I recall how once in a galaxy of my very early twenties I had been blessed with a dubious honor of refactoring & adapting one pretty ancient FORTRAN code from the area of kinetics and reactor design modelling. After some puttering about it, we discovered that whichsoever model it had been fed with, the result initially was the same. As it turned out, because of GOTO-not-so-graceful exit from a deeply nested loop that led one level higher than would have been healthy. Such is the dark power of GOTO considered harmful.