In 1971 I arrived at university intending to read engineering and economics. I have spent my life developing mathematical models of physical systems and making “engineering forecasts” or predictions, as engineers call them. It is standard engineering practice to attach an estimate of error or uncertainty to a central estimate – to provide only a central estimate can be highly misleading. With proper analysis of the forecasting model it will be possible to attach a probability of exceedance to any central estimate. Andrew GarradBristol• Join the debate – email guardian.letters@theguardian.com• Read more Guardian letters – click here to visit gu.com/letters

Source: The Guardian November 24, 2016 18:58 UTC