The task is to rot13 the first column of letters so that they match the word after for; from there it's rather obvious that the top solutions perform the equivalent of a rot13 in one way or another. If we leave out everything we know that is unrelated to rot13 or, at least, changing text… what are we left with?
Let's break the command in smaller chunks:
(<C-V>)(G)(g?)(ZZ)
\1 is visual-block mode, nothing to do with rot13 or changing text.
\2 extends the visual-block selection to the last line of the buffer, again, nothing to do with rot13 or changing text.
\3 is… what?
\4 writes and closes the buffer, again, nothing to do with rot13 or changing text.
After all that, g? (the remaining chunk) is the only part of this command that we don't know and that could have something to do with changing text or, maybe, rot13. Given the high rank of the solution and the unbelievable terseness of g?, the probabilities that it does precisely what is asked, rot13, are very high.
Or is it?
To make sure that we are on the right track, let's ask Vim:
:h g?
to which Vim responds, in its usual voice:
Rot13 encode {motion} text.
No google, no internet. Only grey matter and Vim; the unbeatable duo!
