My last post argued that you shouldn’t look for perfection, or the one “right” answer, when an acceptable answer will do. Well, I agree with myself. And considering that Voltaire and Montesquieu also agreed with themselves on this subject, I feel in good company.
In this post, I want to reinforce the argument against looking for the “perfect” solution, as it is in many cases suboptimal. In my next post, we’ll see how it can even become downright dangerous.
Perfection isn’t free
For the sake of this conversation, let’s define a “perfect solution” to a problem as one that is both the most effective (it solves the problem) and efficient (the solution achieves more for less).
If a better solution is, well, better than a good one then it follows that a perfect solution is necessarily better than a better one. (Pretty bad writing, huh?) Let me rephrase: All other things held equal, “perfect” is probably better than “better.” But all things aren’t equal, because chasing perfection has diminishing returns, and as you continue investing time and resources looking for perfection, the cost of opportunity rises. Too much deliberation is expensive.
Another drawback of looking for perfection is that perfection is intolerant. Let’s discuss.
Favoring effectiveness over efficiency
If you’re a Swiss watch manufacturer, no tolerance (i.e., no wiggle room) in your mechanical creations is a source of pride. Whether the watch works after it’s run over by a car or submerged in a toddler’s bathwater isn’t the value of a Swiss watch; instead, Swiss watches are prized for their mechanical precision, their hand-assembled, artisan-created craftsmanship.
In many other cases, making a system more robust—making it able to function despite the failure of some components—usually requires making it less efficient.
For example, engineers design electrical grids to operate even if a major component in the grid fails; this the N-1 redundancy principle, where N depicts the minimum number of components necessary for operation. Similarly, the flight control system of some commercial airplanes have triple redundancies. In a more business context, you may decide that your execs shouldn’t all fly on the same plane. Here, effectiveness (an ability for your business to continue operating despite the plane going down) trumps efficiency (getting the execs where they need to be cheaper or faster).
Being less than fully efficient, these systems aren’t perfect by our definition above, but they are excellent, especially in adverse conditions. In these cases, “best in class,” or effectiveness, means being able to operate despite partial failure. “Lean,” or efficiency, means getting the job done with as little resource as possible. And sometimes—most of the times?—you can be best in class or lean, but not both.
If you’ve read my book on strategic decision making, you’ll note an important implication to this point, as this is another instance where MECEness isn’t desirable. Indeed, the MECE principle—“no gaps, no overlaps”—implies that any redundancy is undesirable. But that’s too broad a prescription, because some redundancies can be desirable.
Find your point on the spectrum
Therefore, on a continuum of possible solutions, from horrible to our definition of perfect, you want to be on the right side of the spectrum graphed below, but not necessarily at the rightmost point.
Does this mean you need to make unacceptable compromises? I’d argue otherwise: You need, rather, to adjust your definition of perfect given your specific situation and needs.
Otherwise, a rigid search for the wrong idea of perfection can be downright dangerous.
Golany, B. and E. Tamir (1995). “Evaluating efficiency-effectiveness-equality trade-offs: a data envelopment analysis approach.” Journal Management Science 41(7): 1172-1184.
Montesquieu, Pensée nº 1007, XIII. “Le mieux est le mortel ennemi du bien.“
National Research Council, Terrorism and the electric power delivery system, 2012, p. 26.
Voltaire, La Bégueule, 1772, p. 3 “Le mieux est l’ennemi du bien.”
Yeh, Y. C. (1996). Triple-triple redundant 777 primary flight computer. Aerospace Applications Conference, 1996. Proceedings, 1996 IEEE, IEEE.