Limitations of personality systems
at least 1/3 of the discourse around them is outside the scope of usefulness

Disclaimer: This is a post about MBTI, Enneagram, Big Five, etc, written by someone who likes these tools and finds them useful.
ETA: This post has an entry on the “past follies” page.
Cate Hall just published an essay about the enneagram. I’ve been meaning to write a report on my enneagram type, (fragments included at the end of this post), but I’ve also been frustrated with MBTI discourse for a long time, hence this somewhat salty post instead.
Cate makes a number of very insightful points about personality systems (I’m not disagreeing with her, to be clear) that seem really obvious to me in hindsight. For instance, regarding the memetic fitness of astrology, 1) it seems to describe the personality space pretty well and 2) for the 1/12 people it sorts correctly, it’s useful (common blindnesses, etc) and 3) these people will, due to various selection effects, disproportionally encounter each other, hence some amount of predictive power.
My take, generalised, is that 1) these systems tend to accumulate a lot of “lore” that doesn’t actually improve the system, and 2) people tend to use them instead of actually engaging with the territory, i.e. over-reliance on maps (“How To” manuals for people, in this case.)
A lot of well—adjusted people are turned off when an enthusiast for whichever system tries to get them to take half hour long tests or read book-length descriptions. I know the books helped me with making sense of other people when I was 14. But it’s probably a good thing I didn’t fall too far into the rabbit hole, because a lot of the spaces developed around my MBTI type (INTJ) are disproportionately driven by hurt and resentful people looking for justifications to keep shooting themselves in the foot. (Or, at least they were when I sought them out then. I’m also not familiar with spaces for the other types.) (Caveat, it’s plausible the MBTI is more dangerous in this regard than the enneagram, but I won’t die on that hill.)
For MBTI specifically, the “useless lore” I think it has accumulated is cognitive functions1. I just haven’t seen anything convincing me they exist. Some people swear by them, sure, but I think the selection effect going on there is similar to the one for astrology (i.e. some people find a particular “stack” describes their inner processes well, and then they start trying to predict others’ behaviour with this process, with results somewhat better than chance.)
I think this is a failure mode of overly cerebral types, myself included. It can be very tempting to use an elaborate system like “cognitive functions” to understand other people, in particular if you’re pretty far from the median along important dimensions and your own experience is legitimately abnormal. It’s also a shortcut that can lead to problems down the line. The reason I say that is that I’m convinced (fight me) that every human fundamentally has the capacity for the entire human experience. Every motivation, every fear, the full spectrum of the conditio humana is open to all of us. We just vary drastically in our proclivities for specific slices of it. (Caveats around psychopathology apply.)
If you think cognitive functions are a thing, and that accordingly, your “interior Feeling” is developmentally a seven-year-old and your “exterior Feeling” will never outgrow infancy (or whatever the stack for my type, INTJ, is), you will probably just end up fulfilling your own prophecy. Try some “Fe” things instead, you might be surprised. I know I was.
Regarding the enneagram: Maybe this is me not understanding the system well enough, but I don’t get why the potential wings are limited to the adjacent numbers. As Cate points out, the enneagram becomes more predictive if you use continuous scores as opposed to concrete types. My hypothesis here is that allowing any type to have any wing makes it more predictive overall.
That said, I do have to give the enneagram an uncomfortable amount of credit. I spent most of last year systematically analysing certain aspects of my own dysfunction. It was engaging, like solving a puzzle. It made me feel useful - I might not get much else done, but at least I am charting territory, drawing maps so that others might benefit. It was addictive. I only started to get out of the spiral once I read a description of enneagram type 5 unhealthy behaviour. It was basically “tendency to create ever more elaborate mental constructs, while the feedback channel from reality loses more and more bandwidth. The way to get out isn’t to spend more efforts on the mental models while treating everything about your life like a state secret, it’s to re-open the feedback loop to reality.” Guess what worked.
Tl;dr, these systems (including Hogwarts Houses, …) are useful maps for navigating personal and interpersonal stuff, but tend to attract mythology around them that detracts from their usefulness rather than enhancing it. If the framework contradicts the data, don’t dismiss the data.
I meant to link to Anna Moss’ overview of the research, but it appears her website is gone?! There are some captures of oddlydevelopedtypes.com on the internet archive, but not the page I was looking for. If anyone wants to read either book (The secret lives of INTJs/INTPs), contact me, I have the pdfs/epubs and they’re in the public domain. Also, if you know what happened, please tell me!
> a lot of the spaces developed around my MBTI type (INTJ) are disproportionately driven by hurt and resentful people looking for justifications to keep shooting themselves in the foot
How could communities of INTJs form, without one of these masterminds masterminding a dating app which empowers them to shoot themselves in the foot *together*? Now you have me doubting their supposedly impeccable rationality.
I took the online test long ago and got INTJ, but only learned from your post that there were any communities built around it. Instead of finding those, I ended up in the LessWrong-adjacent sphere; at first I appreciated the tools for better evaluating and integrating my reading matter, and over time noticed a lot of the other people using those tools seemed similar to me.
> I think this is a failure mode of overly cerebral types, myself included. It can be very tempting to use an elaborate system like “cognitive functions” to understand other people, in particular if you’re pretty far from the median along important dimensions and your own experience is legitimately abnormal. It’s also a shortcut that can lead to problems down the line. The reason I say that is that I’m convinced (fight me) that every human fundamentally has the capacity for the entire human experience. Every motivation, every fear, the full spectrum of the conditio humana is open to all of us. We just vary drastically in our proclivities for specific slices of it. (Caveats around psychopathology apply.)
This is where I find the concept of incentives useful. Someone has a proclivity that makes some things easy, others hard. What is their incentive to do the hard thing? If they have one (e.g., wanting to make friends), then they are more likely to do the hard thing. This might also be framed as using a second-order desire to change a first-order desire, or to hobble one first-order desire so that a weaker first-order desire wins (depending on how you define first-order desire).
Goodreads summary
> INTJs: One of the rarest and most enigmatic personality types of all.
Enigmatic? I would have assumed we were considered dull because - relative to the typical person - our values, intentions, speech, and efforts tend to overlap more.
Just realised that there's a way to test whether any type/wing combination makes the enneagram more predictive. Cate links https://1231047546.rsc.cdn77.org/images/Ultimate_Personality_Test/Jungian-vs-BIG-Five-Outcome-Predictions.pdf
The test would be to repeat the above, but have one set with standard enneagram types and another where the first and second highest score are used to form type and wing. Then test against life outcome data. Am I missing something? Does anyone know of a publicly available dataset (ACX survey data? Aella's datasets?) that would allow this?