No Qui-gon, You Cannot Force a Roll of the Dice - How Random Always Wins

 Selection, in the context of evolution, is used in order to maintain the best fit individuals and therefore the best fit population. In theory, by leaning on selection, this would eventually mean that over generations there each individual would simply be a clone of the other: perfect eyes for the environment, perfect height, perfect whatever. However, so much more goes into it than that (else everything would just strange clones of each other...or weirder).

First off, we have to think about the consequences of working solely toward selection, namely in the form of inbreeding. Inbreeding could lead to the "perfect solution:" you know exactly what you're getting, exactly how the phenotype will express, exactly how good of a chance your offspring have of surviving when reaching maturity. There seems to be no downside to knowing the outcome and not taking risks. However, by only breeding inward there's no way for the deleterious genes that a) already existed and were hiding in recessive states or b) suddenly appeared due to mutation to be shuffled out of the population. Eventually these deleterious genes will overtake the population and reproduction and fitness will become lower and lower. Thus, ultimately a bad move overall for the evolution of a species. 

Going off of that, too much of a good thing will ultimately always become a bad thing. Recently, when looking at the cave bugs, the breeder's equation calculated that eye size would increase in a linear fashion, becoming larger and larger over each generation. The trait is beneficial to fitness and there appeared to be a selection for it. However, if the eye were allowed to just keep growing and growing and growing and....

then we would just get very tiny eyeball monsters. 

Just eyeballs. 

But seriously, at some point the size of the eyes, no matter how good it allows the bugs to see, would become a problem. It would be bigger than their body (which was not proportionate) or cause problems with mating, problems with predation or finding food, or even take up more energy and function that might be needed for other body parts...like maybe their tiny brain. All-in-all, too much of a good thing becomes a bad thing, so there needs to be some sort of mechanism in place to prevent "over-selection." Once too much has occurred, something like predators, other bodily requirements, or the outside environment step in to keep an even balance, possibly even changing what might be selected for. 

Something that cannot be prepared for at all is...well chance. Phenomena like mutation, genetic drift, bottlenecking, etc. are all completely random regarding the effect it has on the gene pool of a population. There is no preparation that will help and will most likely lead to some sort of change in what's being selected for. Along with this, at some point selection might not be the most influential factor in a population. Mutation rates can over come selection, possibly for deleterious alleles. Same thing for genetic drift. The environment itself is also prone to change, something that may or may not lead to selection changes depending on the changes involved.   

A number of things prevent selection from draining the genetic differences that are necessary to maintain within a population. The even  by taking out all outside forces (migration, random events, everything) and even preventing mutation, inbreeding and the recessive genes that it would bring forward act as a stop gap. When there's no variation, eventually there will be no population, or possibly even no species.