While I see what you mean when you say that having the investigator know what he is giving the subject has no harm, in actuality the exact opposite can actually be the case. If the investigator knows he is giving the subject the actual product, they may end up unconsciously dropping hints through facial movements, verbal emphasis, etc. and that could skew the results in the end which would hurt the entire study. With a double blind study, the chances for that to happen are much less likely to happen because the investigator has no idea who is getting what, so any bias that could possibly occur flies out the window the in the end, the test gets more dependable results.