The summary is very careful to describe the lowest performing workers as: Every team has someone who at the bottom of its bell curve: an individual who has a hard time keeping up with other team members. Based on this definition, as you replace one, you will still have the lowest performer. Just your measure criteria will be higher. Thus unless you have a team that are all clones of each other, work politics will still find the "idiot" to hold.
Yes, obviously someone else will take over the bottom spot. But remember that a bell curve has an important parameter: variance. Your "all clones" team has a variance of zero. Replacing the low performer with someone closer to the mean will not only raise the mean, it will also lower the variance.
Thus, the measure still stands: How do you treat your lowest performers is a good judgement of the company health. Your preference of "Not worth the time/effort. Replace them with someone better." is quite destructive. By following your solution, the team will be forever stuck with overhead of training up the new guy and loss of knowledge of the rapid turnover.
First of all, what I outlined in my post is by no means meant to be a "solution", much less "mine". Second of all, your rapid turnover conclusion is based on flawed premises. If you replace the lowest performer with a new guy who ends up at the very top of the bell curve, and manage to do that repeatedly, that is unlikely in itself. You're far more likely to find someone who will slightly raise your mean performance, but mostly reduce its variance, making the "next idiot in line" less likely to be as much of a problem as the last one. Also, being "forever stuck", as you state it, would imply that there is some magical, unlimited source of "better people" allowing us to keep raising the performance ad infinitum. Which is obviously not the case.
The real message here is: A large variance in your team's performance is a problem. If your team/project happens to be successful, for whatever reason, you will be able to tolerate a larger variance. But I would argue that the best team is one that performs consistently well, i.e. one with a high mean and a low variance; and in general any team with a low variance will have a better chance of performing well than a team with a higher variance.