More on extremities

A good meal and conversation last night ending with a gentle walk back through the cobbled streets of Durham to our hotel, then a short additional walk for me to the nearest WiFi hotspot (regrettably at the bottom of the hill). Back this morning after a wonderful sunrise over the cathedral but there was no charge in the camera, better luck tomorrow if I get up early. We are now back into a series of presentations so I will reflect as we go today. I won’t comment on everything. Where I do I will give it a headline to help readers sort out what they are interested in.

Are stock market bubbles good for us?

First up is Didier Sornette, Chair of Entrepreneurial Risks in Zurich. He is looking at why stock markets crash both in terms of proximate and fundamental cause. The latter is represented by the maturation of a market towards instability, the former compromised several factors, but imitative behaviour by traders is a major one creating the conditions for rapid and catastrophic change. All interesting but then a fascinating insight into the utility of market bubbles.

Normally supply and demand represent a form of limit cycle attractor, as one goes up the other goes down due to price changes. However during a bubble as the price goes up, so supply and demand go up with it creating the conditions for instability. Very interesting idea floated in response to a question: bubbles are good for us, they make us take risks we would normally avoid. Without bubbles there would be no progress, a good argument against equilibrium as a preferred state and a implied reference back to yesterdays lecture on the need for random innovation and change to prevent imitative behaviour creating bland conformity.

Do we really need corporate IT to manage networks?

We then moved on to Damiani and Buzna on extreme events in computer systems and networks. Some very important work here. I (and others) are increasingly arguing that corporate IT and managed networks “within the firewall” should be confined to transaction systems and lock down security on key data items. Social interaction within and without the organisation (and I include email, blogs, communities etc here) is best handled within the complex ecology which is the internet. Lively discussion on this with a general conclusion that the internet is more resilient, faster to adapt and ironically more secure that a corporate system. This for at least two reasons.

• Firstly human behaviour, you are more cautious so human factors and judgement come into play. Other evidence supports this. Social experiments are showing that removing traffic lights in urban areas improves road safety: drivers have to pay more attention to their total environment.
• Secondly the number of interactions and the number of operating agents within the net as a whole provides create capacity to cope with intrusion.

I will be picking up on this and related issues at my keynote address at KM World in San Jose next month (and KM Asia this month).

Violations and Systems Failure

Denis Smith from Glasgow is now up, great guy always good company aside from a perverse liking for Rugby League when the glories of Union are available. He is a deep specialist in crisis and risk management. Issues like how to evacuate Glasgow in the event or terrorist attack, adverse events in health care and the like. How can people like Shipman find their way through multiple protective mechanisms to become the UK’s most extreme mass killer. He is talking about how you prepare for systems. Lovely phrase: management is a degraded function, systems operate in degraded mode. If you look at most extreme events as single events then in each case management had to operate in a sub-optimal way. MBA’s train people to operate on the assumption that optimal behaviour and non-degraded functionality is possible.

Now referencing Weick and issues of order. I think this is one of the weaknesses of Weick’s normative approach in that he tends (not always explicitly) to talk about correctable error which while true at times is not universal and is only a small part of both the solution and the explanans. However his point that increasing dependence on order and structure is one of the ways that disaster happens. Tight control means people have to break the rules to get things done and the system is vulnerable in consequence. Now there is a lot in this, the informal networks that develop in any organisation are stronger in more rigid structures as the ability to work around the system is necessary for survival. That creates vulnerability to intrusion, but it also creates resilience in the event of failure.

Denis is now arguing that process is eroded over time as a result of structured management (my words as summary). Issue is not the controlled limits of contingency planning, but the period beyond those constraints (the point of inflection such as the a police officer closing the gate which exposes the weakness of the ground in Hillsborough).

Denis has a great metaphor of overlapping slides of swiss cheese to explain the way you achieve failure through multiple levels of defense. Each slice of cheese has a hole or two (that is in its nature). The crisis happens when holes in each slice align and allow penetration. System variety can create emergent pathways that cannot be anticipated but create potential for future.

Hierarchy as a means of energy dissipation ….

We now have a Cognitive Edge practitioner up, also a Doctoral student here in Durham. Chris Hamlin whose wife Penny was on the accreditation course in London this week. Chris is looking at network analysis. He is looking at company distributions in the petrochemical industry. Drawing a conclusion that innovation is hard, but once done it will spread quickly. he is using social network analysis on raw data here (which I think is legitimate, whereas basing it on answers to questions is not. He has an interesting entropy based hypothesis: Structural hierarchies evolve to facilitate energy dissipation and are therefore fractal in nature. This means that energy input would cause system change.

Using entropy has created controversy in the group, Boisot up first on the need for entropy levels in the physical system and the representational (information) system to match but they rarely are. Good point in response, namely that physical systems handle energy changes by phase shifts but we find that difficult in information systems. Massive dispute as to whether entropy or not works in social systems. Vote is against at the moment but its a fierce conflict. Peter Allen making the case against, to the effect the entropy is difficult if not impossible to define in far from equilibrium positions. He argues you should not transfer a thermo-dynamic law other than as an assumption, in reality open systems are symmetry breaking, while entropy is about symmetry. Powerful argument; I think I am with Peter and he has promised to pick up on the theme when he is guest blogger later in the year.

I am going to offend everyone who is not Welsh or Scottish

So opened Bill McKelvey of UCLA so this is going to be interesting. This is all on power laws (loosely defined these deal with non-normal distributions where the long tail of a Pareto distribution has significance. The argument is that power laws are ubiquitous covering everything from word usage, book sales, web sites, Cities ranked by population and so on. Bill is now comparing good and bad economies using power laws. I am impressed but not confident I understand on the screen. We have lines for the US, India, Bulgaria, Bangladesh, Mexico and the UK. It turns out that the UK has the worst profile and is thus broken. Now the original six EU members are up there and then the rest and the UK is worse than Malta. I really don’t get this. I am going to Mexico next week, but its not safe for me to drive over the border from Texas (or so I am told). Right, not its questions hopefully this will help me understand this. Actually the model is working with cities as a measure of economic performance. Now that is disputable. The US is a country of cities, the UK is more distributed, this could be cultural imperialism. The commuter range to London extends up here in Durham. Apples and Oranges I think. Didier is coming in now, arguing that a power law is a very poor indicator and general controversy. Bill is one of the best people I know on complexity, but I am not so sure about this presentation.

What happens after a collapse?

Pierpaolo has been very good about managing the agenda and cut his own presentation yesterday so we could move on. Now he is up and talking about a very interesting problem. One explanation is the barren landscape paradigm which says that after the collapse nothing is left. Looks like he is not happy with this. Looking at transferable skills. For example the way that watches (working with gears) allowed a complementary step to bicycle manufacture. In other words a capability can transfer but only when there is a collapse. Recombinant innovation allows surviving skills to re-architect around new applications.

His real argument here is that you should not assume that rebirth of an industrial area after collapse should not work on the basis of a green field, but should use the energy provided by the collapse to build in a recombinant way on what survives the collapse. Interesting issue of serendipity, is it just accidental or there permissive aspects of the system which enable the encounters. Interesting discussion here. If we look at the collapse of the coal industry in Britain in the 80s then where the mine was isolated (Bolsover) from other mines regeneration has taken place. On the other hand South Wales, where each pit was surrounded by other pits this has not taken place. Now this needs more research, there must be some level of diversity (requisite?) for recombination to take place and that in turn by be modified by questions of proximity.

Right, thats the talks over. A seminar tomorrow but now we have to work out how to do something with all of this so its time to end the conference blog until tomorrow.