CommentStreams:5d5e79b31feabfe485874ac1228cd16c

It's an ideal, but human biases can interfere. Confirmation bias, publication bias (only positive results get published), and politically-motivated funding can all corrupt the process.