All human decision making, such as 'should this roster be modified or not?', or 'how good is this roster on a scale of 1 to 10?', will involve both bias and noise. Bias can be seen as the average error, whereas noise is the dreaded variation from time to time - even though we use identical input. Between, and within, individuals.
Daniel Kahneman, in the recently published sequel to the best-selling book Thinking Fast and Slow, argues that it’s time we pay more attention to noise in decision making. In the book, called Noise, he and his co-authors claim that the mean squared error is equal to bias squared plus noise squared (see picture). Something that sets noise as a big problem as soon as our judgement is involved. The inconsistency in our decisions, the noise, also effectively hinders us from assessing and understanding the bias part - which then blocks us from improvement.
Models, such as bio-mathematical fatigue models, or cost functions guiding an optimizer to the best overall crew roster solution, have a big role to play here. These models enable us to quantify risk or 'goodness' in clear exact numbers, even if those numbers may be somewhat abstract. We can follow how these numbers evolve over time and detect positive and negative trends. We know that they are off the 'true average' that we try to reflect - but they have the advantage of providing zero noise. The same input will always render the same output value. And that's a good thing; that is what makes it so much easier to systematically, step by step, reduce the bias by tuning and improving the model.
Model usage comes with many pros, one of them being the speed which is so critical in large-scale crew planning, but it also comes with a number of caveats. These caveats include a necessary awareness of the users of the bias and the need of customization and tuning to take 'ownership' of the model output. Please consider learning more by attending our extensive training courses for crew management solutions as well as fatigue risk management, found here. For a shorter interview with Kahneman about the new book and why reducing noise is so important, look no further than here.
Please also find, through this link, a document explaining more about how fact-based decisions are made in leading crew management solutions that daily affect some 450,000 crew world-wide.
Comments