And that's okay, if the goal is a culture war rather than a science discussion, because even in Darwin's time it wasn't all balloons and ponies for Natural Selection. It was years later that evolutionary biology got help from an understanding of genetics and Natural Selection became accepted after rigorous scientific investigation.
But if the discussion is a science one (and it is, else you wouldn't be here) you recognize that Evolution is a process, not a roadmap, and Natural Selection is just one mechanism in the process. Because it's just one process, things can happen even if they are not beneficial and, because Ma Nature sometimes has a sense of humor, things can even go to a weird place on occasion.
I'm talking about genetic drift, evolution's random walk, and it happens because each of us has two copies of each gene which can be the same ... or different. When we have a child, the little critter gets one member of each pair we have but which one from each parent is completely random. If you're the gambling type, it means nature is basically flipping a coin. Sure, if you have 100 children the distribution will be about even but we don't get to have 100 children these days - believe me, I asked and got a firm scientific answer of "impossible" from my wife - and the more common instance of a couple only having one child means there is an absolute certainty one form of each gene (allele) from each parent will be lost. Now do that for every child bearing couple and you can see where genetic drift gets interesting.
If you're not the gambling type and need a different description, genetic drift is evolution's equivalent of a sampling error. Somewhere between 1 child and infinite children in a population is a perfect example of how evolution can happen without benefiting from natural selection's input at all. Take that, Sir Charles!
The fine folks at University of California Berkeley did this excellent graph showing genetic drift using marbles.
Why does it matter? Because genetic drift can cause changes that have no use at all but that become permanent in a population. The smaller the population, the more impact genetic drift can have. Back to coin flips. Say you want an inherited gene to basically be 'tails' on the coin. If you flip heads 11 times there is only a 1 in 4096 chance it will be heads again - so tails is a good bet but heads can still happen because you still you have the same coin with two sides. But genetic drift, the random walk of nature, means you could eliminate tails completely much sooner. If the parents have one child and each parent has a heads and tails form of a gene, there is an equal chance the tails allele could be gone right away.
You can imagine that in a small population genetic drift could overrun natural selection but how do we prove it? Well, that's the debate and part of the terrific nature of science. The different shape of leaves that seem to serve no distinct purpose could be genetic drift but it could also be some form of natural selection that is imperceptible to us today.
The hard part is deciding what is truly genetic drift. We don't really know if something in evolution has no purpose; we have a lot left to learn, including the purpose of plenty of things. There may also be genetic drift in left over genes that serve no function today so it is unknown to us. What will really be interesting is a future where we can actually determine the 'combat' between natural selection, which essentially wants to improve fitness, and genetic drift, which doesn't give a darn about anything except how the dice land.
There are plenty of other mechanisms in evolution; too many for one article, but genetic drift is my favorite; a little randomness is what keeps the world interesting.
Can't get enough biology? Check out 30 Days of Evolution Blogging And Darwin Day 2009.