Actually, that's where you're wrong. Each phenotype confers a probability of survival upon its possessor and that makes natural selection random by definition. .
This is backwards. Natural Selection is the filter that strains phenotypic expression over time by favoring expressions that are more adaptive/fit. Phenotypes confer nothing but themselves. Selection is the engine of evolution. (Two competing metaphors in one para. sorry)
Thus Natural Selection is the opposite of random. There are elements WITHIN the process that are random, or undirected, such as random genetic mutation, but this doesn't make the process as a whole "random". A longer necked giraffe gets a few more leaves than his cousin, and thereby has more energy to pass on his genetic material, either by runnning faster or . . . whatever it is that giraffes do to get along. The neck length phenotype may have been a random mutation, but the process of survival and the effects of genetic success on species adaptation is highly directed, stochastic, non-deterministic.
I suppose if lions hunted by blindly swatting the air for giraffes then you could consider the process random . . .
When you go to Las Vegas and play roulette, the outcome of the the roulette-part of the process can be considered as a probability distribution. I think this is how you are using the word "random". The roulette can be thought of as "random mutation". If you happen to win, (fat chance) the outcome of the weekend will be different than if you lose. But what happens to you in Vegas, the booze, the hookers, the police, and how many criminals you may choose to enrich with your life savings, or if you survive long enough to pass your genetic material on to your spouse, (or they may not want to have anything more to do with you, once again, depending on the outcome at the Wheel), that's Darwin at work.
