Just before Christmas I got a paper from Bill McKelvey of UCLA with a request to see if I could get it to the attention of Lord Turner who recently said in Prospect Magazine that London’s financial sector was swollen beyond its socially useful size. His paper addresses some of the reasons why that might happen, with implications for the strength of the pound, British membership of the Euro zone and so on. Now Bill is a key member of the academic group working on social complexity and the above view is not a consensus one. I think he is a bit hard on the UK, Pierpaolo Andriani, Paul Omerod and Max Boisot who were also a part of the last Durham Group meeting disagree with the proposition. However there is a germ of something here. So I will set the context here, and then you can read the paper and make your own minds up.
We are all familiar with the so called normal or Gaussian distribution; the bell shaped curve with its emphasis on the mean position provides the basis for risk assessment and much research the world over. The basic idea is that anything beyond a certain range (usually six standard deviations) is an outlier event that can be disregarded. The attraction of this in respect of managing for extreme events can be seen, as can its dangers if we look at the risk management around the New Orleans Levees. Of recent years people have become interested in another type of distribution, namely a power law or pareto distribution. Yes I do know about Zipf’s law etc but I am simplifying here. The important thing is to realise that in nature a lot of things turn out to have a power law rather than a gaussian distribution.
In fact if you take most population distributions (size of cities in a country, animals, financial trades etc.) and produce a double log scale of frequency against distribution you get the picture above. So there are a few large elephants and a lot of fleas, or a few large cities, but lots of hamlets and so on. There are fractal variants of this normally illustrated by Broccoli florets and the like, but the essential point remains the same.
Now the net result of this is a probability ranking as shown in the graph to the left. The key thing to realise is that so called low probability events on a Gaussian distribution turn out to have a much higher probability. Outlier events, disregarded by most research processes turn out to have far more significance than we thought. Events disregarded as highly improbable on a gaussian distribution turn out to be more probable than we thought. There are some pretty major implications from this, one of which is to realise that failure is far more likely that we think. Therefore a shift from a strategy based on preventing all probable failures may be wrong, instead in several cases we need to assume failure and plan for rapid recovery. In other words a shift from a strategy based on robustness to one based on resiliance. To give a practical example, if you have built a house on a water meadow then it is going to flood (its why its called a water meadow, its not estate agent speak). So instead of a conventional build you have stone wall, stone flagstones and the the electricity is in conduits that come down from the ceiling. Funnily enough that is how our ancestors used to build them (sans electricity).
Now there are lots of interesting implications here, and one of the fascinating links is that between SenseMaker’s ability to capture outlier events and power laws, both of which are being seen as two of the key new research tools in a world made aware of complex adaptive systems. I did a session on that with a group of academics at the Academy in Chicago last year, and we are planning a larger follow up in Montreal this year.
Bill is one of the academics involved and he understands power laws a lot better than I do. In this paper he moves things on a bit, arguing that the City of London has become disproportionately large. In effect he is arguing that a Power Law provides a natural constraint that could inform policy makers. its an interesting argument so if anyone has access to a policy maker in the City or Lord Turner and can get this paper to them please do. Otherwise all and any comments will be welcome.
My linking of fleas to the City of London by the way is gratuitous and deliberate in the light of recently announced obscene levels of bonus payments. These days the rats don’t just leave the sinking ship, they also want us to pay for the replacement and then charge us for salvage.
Wikipedia is a funny place. My first encounter still lingers with me and I ...
Just before Christmas I started off this thread by summarising the reasons that people ...