I don't think there are any fundamental agreements here, just some really unclear debating.
With any bonus, you may have the nominal value, the risk-adjusted value, and the net expected value.
The nominal value, is, obviously, the face value on the bonus.
The risk-adjusted value (better term? I just made it up) only applies to stickies. This compensates for the fluctuations necessary to gamble it, and depends heavily on how it is played. The risk adjusted value of a cashable bonus is 100%. The risk adjusted value of a full bankroll coinflip is 50% of the nominal value.
The EV is, of course, the risk-adjusted value of the bonus, minus the house edge multiplied by the wagering requirement.
So, mick's (and snyder's, grosjean's, everyone's) point about stickies is that betting ballsy (quadrupling up in blackjack, single number roulette) will increase the risk-adjusted EV of a sticky bonus, up to a point approaching, but never quite reaching, the nominal value. The downside of this is a potentially dramatic increase in variance. (I play them fairly tight myself, not enough emotional fortitude)
But, you still have bonuses that have wagering requirements (sometime cashable, sometimes sticky). And if the wagering requirement times the house edge is greater than the risk-adjusted value of the bonus, then it's still a bad deal, right?
I think mick's example of two people playing roulette is a red herring. In solo bonus hunting, you'd play a bonus at one casino for one spin, then move on to the next casino for the next spin.