For a CIO, the years can be marked by the birth and then fading away of three-letter acronyms such as CRM, EAI and so on. They rush in with a flurry of articles and consultancies, pushing the next idea, and then gently fade away to make room for the next one. Few such waves are evaluated with hindsight once they have been superseded by the next one.

One that led to particular insight was SCM or Supply Change Management. This wave led to many successful systems, but almost without exception the predicted savings were greatly overestimated. Why? It turns out the costs of the system were dominated by exception processing. Although the run-of the-mill stuff went through efficiently with great cost savings, the very act of automating the system meant the cost of handling the exceptions ate up a lot of the savings made by the SCM. The system suddenly had to carry a post-it note down the chain and understand it at each point: "I know the screws are XML definition 3543 but make sure they are not magnetised like last time, there is no XML for that, and we can't use them if they are."

Read Mike Lynch's views on social media generation

A world that should have flowed as perfectly as intermeshed celestial spheres was defeated by the realities of our daily lives. The problem in IT is that we so want the world to be ordered.

We like a world of certainties, ones and zeros, blacks and whites, rows and columns, taxonomies, ontologies, rules, processes: a clockwork world of perfectly intermeshing brass gears like Mr Babbage's difference engine. In fact, certainty is our safety blanket - it may be looking a bit grubby but we carry it with us everywhere.

Our certainty is an illusion; we prefer exactness even if it's exactly wrong. I'm sure Enron's spreadsheets added up. They were certain, exact and yet wrong.

The reality is we must let go of this illusion of exactness and learn to embrace the uncertainty of the real world. It's time to leave the safety blanket at home.

This may seem radical. After all, we have some of the most eminent names in the history of IT currently trying to make a semantic web where we all agree on how to define everything in the whole world in a standard, agreed way. Somehow if we could only all agree, kill the butterflies of human ideas and pin them in the rows and columns of a defined display case each with label underneath, all would be well.

For those of you flat-earth, butterfly-collecting, ontological, semantic-tagging cultists, rather than my explaining why this doesn't work, let me point you to a far more eloquent version by Doctorow called Metacrap, easily accessible via that most vibrant butterfly, Wikipedia.

I remember in the early days of Autonomy our then CFO was asked what the effect of the dollar/pound rate would be on our results. He went on to discuss his view that the dollar would strengthen. He had missed the point; the job of a CFO was not to come up with an exact view of the world and its future course but to run the company to deal with the possible outcomes. Many economic experts spend all day trying to predict the dollar rate; his job was rather to hedge the business so it would be solid, no matter (within reason) which way the dollar went.

The same is true of the CIO. It is a mistake to assume a certainty and our ability to understand it. The job of the CIO is to embrace uncertainty and make sure the organisation and its systems are robust to respond to that uncertainty.

Lynch on cloud computing

Thirty years ago it was a safe bet that a large firm would be in the same business if you came back to look at it 10 years later. This is no longer true. As we know from events over the last two years, things are less predictable. We may not have expected the Spanish Inquisition but it was probably more likely than the collapse of Lehman's.

The Institution of Engineering and Technology, in an only semi-joking article, proposed the SI unit of uncertainty, the Maybe (symbol: My.) 50:50 is 1My, bears relieving themselves in woods 100My and Jeremy Clarkson becoming head of the Caravan Club 0.0001My; you get the idea.

Many ideas and events are coming together - the rise of unstructured information, agile development, meaning-based computing, real-time optimisation, closed loop analytics - all pointing to a new world where robustness and agility are intertwined.

Gartner's new big thing is Patterns, a strategic framework acknowledging the move to handling subtle, complex and uncertain reality.

The illusion of exactness may, like our safety blanket, keep us happy, but where there are real monsters under the bed it's now clear we have to face up to them. The answer is to think more about how to handle the maybes than to call which maybe will be the one.
We need to learn to embrace uncertainty because uncertainty is the only certainty.

About the author

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