i don't know where the exact cutoff between min/max int is before it's converted to a float; i figured it would be -2147483648 and 2147483647 but those seem to be converted to floats as well
I think it's 1 inside those.
Floating point numbers (floats) can represent both small fractions and very big numbers, but not at the same time. If your number is above the ~2,000,000,000 limit of an integer, so that MMF represents it as a float, you're not going to reliably be able to add "1" to it because there won't be the precision available to do it.
Though I know most of MMF can handle double-precision numbers now, which have 48 bits of precision instead of 23, so you'd probably have to go over 281,474,976,710,656 before you couldn't add just "1" any more.
EDIT: No idea about the limit on paragraphs if there is one, sorry.