Now, I realize that language is like a living organism that changes and adapts, and that even a so-called dead language like Latin can adapt to form new words, but I am having a lot of trouble accepting the idea that the gender of the noun doesn't matter in passive periphrastics. The default should not be the feminine.
The immediate cause of my aggravation is a new "delenda est" applied to the noun "Islam" at Pacificus Inclined to Peace blog.
P. Gainsford has kindly written to confirm what I thought, which is that "Islam" should be treated as a neuter noun, so the passive periphrastic that would go with it (and without getting into the absurdity of this or any of the other propositions) is delendum est, NOT delenda est.
In his Laudator Temporis Acti, Michael Gilleland blogs references legitimating our use of the famous passive periphrastic ("Carthago delenda est") of Cato who thought that "Carthage must be destroyed." Gilleland says further that "Unio Europaea delenda est" (the E.U. must be destroyed -- to save Europe) is also fine and a good effort at applying the rules about the gerundive and the "to be" verb.
Carthago delenda est