(There are some fractions here, please persist. It’s worth it.)
Imagine that you have a daily drive. Half of it by distance is on dirt roads where your car can drive 10 miles an hour. And half of it is on a good road where you can drive 50 miles an hour.
Which is a better choice: Trading your car in for one that can drive 22 miles an hour on the dirt road but no better on the highway? Or one that can do no better on the dirt road but 200 miles an hour on the highway?
Imagine that your factory has two kinds of machines, all fully busy. Half of them can process steel with an accuracy rate of 2 in 10. The other half can do it with 80% accuracy. Which is a better investment: Tuning the lousy machines into 30% accuracy or making the newer machines perfect, with no errors at all?
And finally, what’s the best way to improve fleet mileage? To get the 14 mile per gallon Hummers to upgrade to Toyota Camrys, or to get the Camrys to convert to infinite mileage electric cars?
In all three cases, because you can’t average averages, the answer is to improve the laggards.
Here’s the arithmetic if you’re curious.
And we should be curious. Because it feels safer, more productive and easier to go after the devices or systems or people that seem to be so close to getting it right. But it’s the laggards that cost us the most.