical hierarchical or process based structure you would need to know almost everything and everybody included in the hierarchy/process chart.  You would need to have access to information that we typically don’t have. Hierarchies and process charts are thus not efficient ways to organize knowledge work. They are not transparent enough.

For interaction, the challenge is engagement. Widening the circle of involvement means expanding who gets to participate. It is about inviting and including relevant, new and different voices. The measure is built on the social graph: how many of your friends know each other?

The network design principles successful organizations follow are: ( 1 ) shortening the distance between two randomly picked files/nodes/people; ( 2 ) getting more people who you know personally, to know each other.