by Mike Lynch

Only CIOs are in a position to take a look at the whole picture

Sep 12, 2010

The job description of the CIO seems rather mercurial: they have to understand technology, business, legal and accounting and project management, but most of all they have to be practical experts at understanding systems. This ability to look at a whole system encompassing many components from technology to people to policy to markets may be a much-needed skill in this age of complexity.

Ever since I was a boy, I have had a morbid interest in plane crashes, not for the drama, but for the intricate process of elucidating the cause. Perhaps it was this that helped launch me into a career based on the mathematics of complex systems.

There are shelves of books on such doomed flights and reassuringly the final chapter always gives a simply stated cause. However it is rarely that simple. Yes, the turbine blade exploded, but what if the flight control position had been designed to run further from the engine? What if the pilot had tried to use engine power to steer? What if the inspector had spotted the turbine manufacturing fault 17 years earlier? What if he had not been getting a divorce and had been sleeping better, and so on and on.

We are taught that we must always have a clear, stateable and defendable reason for doing something. But studies have shown that how we make decisions is actually very different. We often fold in a large amount of subtle, unstated information, make a gut decision and then, when asked to justify it, retro-fit a simple but illusionary piece of reasoning. The world is too complex for the simple, justifiable explanation, but the subtle one is not acceptable to state out loud.

Such retro-fitting was rife during the financial crisis, when explanations ranged from bankers’ bonuses, collateralised debt obligations and inadequate capital ratios to the very wiring of the male brain. But we crave our daily dose of reason because with it comes a solution.

Since the invention of fractional banking and the innovations that helped fund the Tuscan wars, the financial system has become exponentially more complex. Now each year we trade in exotic financial instruments of a value many times that of the world’s actual output. Our technology also means it executes faster and faster, doing those trades in milliseconds. Within all this lurks a deadly killer: feedback.

Those of us unfortunate enough to have held a microphone at a wedding in front of a speaker and have heard the howl of feedback have some idea how a stable system can career into chaos. In the financial world we have seen even baby feedbacks bring down systems, for example the Lloyd’s reinsurance spiral. Many CIOs know this only too well, as they have seen the unintended effects of feedback as systems such as the corporate network operate in exactly the opposite way to what was expected.

So how does feedback arise in the financial world? Bank A takes on some exotic financial instrument and lays off a part of the risk with Bank B. Then B does the same with Bank C and so on. Eventually A gets a new deal with Z, not realising that buried in it is a part of its original deal. The feedback link is complete, the uncorrelated becomes correlated and chaos ensues. This effect became obvious in the last meltdown and helped bring down the Too-big-to-fails.

Even a relatively simple feedback system can keep maths PhDs off the streets just answering the question: “Is the system stable?” The problem is that feedback in such systems is unpredictable in its effects. Reducing risk in one part of the system can actually make the system less stable. Counter-intuitively, it means that actions such as improving capital ratios, reducing bankers’ bonuses, circuit breakers and so on can make the system even riskier. To make the right decisions we need a system view, not a component view.

The arguments for treating each part of the system separately were blown apart by the collapsing of de-correlations during the last crisis, but most of the people involved in solving these problems have, unlike the CIO, no practical knowledge or understanding of complex systems.

In short the ‘clever’ people such as the bank chairman or regulator, who reassure us they understand all this stuff and have it in hand, do not. Few seem to get this system view and so we will continue to tinker with the components.

So where are the certainties? Firstly, the innovation of mankind will lead to more complex financial instruments and systems. Secondly, our technology will execute faster and faster beyond the speed at which regulators can react. Thirdly, there will be more feedback.

Without the kind of system-wide thinking that CIOs use every day and that so many others cannot comprehend, perhaps we are doomed to the avoidable result of the increase in complexity. Perhaps it is time to look for different ?skills in our financial system’s pilots.

About the author

Mike Lynch is the founder and CEO of UK software company Autonomy