Evolution as design?

Ars Technica reports on February proceedings at the American Association for the Advancement of Science, on the topic "Communicating Science in a Religious America". http://arstechnica.com/articles/culture/communicating-science-polarized-...

An interesting bit from Ken Miller, Brown University cellular biologist, on combating Intelligent Design. Specifically, he addresses how to counter the doubt over Darwinian evolution created by ID's tactic of portraying it as an improbable "chain of accidents".

The solution to this, in Miller's view, is to describe evolution as a design process. Evolution is not an accident or a mistake, but a predictable property of nature, in which the process of exploring fitness space necessarily designs adaptations. Miller feels that the human body is designed; it's just designed by evolution using a fish as its raw material, an argument he borrowed from a recent book, Your Inner Fish by the discoverer of Tiktaalik.

At first glance, that view of evolution as design looks good to me (though I wonder what counter-arguments other biologists might offer). It also seems to me that we could point to somewhat similar examples elsewhere in nature, in case doubters claim you can't have a design process without an intelligence.

Here's what comes first to my mind: Look at a river bed full of smooth, round rocks. Did they begin as round stones? Of course not; something made them that way. An Intelligent Designer? Uh, no. Okay, then were they made round by the random chipping and wearing of their edges over the eons? Well, yes, but not quite "random": the process of water and grit flowing over rocks produces the predictable property of roundness.

I think any IDist will have to agree that a whole river bed of smooth rocks is not pure chance. (Feel free to ask him: "What are the chances of this one perfectly smooth rock forming? Now what are the chances of a million forming, all in one place?") And he'll also have to agree (we hope) that the Sky Father was not reaching down and hand-smoothing each rock.

Clearly, the smoothing of each rock is a "random" process of tumbles and scrapes and abrading, yet the smoothing is a predictable property of nature.

That's just my layman self talking, though. Is the above an apt example for better understanding Miller's statement about evolution as a design process? Or is it too far removed to be of use?