When I worked in the oil industry, a great deal of money was made trading crude oil. At the time, each barrel of crude would change hands about 5 times on its journey from the oil well to the oil refinery. Oil companies have refineries all over the world and each one is different; some cannot deal with certain crude oils, and the value of the products each one can make from any crude is different. I was working with an oil company and it was obvious to me that the value of any cargo could be established based on the distance to any refinery and the value of that type of oil to that refinery, and this information would support any buy or sell decision. What was not at all obvious was how the oil traders made their decisions to buy and sell any cargo.
When we looked at what was happening it was clear that the refineries weren't always getting the best crude oil, largely due to trading decisions. When we suggested a method of calculating this value for each cargo to help traders make better decisions there was a lot of resistance. This was surprising because, at the time, it was common for each refinery to optimise its operation based on any particular crude oil. So whilst it was sensible and commonplace to use optimisation to deal with a problem, it wasn’t the same when it came to avoiding the problem in the first place.
This is, unfortunately, very common in other sectors. Big, often irrevocable, decisions are made using little more than spreadsheets and then a lot of effort is applied to improve performance. Should it not be the other way round? Technology is available nowadays to model options and uncertainties, but I fear we have same problem that we had with the oil traders. People who are highly paid to make decisions don’t always appreciate the possibilities and how optimisation technology could help - not replace - them.
