I was thinking more about social experiments than medical ones. A lot of problems with current social science research comes from telling people they're part of a study. Humans change their behavior when they know they're being watched (not to mention selection bias; people who agree to be a part of the study are often a very specific population subset, on the intersection of people with too much free time and the kind of people who like taking part in experiments). I think we need more studies like that infamous Facebook one (the one which media took and turned into an overblown ethical issue) - tweaking something and observing the dynamic response of the system, without the system being aware that it's being explicitly influenced.