A high profile exercise in replicating the results of some high impact papers has found that only 2 out of the five such results could be replicated. Nobody has claimed errors or fraud in the original studies. It simply shows the difficulty of replication in life sciences, where systems themselves are prone to change and even evolve with time. It does show that those looking for quick bucks in medicine are confusing how science needs to develop and in trying to convert science outputs to marketable commodities – such as drugs or diagnostic tools – we may do harm to development of science itself. It also calls into question attempts to introduce randomised control trials in social sciences, as a kind of gold standard in science.