That would be my first thought, but that's not how it works. If 50% of people had placebo and the drug had a real effect on 10%, then: - Placebo results: 50%

That would be my first thought, but that's not how it works. If 50% of people had placebo and the drug had a real effect on 10%, then: - Placebo results: 50% - Drug results: 50% + 10% * 50% = 55% To detect a difference of 5%, you need way more people than to detect a difference of 10%, so your testing is now costlier and possibly slower. In the extreme, drugs that only benefit a few, or with relatively small but real improvements, will never get tested or fail testing, because of these placebo-folks