Some people may score same results for the entire life, just as some unlucky molecula of oxygen can stay in same cubical centimeter of the room for a long time. Statistically it's possible. But it does not matter, when you need predictions for the entire ensemble. If real predictive power of MBTI is low, if the cost of mistake is high (e.g. when you do not hire a star because of this test or when you promote to management position a guy with promising test score), then this typology does not make sense and must be avoided. Since there's no scientific proof of predictive power of this theory, then it's garbage.