Is your time-discounting function a value, or a value parameter?

I agree, this is not a continuation - the argument in the other thread is indifferent to this, since if we ignore the possibility of later things being valued exponentially more, a polynomial discounting function from any source means that the "real" (context-independent) discounting function is polynomial (including 1).

As for "mathematically wrong," you should rather say "bad if preference reversals are inherently bad." "Inconsistent" means multiple things here, and the "mathematically wrong" inconsistent is not the same as the "preference reversal" inconsistent.. See this Wei Dai comment.