Rebol is a battle on many fronts, and I worry that if we get too sidetracked by syntax modifications like this we risk missing the "main event".
** may not be ideal for exponentiation but it works. Please use one's imagination regarding that looking like a hat., perhaps in the negative space between the asterisks.
The problem isn't exponentiation, it is when I need to insert a lot of markup in a valid Rebol file: think if someone is accustomed to read things as int_0^1 x^2 dx (TeX) or this is a ^sup^ (AsciiDoc). And think if someone needs copy-and-paste from TeX or AsciiDoc into Rebol. And think if that someone is me!
Even if ^ were allowed as a symbol, we would not allow 1^2 or similar, without spacing.
Rebol is simply too lexically saturated for it to be realistic to paste other formats verbatim in the middle and have them work. Whether that be HTML, TeX, AsciiDoc, or otherwise.
There has been some work in Red on a "pre-load hook". I don't know how successful this has been. But if one has some "wacky" ideas for new notations, I'd prefer it be done with preprocessing rather than getting invasive into the C code at this time. I'd suggest looking into what they've done with this, and how it has been working out for them.
Well, most things we can paste between { } unless there are mismatched curlies, and the issue is whether it's possible to allow the escape character to be changed.
It's too much work for this current effort to reach parity, but perhaps it could be looked at for the next big leap forward.