I just re-tested myself with the abridged version of the MBTI located at:
http://www.humanmetrics.com/cgi-win/JTypes1.htm
Funny thing is, for the first time in quite some time, my scores have been consistent. I tested as an INFJ for the third time in the last couple years, and when I read the portrait description, it fit like a glove. I skew towards the detail oriented portrait in that I tend to get annoyed when people ignore details. I've been doing my best to mitigate that by educating myself on the different personality types.
Working together is hard, but when done correctly, can really boost productivity and keep your team together. I'm very curious as to what the distribution of the HN community is.
For all of the "science" that goes into building the models for these tests (not just MBTI), any that rely completely on self-evaluation (especially multiple choice questions), should be taken with a huge grain of salt. They're fine for entertainment, but for anything practical they should be regarded as one step above astrology or haruspicy.
Just to provide an example of the worst case, I've seen companies hold personality test workshops, only to have that bind their thinking and give them reason to quash dissent ("Of course... that's what a XXXX would say" or "I really need a YYYY on this project"). Too many times these things turn into a substitute for actually getting to know your employees' strengths and weaknesses.
I don't want to be seen as overreacting to a simple harmless poll, but I do want to make sure people aren't giving too much credence to their (or others') results.
[1] http://en.wikipedia.org/wiki/Forer_effect