yeah, but there are still a lot of constants even with the si units, which is because they predate the discoveries of many of the constants. more constants generally means more arithmetic operations you can accidentally leave out of your program and constants you can get wrong. this was a bigger concern in the slide-rule era, but the humans are still infallible, so it hasn't gone away entirely

i don't think it makes sense to describe miles or meters or stoney lengths as 'wrong' or 'right'; you can express the fitzgerald contraction or orbital speed or whatever equally well with any of the three, but one of them simplifies the fitzgerald contraction somewhat

I think a key tho in a measurement unit that’s desirable is its relationship to something consistent everywhere in the universe, such that the reference unit is reproducible by anyone. This doesn’t matter much day to day but as a system we should converge to one that anyone anywhere can agree how long a distance may be without a mutually agreed upon calibration but can calibrate without access any specific reference unit. The triple point of water is always what it is, and a colony on mars without easy access to something locked in a room in Paris or whatever can create their own reference measurements from whole cloth and they’ll be the same as the colony at Alpha Centauri now and in 20,000 years.

yes, even before alpha centauri, a big problem with the kilogram artifact locked in a room in paris was that it keeps changing its mass and nobody knows why or exactly how much or in which direction, although it's probably a fair fraction of a part per million by now

agreeing on distances with a faraway colony is actually a considerably easier problem than the kilogram; if we transmit them a radio or laser message, they can measure its length to within parts per billion, and we can do that with 01950s technology, while kibble didn't invent the kibble balance (previously known as the watt balance) until the 01970s

(you do have to worry about redshift: one part per billion of redshift is 300 millimeters per second, so you have to know the relative velocity of alpha centauri to within meters per second to correct for it—and, while that's also the 01950s technology of measuring spectral line frequencies to that precision, that same technology is what allows you to do without the radio message entirely)

transmitting or storing the number "9192631770/299792458" can be done with significantly higher exactitude, of course, and can be done over twenty millennia more easily than transmitting a radio message can