Reproducibility in Data Science

I think SMBC comics made a joke along the lines of “p-value acceptance limit dropped from 0.05 to 0.03, half of science invalidated”