1. The Myers-Briggs personality types rests on the assumption of four independent dichotomies, namely:
    1. Extraversion versus Introversion
    2. Sensing versus iNtuition
    3. Thinking versus Feeling
    4. Judgment versus Perception
  2. At its core, the «personality test»1 consists of about 90 binary questions (that is, there are only two possible answers).2
  3. In theory, a person taking the test could answer to any of those 90 questions in any way.
  4. 90 binary choices give rise to 290 ≈ 1.23 × 1027 different results.
  5. There are approximately 7 × 109 people on Earth at the moment of writing.
  6. It is commonly believed that all humans are unique in some way.
  7. If all humans—and their personalities—are unique, then the different test results are enough for 290/(7 × 109) ≈ 1.76 × 1017 planets worth of people to have different results.
  8. The MBTI assessment then reduces them to merely 24 = 1.6 × 101 categories. This is a reduction of about 26 orders of magnitude.
  9. For comparison, 26 orders of magnitude are what separate the length of an X chromosome ( 7 × 10 − 6 meters) and the diameter of our galaxy ( 9.5 × 1020 meters)


The Myers-Briggs should be reclassified as a high-compression algorithm, able to take in several planets worth of inhabitants and (allegedly) classify them all using only half a byte. The potential applications of this algorithm in Computer Science remain to be discovered.


Yes, I’m aware that, since the questions ask about emotional preferences, they cannot be completely independent.

No, I’m not sure these personality types are an accurate descriptor of personality. I’m not sure they are not. This is more of a comment on how reductive «personality types» can be, both as a good and bad thing (remember, «all models are wrong…»)

Catalan's conjecture ⇐ Part of Brevity Quest 2020 (288 words) ⇒ ?

  1. Formally, a «Myers-Briggs Type Indicator Instrument»

  2. For simplicity, I will ignore momentarily the possibility of skipping questions.