If people get entrenched in long-term comment exchanges then they have less need to keep a-checkin' the publication's new output for sustenance (and so won't be looking at what the editors would prefer them to look at).
An ad eyeball is an ad eyeball. I'd need to see some very solid metrics showing that eyeballs on old pages aren't as good as new.
Say we limit this to juicy "native advertising". Then you actually have to read new articles to be advertised to because ads come your way dressed as articles.