> I suggest:
> Level 1: long year (no precision).
> Level 2: long year with precision. Both using "Y".
> Anyone see a problem with this change?
That seems a good decision.
> an atomic clock. This is not compatible with the Gregorian calendar. Thus it is invalid to state the date of the big bang in Extended Date/Time Format because the number of rotations of the Earth since the Big Bang is undefined; the Earth did not form until long after the big bang.
True! This reminds me of the duration calculations within XML, where "time duration" has two different units: second (and its multiples such as hours) and month (and its multiples such as years). The problem within XML is not exactly the same of course, but measuring time in seconds (and its multiples) makes sense as a different unit than (Gregorian) years.
Indeed, the big bang is rather a use case for the need for another set of time notations out of reach for the (proleptic) Gregorian calendar and thus NOT covered by today's EDTF. Different relativity theories make things even more complicated. Therefore I suggest to write in the specification that the big bang is a use case for a future EDTF.next.
As of today, the most accurate notation for the big bang needs an open interval.