Efficiency Can Get You In Trouble
JeanFrancoisPuget 2700028FGP Comments (2) Visits (5911)
This post was triggered by a recent tweet by Laura McLay. Laura was pointing to John D. Cook entry on how Effi
In short, someone was punished because he used his brain to find a more efficient way to perform a business function at the price of having to change some rules.
This in itself is certainly worth blogging about as John did. But why blog here?
Well, think about what happens if you replace brain by, say, mathematical optimization in the above sentence. You get:
In short, someone was punished because he used his mathematical optimization to find a more efficient way to perform a business function at the price of having to change some rules.
Could someone be punished because she used mathematical optimization to discover a better way to perform a business function? The answer is unfortunately yes.
While I was working as an ILOG consultant for some railway company, I saw the following. After some initial success thanks to our work, people started to look at the use of OR to improve various functions. Someone looked at an interesting problem about track maintenance. One of the tasks is to make sure you pour herbicide regularly on all track sections. It means that you need to have a special wagon run over all tracks regularly. The wagon is basically a tank with small holes, by which the chemicals flow on the track underneath. This is a nice combinatorial problem, where one tries to minimize the overall mileage of the special wagon (and its associated locomotive) as it minimizes the consumption of chemicals.
The person looking at the problem was able to reduce mileage, resulting into a 7% drop in chemical consumption. This was a remarkable achievement. Yet, that person was fired after explaining the result to his colleagues and management. Was it because his manager feared to be accused of having done a bad job so far? Was it for other reasons listed in John's post comments? I don't know.
But this exemplifies one of the issue practitioners must think about. When starting a mathematical optimization project, make sure that the people in charge of implementing the result computed by mathematical optimization are in support. This includes managers, but also people whose work will be impacted by the new way of doing things. Failing to do so will result in people trying to prove that the new way does not work, if not worse.