> That's not how science works. If you have a hypothesis with no experimental evidence, discussing it as a fact is akin to doing astrology.

Sure, but you can discuss it as theory without implying that it's proven fact. Most (if not all?) theories that are now considered scientific fact initially went through a period where they were lacking in conclusive experimental evidence.

> I'm sure you're aware that most people recognize themselves in horoscopes descriptions of their personality. This is quite similar to that.

I am. And that's why I think the plausibility of the theory is important. There's a difference between saying "people think in way A and way B" and "whether people think in way A or way B is determined by the position of the planets at the time they were born".

I guess I would say that it's not quite true that there's no experimental evidence. It's more the case that the academic discipline "experimental psychology" has failed to find evidence. Other disciplines like cognitive science and behavioural economics tend not to have studied the MBTI specifically, but have found experimental evidence for theories that are substantially similar enough to the MBTI's hypotheses that IMO they do count as some evidence.

Specifically there is experimental evidence that:

- There are qualitatively different ways in which people can think and process information.

- People can have a stable preference for a given thinking process. And that such preferences follow a bimodal distribution.

- People can have a preference towards either introversion or extraversion (MBTI's I vs. E distinction)

- People can think either in a "Fast" way (fast, intuitive, perceptually inaccesible) or a "Slow" way (slow, reasoned, perceptually accesible) (MBTI's J vs. P distinction)

That's evidence for ~50% of the MBTI already, and would leave "Feeling" vs. "Thinking" (value judgement vs. descriptive judgement) and "Sensing" vs. "Intiuton" (concrete vs. abstract thinking) as interesting research questions.

The test-retest reliability of the test is bad, meaning that people who retest very often get different scores.

I think this is a very good indicator that the whole thing is bogus.

I'm of the opinion that the test itself is bogus, but that the theory it is attempting to test has merit and is worthy of investigation with better tests. Test-retest reliability doesn't distibguish between those two possibilities.

I always get damn near the middle of the chart, on both axes.

But I think this confirms your claim in my case. For nearly every single answer on the test, my response is "it depends." I really could go either way on nearly every single question because I need more context.

It makes me think the whole test is essentially BS, kind of like a horoscope where you can read whatever you want into it.

Yeah, the test isn't really very accurate. And if you're getting close on both axes that mean the test is actually telling you it isn't sure! (this doesn't mean your personality is actually close - it means the test doesn't know).

There are better tests (like this one https://sakinorva.net/functions) which will give you a confidence percentage for each cognitive function rather than just a best guess overall type. But none of them are that great, and currently the gold standard for determining type is to only use the test results as starting to point and to familiarise yourself with the theory to fine tune that, potentially with the guidance of an type expert.