I am settled into Durham University for three days, with a small group of very interesting people discussing the subject of extreme events. Our guest blogger Max is one of those (if you have not checked out his delightful combination of wit with serious reflection do so) and there are other old friends. Bill McKelvey of UCLA, Peter Allen of Cranfield and others all brought to together by Pierpaolo Andriani of Durham. So expect three days of reflections and comments on this most important of subjects.
Day one is a series of 15 minute presentations (well most of us went to 20 or more) on our research interests. A fair number of the participants are dealing with mathematical models of various types. There is an extensive and ongoing discussions about power laws from most delegates. Those of us from the social sciences range from my use of narrative and cognitive approaches to complexity to a very traditionally minded approach based on decision analysis and scenario planning from George Wright of the business school in Durham I am afraid to say that his presentation represented everything I dislike about management “science” in business schools.
There is an important divide in the group between those who are trying to create models that will predict extreme events, and those of us talking about anticipatory awareness. Now the former group are resisting the word prediction, but they are in the main trying to create statistical and other models that explain past events and by implication this implies some predictive capability. There is also some debate about the relative accuracy of Joe Public as opposed to experts.
- A very interesting presentation on random drift. If you build a model with a set of agents each of which is characterised by a different colour and then set a rule by which each agent imitates another on each time period. If you have no innovation or change, then over time all the agents end up the same colour, i.e you get a bland commonality. If you add in a rule by which one agent is allowed to create a new colour, then blandness is avoided.
- Fascinating hypothesis to the effect that during the early phases of a market the creation of new nodes in a network exceeds the creation of links as a result of which network density decreases. During the mature phrase the rate of connection of linkages increases over that of nodes, network density increases. Now that would explain innovation and conservatism and might give us a way of measuring capacity over time. interesting that ….
- A debate on the ability of a crown to predict is justified by reference to cases (a lot from The Wisdom of Crowds) such as the inability of experts to locate a submarine in the Pacific. Now several of us challenged this. First we have the question of if you have the right experts, the second if the question of selective sampling. OK this is a success, now how many failures were there using the same method?
- Social and management science here are very lazy, they justify their theories by retrospective coherence, explaining past events. Repeatability is the essence of scientific experiment not retrospection. It’s interesting this need for either/or alternatives. Max makes the interesting point about what happens when a crowd becomes a mob. Now instead of multiple diverse opinions where large number mathematics will come into play, we have a coalescence of belief that is a tyranny of little wisdom.
- The big issue now is how to do inter-disciplinary or trans-disciplinary research. One of the reasons we are all here is that the EU rejected a joint research proposal on the grounds it was a series of discrete projects with only token linkage. This is fair comment, but there was little time and the only way forward. We need to create a bridge between the social and natural sciences, but we have no formal method or approach in which we can do it. It’s all too easy to fall back to a multi-project base as its more comfortable for the participants, it represents no threat of challenge.
OK, back to the debate (its getting interesting and frustrating in equal measure) more tomorrow.