@Kiparsky thesis

All /pANini commentators teach that the three words vA , vibhASA and anyatarasyAm mean exactly the same thing: "you may use this or not, both ways are okay".

Now, if you think about it, that could not have been the case in /pANini's times. By the /ardhamAtrAlAghavena principle, if vA and anyatarasyAm had had the exact same meaning for, then sUtra ruhaHponyat... would be just ruhaHpovA .

It follows that these two words dids not have the exact same meaning for /pANini. And that the difference was forgotten later, earlier than the first commentary on pANini's work. What was the difference in meaning? We don't know.

Professor Kiparsky made this guess —

"/pANini uses vA for options he thinks preferable, anyatarasyAm for options he'd rather avoid, and vibhASA for indifferent options."

If that's right, then in rule vAzari /pANini is telling us —

"grammarians agree that both namaHzivaya and namazzivAya are okay, but I think that namazzivAya is better"

while with rule ruhaHponyat... he meant —

"grammarians agree that both rohayati and ropayati are okay, but I think that rohayati is better"

and in rules with vibhASA he had no preference.

It is not sure that this guess is right, but it's worth thinking about.