I completely agree about the always pursuing funding part. I was just computing what it costs to support my average sized group of 6 supported graduate students (I have more that have their own support through their workplace, but my policy is that my graduate students don't pay their own tuition and have a reasonable income to live on.) It boils down to about $350K/year of overhead bearing money (which is fantastically cheap since I teach at a public university; my private school colleagues need at least 50% more than that). Considering most NSF grants are in the $200K-$300K over 3 years range, I have to get approximately 1.5 such grants per year. The average success rate for NSF grants is betweeen 5% and 15%, let's say 10% on average. Thus, assuming an average success rate, a professor needs to write approximately 15 proposals to support such a group. Assuming the average success rate, I've computed that I would get the same expected rate of return if I simply consulted and donated the consulting money back to myself. I would also avoid the reporting headaches that come with a lot of money these days.

I spend approximately a solid month on a serious proposal, assuming I'm writing it by myself (in fact, I should be working on one right now, taking a break). Most successful proposals include significant results that demonstrate that the proposed research works. In fact, I have heard that many proposers have already done the work being proposed and are proposing to prefetch money for the next idea. Most new proposals do require a reasonable idea that has at least some novelty, which makes writing the next proposal difficult. Also, once you submit a proposal, you aren't supposed to submit the same idea somewhere else, building in some dependencies that are difficult to get around.