I think that's the idea -- you want miners to be as useful as construction waste aside from their utility (and return) from securing the network.  Else a potential attack becomes a recoup-able investment.  

Paul Sztorc did an extended analysis I vaguely recall; although that would seem inconsistent w/ his drivechain advocacy, so I might misremember.

elvismercury

bitcoin_Mining

Troy Cross' takeaways from the Heatpunk summit in Denver

k00b

I'm sure you recall, but there was a time when people thought they could change the work algorithm to something like genome sequencing. The main argument against it was that mining needs to be wasteful. IIRC this is also an argument against merge mining and drivechains.

For some reason, I assume the effect of positive externalities to be pretty marginal. I guess it could motivate a hastrate majority that might not otherwise exist? Seems impractical but maybe I'm thinking about it wrong.