So much this! I wish I could remember the site where a medical researcher is going back and trying to reproduce past studies and finding something like 50%! can't be reproduced. A very close friend was telling me about an archaelogic dig he was on with a famous researcher that destroyed evidence that disagreed with his theories. Frustrating. At least our planes aren't falling out the skies so kudos to the physicists!
The Economics of Reproducibility in Preclinical Research
> ... An analysis of past studies indicates that the cumulative (total) prevalence of irreproducible preclinical research exceeds 50% ...