>>195
I was thinking it might be possible to consider only the "exponential tower", since that can be "normalized" and that should get rid of most of the redundancies. But then it is possible that sqrt(2^6^8) is a "special" number and I would need to calculate at least some of its most significant digits, to see if it starts with 2681 or 268 followed by some digit and then an odd(?) amount of zeros.