Yes - it makes about as much sense as any mapping into predefined fictional categories. The anti-elevator pitch for MB is: two imaginative persons of no scientific background read a bunch of psychology, invented a model around what they learned, and no-one would have heard of them since - except, in came a consultant who said "I can sell this! Effing yay, bring the cash in!"
Note the tendency of people to invent all sorts of things in falsifiable fields - like perpetual motion machines, and so on. There's no falsification mechanism in psychology...
Analogously, I'm pretty sure if astrology used some non-falsifiable grounds for it's input variables it would still be a reputable academic pursuit.
The statement about falsification mechanisms is incorrect.
In fact, one of the problems with the Myers-Briggs is that it is continuously brought up in these discussions, even when experts in the area of behavioral individual differences dismiss it because it is lacking in evidence (and it has never really been dominant, or at least not for decades). There's all sorts of model-testing that lends support to some models (e.g., the Big Five, which is mentioned in the essay) and not the M-B, in terms of its internal empirical characteristics and predictive properties. And they do involve falsifiable predictions of multiple sorts.
The problem is that people complain about nonsense such as the M-B being nothing more than a money-making consulting scam, but then don't take that assertion seriously, in the sense that they assume the consultants are scientists.
It's as if con artists were selling perpetual motion machines, and HR departments started buying them, and then we started complaining about physics being non falsifiable, rather than about HR departments and business administration not understanding physics. It's all strawman arguments.
As an exercise, for example, I recommend someone searching for modern basic research using Myers-Briggs uncritically in mainstream psychology journals. You probably won't find it except for as some kind of deceptive ruse in an experimental protocol.
As for measuring behavior, any measurement throws away information. That's the tension: weight, BMI, blood pressure, temperature (under what pressure?), etc. The problem isn't in the measurement, it's in how that measurement is used and interpreted, and how much information is thrown away.
So, nothing is wrong with getting a measure of emotional-behavioral state. The problem is overgeneralizing from that, across time or situations, overestimating its predictive information, failing to consider uncertainties or biases of measurement, and so forth. BMI is an imperfect summary of someone's physical health, but it does have utility. The danger isn't in BMI per se, it's in assuming it won't change, assuming things about the reasons for a given BMI, ignoring how any given BMI was calculated, and so forth.
The problem with non-falsifiability is that let's say we avoid 'overgeneralizing' from a categorization as you suggest. And instead we stick to whatever tightly constrained region of classification or prediction you'd consider acceptable. And so we take a Myers-Briggs test and it says this individual should exhibit this class of behavior. And it turns out they don't. Would this pose a problem to Myers-Briggs? Not in the least. Okay, what if it was a hundred? A thousand? A billion? There's no magic number where it's suddenly a problem.
I do agree with the person you're responding to that astrology would likely still be considered somewhat scientific if it didn't rely on things that we know to be false. For instance astrology mixes Mercury starting to go backwards as a key player in its predictions. The problem being there that Mercury doesn't go backwards. It was/is an optical illusion based on an inaccurate understanding of our solar system. But outside of getting some things fundamentally wrong astrology is the same as any other unfalsifiable model. Being wrong doesn't matter, and you can just constantly add onto it and claim you're 'refining' it.
Maybe even that geocentric model though is the same story. Part of the reason the geocentric model of our universe lasted so long was because, with the technology at the time, it wasn't completely falsifiable. Mercury needs to go backwards to make this model work? Other planets need to go into crazy floral sharped curvy patterns to make it work? Well okay then I Mercury goes backwards and planets go in floral shaped curvy patterns. If you wanted to suggest a different model, such as a heliocentric one, that'd involve throwing away literally centuries of work and entirely discrediting astrology (which was at one time a pursuit as scholarly as any other) as a science. People blame the church for the geocentric model, but there was much more to it than just that.
At least weight, BMI, pressure, temperature are measurable quantities. They don't tell the whole story, but what they do indicate is reasonably accurate and comparable and reproducible.
I don't think that can be said for characteristics of someone's personality. It's hard to even define them, much less measure them.
Note the tendency of people to invent all sorts of things in falsifiable fields - like perpetual motion machines, and so on. There's no falsification mechanism in psychology...
Analogously, I'm pretty sure if astrology used some non-falsifiable grounds for it's input variables it would still be a reputable academic pursuit.