Image via WikipediaWho is in charge ? The boss ?
Until some decades ago a farmer was totally in charge of his farm. He knew how and wen to work the land, how to harvest, when to put the bull and the cow together etc. He knew all the details.
The modern farmer nowadays has to rely on computer data to know when to inject what in which animal, on soil analysis data to know what and how much fertilizer to use. His is no longer in control. Because of the optimisation of the way of doing agriculture and the accompanying complexity he has to delegate a lot of tasks to others, to computers.
And here I am still talking of a one man agricultural company.
What does a CEO know about his 5,000 employee enterprise ? How much of the details does he know ?
Almost everything is delegated to his employees in an official hiërarchical structure and an unofficial but vital I-know-who-can-solve-this-problem cooperation network.
The result is that a lot of CEO’s get fired (they do not call it that way and they get huge ‘goodbye’ bonuses but the result is the same) becaus they cannot handle the complexity any more. Even with all the delegation down to all the hierarchical layers, he is bound to fail if he does not know enough of the structure and their interdependencies to be able to predict the consequences of his decisions. As the enterprises become still bigger, and altough things like knowledge management and enterprise 2.0 are getting more and more attention, the latter seems more and more impossible.
How did evolution solve this problem? There is not much hiërarchy in biology. You have the leader of the herd and that’s about it. Or not ?
It seems that there is a whole lot of hierarchy going on in our brains. When we want to walk, we do it just like that. No need to decide which muscle has to contract first, second, with what strength and speed. How does our brain accomplish this ? Apparently this complex hierarchy is made possible by a complex network of neurones that is sufficiently flexible to adapt and learn without collapsing.
In “Why the Demise of Civilisation May be Inevitable” (and see discussions of it by Al Fin and Detainees) scientists see the hierarchical complexity of our civilisation as the reasons of possible breakdowns of -little- parts of our civilization and the increasing complexity of our networks as the reason that this breakdowns will cause chain reactions of breakdowns up to the destruction of the way humanity now exists.
They can be right, they can be wrong (read the discussion) but why is the network + hierarchy combination in our brain so robust (I know people who lost parts of their brain and stil behave more or less normally) and why is the network + hierarchy combination in human societies so vulnerable ?
The difference between the two lies in the structure of the network. Human-made networks are … well, made by humans. This means that humans made networks with a much too rigid structure. They contain too much strict rules. It makes me think of the spaghetti programming style that was used in the 70’s and 80’s of the previous century. If you “pulled” on one string of the spaghetti, all the rest was affected. There were programs no one even dared to touch because you could not tell what the effect on all the rest was going to be.
Our brain networks are not designed, they are grown, just like everything in biology, just like ecological networks. It took evolution thousends of centuries to get to it.
Humans do not have that time, so we design everything to work optimally, but only under the given conditions. But conditions change from time to time.
The biological solution is not the optimal solution, but the one that is ‘good enough’ even in changing circumstances. That is “Why our brains are so clumpsy“.
That is also why data miners nowadays “grow” models in a similar way. In stead of using one statistical equation they use artificial neural networks, ensembles of weak learners, to grow some sort of holographic models : black box models where the rules are distributed troughout the whole model but which are very robust. For this robustness is exactly what data miners seek : form a given set of data (=specific conditions) make a model that can pinpoint the as correct as possible outcome for other (altered conditions) data.
Our financial networks need a similar solution to become robust enough to withstand events as a simple credit crisis.
The problem with our economical and financial networks is that “as correct as possible” is not good enough. The balances on our accounts have to be correct up to the penny. As long as no one will accept a payment of approximately 1,000 dollars we will live with the sword of Damocles above our head : will the system hold or crash ?