Lets try a thought experiment.
You regularly carry out a particular action on particular subjects. An alternative method is suggested. It is clear that overall this method produces better results across the range of subjects than your method. Indeed, studies show that overall it is S% more effective.
However, over the years you have become quite experienced in the method you use to carry out the action. For X% of the subjects it is A% effective, for Y% it is B% effective and for Z% it is C% effective.
For an experienced practitioner of this method it is D% effective for U% of the subjects, for V% it is E% effective and for W% it is F% effective. Overall it is M% more effective.
Generally, it takes T years for a practitioner to become as expert in this method as you are already in yours. G% of practitioners never become as effective, H% become as effective and I% become more effective.
Now there are at least a couple of ways of looking at this problem.
1 – Accept that the method is better overall and change to using the new method.
If you do this you are accepting that in making the change there will likely be a dip in performance and output before the improvement takes hold. You are also accepting that in some circumstances the outcome for some subjects may be sub-optimal compared to your use of the previous method. You are also accepting that you may not end up performing at the anticipated level.
2 – Stick with your existing method
In doing this you are accepting that the outcomes for your subjects may not be as good as it could be. You are accepting that your own performance could be sub-optimal.
Each of these approaches has their merits, and, depending on the circumstances, is defensible.
One, changing method, looks at all subjects and seeks to deliver the overall best outcome possible. It is a sensible approach when it is not possible to easily (or reliably) identify the which subjects would best benefit from which method. The other concentrates on the individual subject and seeks the best outcome on a one-by-one basis. It is a sensible approach when one can easily (and reliably) identify which method is likely to deliver the best outcomes for that subject.
So we can introduce another variable to assess the likelihood that we can correctly assess which method is best, K.
So now we have A. B, C, D, E, F, G, H, I, K, M, S, T, U, V, W, X, Y and Z.
Now there are cases where the rationale for change is overwhelming, where the change mechanism is so simple, where all the percentages clearly line up in one direction. In those cases change is a no-brainer.
But arguments for change are rarely this simple.
Where the percentages are close, or in conflict the answer isn’t always as clear cut. Perhaps where T is nearly as long as the remaining period you have carrying out any method then the rational approach is not to change. Or perhaps in circumstances where a short term dip in performance is likely to be very heavily punished in some way then the rational approach is not to change. Or perhaps where the new method is itself likely to be transitory. Who wants to put that much effort into change when another one is going to come down the pipe as soon as you acclimatise.
This isn’t an argument against change. It is a suggestion that perhaps change needs to be better described. It is insufficient to quote just ‘S’ as the justification. ‘S’ doesn’t always over-ride A, B, C, D, E, F, G, H, I, K, M, T, U, V, W, X, Y and Z.
This video is illustrative of the problem.
The bricklaying machine can lay bricks up to 30% faster than a human bricklayer. ‘S’ = 30. Job done. Get rid of human bricklayers and just use the machine.
But hang on. What if you aren’t just building straight walls? What if you are working on a restricted site? What if the machine breaks down? What happens to ‘S” then?
What I would argue is that the headline figure is always just the starting point, it’s not the end of the argument. I can agree that a mechanical bricklayer can lay bricks faster than a human bricklayer and still hold the rational belief that replacing all bricklayers with machines is a bad idea.
All the above arguments have multiple applications, which we can call ‘N’.
And are likely to be correct in R% of cases.
Now I’m starting to run out of letters…