A double-blind study is necessary in cases where the investigator may become biased if they had knowledge of which is the effective drug. For the example presented in the question, with a drug to treat Alzheimer’s the investigator may selectively distribute the drug. In the unlikely case that they personally know someone in the study, they may choose to give that person the effective drug because they want them to be treated. Additionally, the investigator may skew the results and not randomize the distribution so that their data is favorable to the hypothesis.