The prime example is Edison calling the air plane a toy and therefore declining to work on it.

No results that has not been replicated should be trusted. The problem is not that errors are made but rather that results are trusted before they are replicated.