I mean it just doesn't seem to me that "b/a = 10^23" is harder to use in a numerical or intuitive sense, compared to "a = 31.5, b = -26.7; b-a = -58.2". Who cares if the important number is the exponent instead of the coefficient, just take the natural log of it and use that as your "useful value" instead, or something.
Like I could see if there were useful derived units that were a few layers of math away from candelas; off-hand, it seems intuitive to me that since a candela is Power/Radian^2 and since parallax measurements are an angle of arc anyways, you could get something that usefully replaced "magnitude" as a derived unit using "candela" + "diameter of Earth's orbit". (Not 100% percent sure how I'd go from arc angle to solid angle there but I dunno what would be most useful). And, like, maybe using a Luminosity Function weighted off like hydrogen absorption lines, rather than the wavelengths of light the human eye is most sensitive to; or outright dropping Luminous Intensity for Radiant Intensity to measure the total photon energy of radiation instead of "visible light."
Like the "well we keep using it because we've always used it" is a valid argument simply because there's all this knowledge inertia in using the same units we've always used and understanding what the Old Timers already wrote -- I still use "horsepower" to quantify how much electricity a motor should consume, even though the motor replaced the horse like a hundred years ago.
But it's like... "some guy put all the stars he could see in four buckets, and we took the average apparent brightness inside each of those four buckets and ran curve-fitting on them, and the best-fit logarithmic that came out of that is what we use to talk about how bright stuff in the sky is" just seems like a... painful way to measure stuff, no matter how important That Guy was.
It just seems weird to me, as like an amateur Space News Follower, that I've never even heard of an astronomer saying "Magnitude is a pain in the ass, here's a fresh new way to do it that we derived with dimensional analysis from fundamental units" even though that has been basically the entire argument that the SI Academy has been pushing in all the other sciences. That's why it's kilowatts instead of horsepower, joules instead of BTUs, etc. (Well, "as well as" not "instead of", but you know.)
Like maybe this is just me thinking about building instruments, but it seems like you frequently would care about "objective number of photons reaching you" for pretty much all measurements all the time these days, because "how many photons hit this particular pixel of the camera" is the basic measurement you're taking for each pixel. I mean even if you went back to plate exposures you're looking at how how many photons hit the plate, right.
And it seems like Astronomers frequently want to talk about "Absolute Magnitude", which seems to me less, I dunno, useful than Radiant (or Luminous) Flux, in a "plug it into equations and go actually noodle through the math you're interested in" sort of sense, when meanwhile getting from Candelas to Flux is pretty straightforward and isn't harassing you with intermediate complexity before you even get to the part you're trying to think about.