October 18, 2006
Consensus Management - Why It Doesn't ScaleWay back when in the dim and distant mists of time, I posted some thoughts on how megatrends happen. To quickly recap the main points, I used a model that some physicists had created to help them understand the behavior of forest fires. One question in particular had them foxed: why were most fires small, burning an acre or so, but every once in a blue moon they'd get a huge fire that would lay waste to millions of acres?
So they created a very simple model to simulate what they saw as the key variables in the process. It was a game, where in each move a tree was randomly planted in a grid of squares. Every few moves, they wuold drop a match randomly on one square of the grid. If it dropped on a square with a tree in it, then that tree would burn. If there was a tree in an adjacent square, there wuold be a 50:50 chance that it would burn too. So a tree next to a tree next to a tree that a match landed on would have a 1/4 probability of catching alight. And a tree seperated by 3 degrees would have a 1/8 chance, and at 4 degrees removed it would be a 1/16 chance, and so on. I calcuated that the odds of a grid of 8x8 trees all burning because of one match were astronomically small - millions to one against.
But, given enough time, it might happen. And this is how those very rare but very big forest fires can happen. My point being that nobody can predict which match will cause the whole forest to burn. It's highly unpredictable. You just can't bank on anything happening on that kind of scale.
Which explains rather neatly why consensus management doesn't scale up. Let's imagine that the chance of building a consensus between two people is 50:50. I might be able to get a consensus with 2 people. It's half as likely with 3, and 25% as likely with 4 people. Building a consensus among 64 people is very unlikely. In a company of 1000 people, the odds really are stacked against you. Arguably, nobody should bank on reaching a consensus on a large scale. In that respect, consensus among a large population can be thought of as being effectively a megatrend!
And yet we still try...
Posted 14 years, 6 months ago on October 18, 2006