I reject the notion that it's just like a horoscope. I've routinely tested as INFP, leaning towards an INTP (based on their cognitive stack, a grey area between the two makes perfect sense), and reading about the two feels like reading a very accurate blueprint of my brain, whereas reading, say, an ESFP, feels totally foreign. Even something that's apparently "close," such as an INFJ, still feels totally wrong, like a shirt that fits well except for being too tight across the chest. If it were really like a horoscope, I could pick any of them out of a hat and find something to relate to, but -- I mean, I guess you've gotta trust me here -- that simply is not the case.
Perhaps I'm the "time at which the broken clock is stuck" and am benefitting from an accidentally correct explanation, but learning about and deeply understanding my "cognitive function stack" has allowed me deeper insight into my strengths and weakness (not what they are -- I already know that -- but why, and how they manifest in ways I don't always readily see), and has genuinely made my life better -- more so than any therapy sessions, and that's saying something, as I easily count seeing a therapist as one of my best decisions in life.
I suppose that my type more or less predicts that I'm the kind of person who would find value in tests precisely like these, and my test always has the "-T" flag (for "turbulent," meaning "not secure in oneself"), so it's certainly a chicken and egg problem. I take it with more than a grain of salt, but not a mountain.
I'm also not a "personality test" junkie at all. I remember a decade ago I tried a few random ones, and they were clearly all bullshit, more along the lines of a horoscope. MB is a different beast.
Furthermore, I later learned that I'm actually friends with the types of friends I'm "supposed" to be friends with, and in hindsight, we had bonded precisely the way the test had predicted. Same for who I'm attracted to, and who I'm repulsed by. It's not like I go around asking about peoples' types, but the handful of times I have, the conversation has been pretty eye-opening.
All I'm saying is: don't totally discount it just because you heard someone compare it to a horoscope once. There are a lot of flaws -- the most obvious one being that its accuracy depends upon the ability of the person taking it to answer authentically and truthfully, rather than what they wish they were like.
But my gut tells me that it's a good direction for science to go. Are there probably too few many types? Absolutely. But like I said, it appears to be an extremely high level abstraction of actual psychology, just as so much in psychology is (Jung's Archetypes come to mind).