Interesting talk. Usually when people are challenging conventional wisdom, it's not difficult for me to see where they're coming from. In this case, however, I find myself struggling to agree.
Yes, there are benefits to autonomy, mastery, and purpose... but c'mon, isn't that blatantly obvious? Is anyone actually surprised that employees are more engaged when they're working on their own projects?
Regarding the "candle problem" that Pink brings up, I'm not sure if easy-candle-problem vs hard-candle-problem is analogous to mundane-career vs challenging-career. The candle problem takes minutes or even seconds to solve: It's not difficult to see why those under pressure may jump into action prematurely, before considering multiple possibilities. The correct solution is non-obvious, so those who act the fastest actually perform the worst. In other words, the people with the higher incentives are actually working harder, but it's backfiring because this is a special case. I doubt employees at big companies run into this issue when offered incentives, as their projects take months/years to complete, not seconds.
I haven't seen anything to suggest that incentives are ineffective. The "old" wisdom isn't wrong: people are willing to work harder for higher returns. If anything, incentives work too well. When poorly designed, they can actually go against the best interests of a company. If you tell ProgrammerJoe he'll get a raise for working quickly, then he's going to write craptastic code as fast as possible. If you tell him he'll get a raise for every new feature, then he's going to create a barrage of unnecessary and poorly-designed featured.
By all means, dangle carrots in front of your rabbits. Just make sure you're leading them in the right direction.