Let me just repeat the comment I left there: Given the pointless discussion user 'humble' continues posting more and more comments as new answers, I vote to close as "no longer relevant".
(It keeps bumping the question on the top without any good reason, really. I think the diagnosis of quid and Ben Green is very accurate, and in any case the answer to the original question given by quid is surely sufficient.)
Right. Quid's answer suffices. In fact, my understanding was that numerical evidence supported Granville's constant coefficient fairly well, and tends to discredit a coefficient of 1. Not sure where I saw that, but it is just computer calculations anyway.
So I think the original question was a bit wrongheaded in the first place. Also voted to close.