It should be noted upon mentioning curve-grading employees against one another that there's a big difference between being in the bottom 10% of a group and an employee being actively destructive to productivity and morale. Planning from the start to get rid of 10% of the workforce doesn't seem very smart. Aspiring to hire 100% good to great workers makes more sense. Getting rid of people with tendencies actually harmful to the group is good, but getting rid of a productive employee who happens to be 5% less productive than someone else is just silly.
While what I wrote above promotes evaluations for the most part, I wouldn't promote intentionally hiring 10% of a workforce for them to fail and be fired. I'd want the documentation available to support firing people who actually, for example, cause productive members of the team to want to leave or who introduce more work for others than they accomplish themselves. That's hopefully a rare person who doesn't slip past the recruiting phase very often, though.
Mostly evaluations should be about two positive outcomes. The first is finding the problems otherwise productive employees are having that management can address. The second is about rewarding the great employees who stand out from the good and maximizing their contributions.