It was bad. Really, really bad. The local chapter had a tiny handful of post-retirement members. I knew a ton of folks in my community at the time, a bunch of them were smart, active people, and none of them were members. It really looked like the only thing the Mensa folks had going for them was having done well on a test.
The newsletter was hilarious. One of them had a feature article on which was the better superhero, Batman or Superman? The article and the responses to it mostly centered around which of the fictional comic book characters had a bigger IQ. The writing was juvenile (and not because it was written by kids), and there was a regular feature that would pick some kind of trouble in the world and ask readers how they'd solve it. None of the responses ever argued from any kind of expertise or domain knowledge or offered any new insight; the theme tended to be, "well, I'm pretty smart, so this seems like an easy problem to solve..."
The discussion on HN on any given topic -- and the topics themselves for that matter -- tends to be a heck of a lot more interesting. I keep coming back here because of the people that just randomly show up in discussions who are actual experts on the topic. That doesn't really happen in Mensa.
My takeaway from that whole thing pretty much agrees with what you're saying. People who are Mensa members are members because their IQ is really important to them to an unhealthy extent. That would seem to correlate well with other psychological and physiological problems.
Probably anybody who reaches the level of 1 dan or higher in Go has at least a Mensa-level IQ. Researchers wanting to do further work in this area could try polling those folks too.