I attended a lecture from Conway where he described this approach -- but I don't recall him calling it FRACTRAN. I don't even recall what the particular example computed, but the program was around a dozen fractions.
This was probably at the end of the 70s while I was a Maths undergrad at Cambridge.