I have a hypothesis. Implementing this policy won't change the flow of email.

I support this hypothesis as follows:

Most Stack Overflow users pass through three instars: larva, pupa, and moth.

Larval users have such low rep that a downvote feels like a real hit. This feeling is amplified by the initial shock of that first bit of negative feedback.

Pupae have enough rep that the impact of a downvote is no longer a significant hit to their overall rep. Further, in the process of gaining that rep, they've met enough negative comments and downvotes to have some thickness of skin. (Unless, of course, they've gotten their rep entirely by throwing softball questions.) However, they are still coveteous of their rep, and so downvote sparingly, if at all.

Moths have enough rep to spend it on downvotes, and are *really* not perturbed by them.

Thus, my claim: even polite, low-ratio downvoting will continue to generate plenty of complaining email. Because it is human nature to complain about the perceived unfairness of negative feedback. (And I bet that your email comes from a bimodal distribution: larvae and the constitutionally hypersensitive.)

On Meta, all the time, we write 'votes are on content, not people.' But the rep system guarantees that people will feel the sting. If you want happier users, then you need some way of getting bad content to sink without ruffling the feathers of its creators. This strikes me as a quest for a square circle.

If you are going to limit downvotes, I wish that you'd put a symmetrical limit on upvotes. Require more rep for them, or limit them. I mostly downvote to counteract the effects of what I see as ridiculous upvotes. I'm ready to leave something mediocre at '0', but I am provoked into spending rep when I see positive numbers on content of negative value.