The survey results are in, so I am updating this:

If you scroll down to "MYERS-BRIGGS" you'll see that there are 436 people in Yvain's selection of results (of greater than 10 people for each type, leaving out a total 3.1% of the survey data). That's what these figures are based on. (The raw data is missing around 10% of the responses due to people wanting anonymity, and the graphic provided to show more detail has some issues so I used Yvain's selection.)

• Ballpark 90% Introverts: Correct

  371 Introverts (85% of 436)

• Most common type: INTJ: Correct

  163 INTJs (37% of 436)

• NT types > 75% of the population: Correct

  371 NTs (85% of 436)

• NF types - a handful or none (possibly more than ISTJs) : Correct

  51 NFs 436 (12% of 436)

• ISTJ a single digit percentage of the LW population: Correct

  14 ISTJs (3% of 436)

I wasn't sure exactly how I should interpret the somewhat vague "a handful or none" for NF types, but I see that I used enough numbers to be able to do a literal, mathematical interpretation so I chose that method. I had predicted it was possible that there would be more of them than the ISTJs who I had predicted would be in the single digit percents (implying that 10% or more of them wasn't outside the range) and that there could necessarily be no more than 25% of them because it would contradict the NT prediction, so since they were within the numerical bounds, I interpreted this as correct.

Another interesting thing to note is that each personality type in the top 98% of LW personality types is in the same order as the type list I wrote here. Unfortunately that comment had been previously edited, so whether or not you believe that I did this intentionally will be based on how much you trust me not to lie and what you think the probability is of me having the ability to correctly list the personality types of 98% of the LessWrong population in same order as we'd see on the actual personality test results after having proven to you just now that I can make correct predictions about the Myers-Briggs personality types on LessWrong.

What's really interesting though is that our personality type pattern matches the pattern Mensa discovered when they did a personality type survey, and the pattern that Mensa and LessWrong share is very different from the ordinary personality type statistics. This makes the IQ figures on the yearly surveys more believable.