I hear the imperfect measurement argument used in two completely different ways. Some people use it to argue that IQ scores are basically meaningless and should be ignored entirely. Others say the imperfection is minor and the scores are still the best tool we have. I want to know what imperfect actually means in this context and whether the limitations of IQ tests are serious enough to undermine their usefulness or just minor caveats that careful interpretation can handle.
The people using imperfect to mean basically meaningless are making a much stronger claim than the evidence supports and it is worth pushing back on that 
Every psychological measure contains some degree of measurement error. The question is whether the error is large enough to swamp the signal. For well designed IQ tests the answer is clearly no. Scores predict academic achievement, job performance, health outcomes, and life expectancy with effect sizes that dwarf most other measures in social science. The imperfections are real but they are the kind that careful interpretation handles, wider confidence intervals at extreme scores, some sensitivity to test anxiety and fatigue, modest cultural familiarity effects on specific subtests. None of those undermine the core validity of the construct 
Saying IQ tests are imperfect therefore useless is like saying maps are imperfect therefore navigation is impossible 
The map does not need to be perfect to get you where you are going. It just needs to be accurate enough to be more useful than having no map at all 