Some prostate cancer is very slow growing. The man will die with it, not of it.

Treatment is harsh and aggressive and carries risks and has unpleasant side effects.

In that case doing nothing is better than doing something.

A very good question to ask of a doctor is "What happens if I do nothing?"

It is unlikely to work less than the placebo, just for 20 people you happened to choose some just got better anyway - and we're in the wrong group

That's why it's important to publish all trials - if the 20 people who got better anyway were receiving the active drug do you think they would be saying our drug might not be that effective?

It can have adverse effects worse than the nocebo effect (nocebo in Latin means "I will harm". It is the negative twin of the placebo effect).

I thought the nocebo effect described that a placebo even worked when the patient knew he was getting a placebo.

“In medicine, a nocebo is a harmless substance that creates harmful effects in a patient who takes it.”

http://en.wikipedia.org/wiki/Nocebo

It could be chance. Remember if you have a small sample size, it's possible that (by chance) the placebo control group just happen to get better faster, making it seem like the placebo works 'better'.

If the drug makes the condition worse?