I'd say that sucky features don't lose you customers, they lose you time. My users have never (to my knowledge) quit in disgust over a poorly implemented feature, but they complain a lot about them and send me support requests if they're confused. This means I'm often re-explaining the same things or dealing with the same problems over and over. You spend your time either supporting or fixing it.
3 and 5 are corollaries, and probably the most useful points in the article. It's so easy to get caught up implementing customer requests, especially when they wave money at you. It's much harder to stick to your guns and/or distill their requests into feature sets that do what they actually want, rather than what they asked for.
I enjoyed this: "The fastest racecar can’t move if the gas-cap gets stuck; your product is only as good as its worst component." It reminds me of Dave McClure's "something in your product sucks, find it, kill it" (paraphrased).
Customers will always solve your problems poorly. If they didn't they would be in your business.