Of course this depends on how the supply demand curve looks like for their specific business.
An angry bird informed me about this :)
Though in this case, it's likely that Twitter has repeatedly run internal studies about what a paid-tier would look like, and each time concluded exactly what the GP is passing on: that it would be a net loss because the most-valuable-to-advertiser accounts would be the first ones to pay.
Even just psychologically, it would be a major blow to their sales pitch, going from "pay $X to put your name in front of the top movers and shakers in the world" to "pay $Y to put your name in front of all the Twitter users too cheap to cough up $8/mo to not have to see you." Not hard to see that $Y is going to be less than $X... probably a lot less.
I think another factor is the paradox that the user who has enough money to be willing to pay to not see another ad is exactly the most valuable user to advertisers— they've pre-qualified themselves on multiple axes.
So there is always going to be an enormous moral hazard here with the ad-driven-freemium model— always a temptation to run an occasional high-value ad to the paying users, or to disguise the ad as a "recommendation" or position it instead as a "sponsorship", or whatever else. Other than users reacting swiftly and decisively against this kind of thing, it's hard to know how else to keep companies away from it.