How to Forecast the Future

Computerworld - Silicon Valley is littered with the corpses of companies who mistook a clear view for a short distance, says Paul Saffo in this months Harvard Business Review. Even when you see a change coming, its difficult to estimate how quickly it will arrive, Saffo says, but understanding forecasting can help you turn uncertainty into opportunity. Saffo (www.saffo.com), a veteran forecaster who explores the impact of long-term technological change on society, shared some of the tricks of the trade with Kathleen Melymuka.


What is forecasting, and how does it differ from predicting? Predicting is about certainty, and forecasting is about appreciating uncertainty. Forecasting gives a context for decision-makers to act in the face of uncertainty. In the business world, uncertainty is our friend, because uncertainty is opportunity.

Why should a CIO care about forecasting? Forecasting is more important than ever for CIOs today because the CIO job function is central to enabling the corporate strategy, and the underlying tools that CIOs use are changing at the rate of Moores Law. CIOs live in the eye of the hurricane.


What is the role of intuition in forecasting? When youre making a decision, its very, very rare that you can have 100% of the information. If you wait until you do, that opportunity will have passed. So any strategic decision is going to require action before all the information is available, and it will require an intuitive leap. Its folly to rely on blind intuition, but forecasting makes your intuition informed intuition.


What is the cone of uncertainty, and how do you use it in forecasting? Its a way of thinking about what lies ahead. Youve seen it on the weather map: They draw a cone showing all the places a storm could go. Its also the cone of possibility: the space that encompasses all the possible outcomes that could unfold from some present moment. I think its a good visual aid. It forces me to make sure Ive accounted for all possible outcomes. Thats the essential step in forecasting.


What are wild cards, and why are they so difficult to read? A wild card is a trend or event that has either a very low probability  less than 15%  or a likelihood of occurring that you simply cannot quantify. Right now, an interesting wild card in IT is quantum computing: How quickly does it arrive? Its coming eventually, but is it a decade or two off? What if it happens in the next two to three years? It would completely negate all the advantages of public-key cryptography. There would be no secrets. The impact would be huge.