Well it's entirely true, so I don't know how you can call it a bullshit position.
It works like this; every generation has small genetic "mutations." I put them in inverted commas because they are so small that they are not having extra fingers or anything like that, it's a case of being slightly taller than your parents, or having a slightly shorter nose. Obviously some physical differences are based on environment rather than genetics (for example, the fact that Europeans are a lot taller now than they were 400 years ago is probably due to an improved diet, rather than a change in genetics). These changes are, at least to our current knowledge, completely random, we can't predict them at all. That's not to say that in the future we may have the technology to predict the changes in genetic structure over generations, but when you look at the differences between siblings (and there are obviously significant differences, as well as similarities) we can see that it's very hard to predict (of course the main reasons there are differences is because of the different active genes that each sibling has, all siblings have all the genetic information from both parents, but only half is active, and again which is active and which isn't is, as far as our current technology shows, random). The amount of variety falls along a standard distribution graph. Hence it is most likely that the changes will be very insignificant, and much less likely that they will be a larger change. All the changes either give the next generation a greater chance of survival, or a lower chance of survival. Again this falls along a standard natural distribution graph. We can see that each generation has an equal chance of being better or worse than it's parents. Natural selection is simply the name given to the fact that the generations that are more likely to survive slowly become more dominate, while the generations that are less likely to survive become less dominate. It is completely non random, given any amount of life forms of the same species, and an accurately known chance of survival for each generation, and enough time, we can easily predict which genes will survive and which ones will not.
Like most things, randomness only comes from not knowing the process that leads to the result. Theoretically, if we knew the exact situation that sperm was released, we could predict which sperm would reach the egg first, however we don't know that, so we label it as random. Likewise, theoretically if we were given enough information, and had enough knowledge, maybe we could predict what genetic information an offspring would contain, however since we do not have the knowledge or the information, we label it as random. Natural selection, however, is clearly not random, but very predictable, providing we know the variables. Perhaps we cannot predict what life will be like in 1 billion years, but that's not because we can't follow natural selection, it's because we can't predict what the environment will be like at that time, so we don't know what natural selection will be like. On the other hand, last year scientists put a predator lizard with a group of smaller lizards, and easily predicted the evolution that they witnessed (natural selection led to the prey lizards becoming faster and faster runners over each generation; there is essentially a 50% chance that each sibling will be a faster runner than its parents and the faster runners have a much higher chance of survival). So natural selection, when we know the information, is not random at all.