The failing is having a simple model that assumes that any one paradigm is ever best.
A multi-paradigm language will always have the capacity to be better, or equal, by any metric.
I also strongly disagree with this sentiment based on experience. There are some tiny niches where functional programming is particularly useful, but outside of them the code is classically difficult to read, write or debug.... a problem made worse by the, very honestly, absolutely appalling quality of almost all example code - normalising one letter variable names, squeezing things onto one line and all kind of other terrible practises.
these are deeply serious problems that fp fanboys and academia seem to glaze over, presumably from a lack of real world experience, or appreciation of how valuable these qualities are ...