User blog comment:LittlePeng9/Higher order set theory/@comment-24509095-20141010045633/@comment-1605058-20141010054510

I think it's reasonable to think that general framework for working with Turing machines can be defined in FOST in, say, 10^20 symbols. Then we can easily define BB(100) in roughly the same number of symbols (e.g. 10^21). It certainly is far beyond Graham number. And Graham number of symbols seems to be good enough to define a lot of useful things, like truth predicate for FOST, SOST and so on.

And now we iterate the function 98 more times.