Is it possible that in the midst of their no-comments, no-test-cases spaghetti they introduce a few bugs that affect the results? If so, they need to release their code to be scrutinized along with their findings.
I don't see where I said it was spaghetti code. Sometimes a goto in C can save quite a few running cycles. Write a quick Runge-Kutta-Fehlberg integrator without it.
