> Traffic flow may be best measured, not in feet or meters, but in multiples of the average car length.
And you've just thrown out the other advantage of metric: universality. The average car length is different in USA than in Europe. If you use "the USA average" to count flow in there, and "the EU average" to count flow in Europe, then you cannot compare those measurements afterwards. If you force EU to use "the USA average car length", then the measure is extra counter-intuitive for EU people. If you just use the "world average car length" then the figure is subtly counter-intuitive for most people.
That's why it is better to stick with a universal measure, and let that measure be easy to operate with. The metric system provides both features. Then just let people do what they're good at: figuring out what the "normal" ranges of values are and work from there.
> "you've just thrown out the other advantage of metric: universality"
Nope. I've done the opposite.
Let me explain: when you set up a traffic flow equation (or many other complex equations), you don't write it in terms of meters or feet or any other supposedly universal unit. You write it in terms of whatever the natural scale is -- whatever makes the math simplest. (You might define the average car length as "1", for example.) Then once you've finished solving your equations, you can rescale everything in whatever units are most familiar to the audience you're presenting to.
And you've just thrown out the other advantage of metric: universality. The average car length is different in USA than in Europe. If you use "the USA average" to count flow in there, and "the EU average" to count flow in Europe, then you cannot compare those measurements afterwards. If you force EU to use "the USA average car length", then the measure is extra counter-intuitive for EU people. If you just use the "world average car length" then the figure is subtly counter-intuitive for most people.
That's why it is better to stick with a universal measure, and let that measure be easy to operate with. The metric system provides both features. Then just let people do what they're good at: figuring out what the "normal" ranges of values are and work from there.