"The proportion of nefarious types (relative to the total user-base) routinely using Tor is going to be significantly higher then the proportion of such who are not..."
He is claiming that nefarious usage is significantly higher than non-nefarious usage. That's the exact meaning of majority.
That is to say that your odds of finding a nefarious packet is greater on Tor than on the broader internet.
That is certainly the view of the security services: anyone using encryption we can't break is a potential enemy hiding something we want to know.
That is why there were export restrictions on encryption technology until it became obvious that was detrimental (crypto developed elsewhere wan't covered, so the restrictions put allied commercial users at a disadvantage to other countries in the industrial espionage stakes without actually affecting the people the restrictions were aimed at at all).
It is also why services that offer encryption are pressured into giving authorities access to the private keys, and those that refuse (or can't due to the design meaning the keys are only in their users hands) tend to get shut down.
I don't know about exit nodes, but at least for hidden services that seems to be true (assuming you count torrents and drugs as 'nefarious'). See http://www.verisigninc.com/assets/labs/Measuring-the-Leakage... / http://www.ieee-security.org/TC/SP2013/papers/4977a080.pdf / http://donncha.is/2013/05/trawling-tor-hidden-services/
Doing anything that the NSA do not like, is "cool"; even more so in the global scene.
Seriously, if you want to get people to use Tor, all you have to say is that the NSA do not like it.
I may be stretching here, but I think most of us would probably agree that monitoring or outright shutting down Tor would be OK if the nefarious usage was 100%. More interesting is, at which point do we consider monitoring morally wrong as that percentage approaches zero?