Not at all sure where this argument is going.
In the process of bringing new drugs to the medical marketplace, there is a well established process. Laboratory tests and proof of concept. Animal studies as proof of efficacy. This is known as the Pre-Clinical Phase. Then Clinical Trials - 3 phases, 1 through 3, each requiring human volunteers, each phase testing for something slightly different, and requiring a different scale of volunteer numbers. I have simplified the process enormously, for space reasons. Each phase is a major undertaking in itself, costing potentially millions of pounds. Wiki has a pretty good page on drug development, if you wish to read further..
http://en.wikipedia.org/wiki/Drug_development
In the phase 3 trials, where a drug is being introduced to treat a known disorder or disease, the gold standard of testing is a Randomised Double Blind Clinical Trial. As part of this process the volunteers will be randomised into a group that receives the trial drug, and a group that receives a placebo - although if you are developing a new drug for an existing and known disease, better practice would be to test against the current best performing drug/treatment protocol - This is something pharma companies often fail to do, preferring to use a placebo control instead, for obvious commercial reasons.
If you are developing a brand new drug for a brand new condition, one only recently diagnosed, any drug would still have to follow the same process as above.
Whether a drug is deemed to be successful or not will depend on confidence levels.
Only in those conditions where the disease is invariably fatal will some of these control measures in stage 3 trials be relaxed, to allow for experimental treatment, where patients can sign up under informed consent conditions.
It would obviously be highly unethical to randomise a trial into a drug cohort and a placebo cohort of patients where the disease they were suffering from was terminal.
Some drug trials where you have a treatment cohort and a placebo cohort are terminated early because the early data from the drug cohort shows a significant improvement in the diease being treated for, since it would become unethical to knowingly continue to offer placebo when a truly effective treatment is available.