I've read a (MBTI) study that self-typings may be more reliable than typings done by "experts" anyway - this was tested by checking the correlation between two tests taking six months or so apart by each individual and two appraisals by the "experts".

Of course that was MBTI and most of us are probably more informed than these "experts"...but if self-typings on the whole are more accurate than peer review, then how are we going to test the reliability of self-typings on teh internet? Maybe we should just leave things as they are?