"except that you can not demonstrate empirically, the mechanism by which something becomes more complex. Did the one photosensitive cell WANT to become an eye?"

There's no need for any 'wanting'. It's a simple equation: modifications (mutations, if you like) occur. There is a very small, but positive, chance that the change will be beneficial. There is a much higher probability that it will be detrimental.

In the case of the eye, I'd say it's sort of obvious that, if you have already got one sensitive cell, having two or more would be an advantage- you have greater sensitivity. Having the sensitive cells recessed slightly is an advantage, as it adds directionality. Further recessing will lead (unsought-for) to the pinhole camera effect coming into play- the organism can now sort-of see shapes, and distinguish between, say, food and not- food (a predator?).

And all these 'eye' types are actually found in nature. Plus of course, the line that's developed a protective (transparent) skin over the pinhole- thickening that just 'happens' to be in the right place and you've got the start of the lens. And all these types are also found in nature.

The 'benefit' or 'detriment' is tested by competition. If the change is detrimental, it will tend to be selected out (NOTE: not immediately and magically zapped). If beneficial, it will tend to remain around, and according to the benefit accruing, will propogate through the population simply because its possessors have an advantage.

Note that the mutation occurs randomly, by chance (on the basis of the possible mutations of the existing genetic material). But the selection is anything but chance- it depends solely on the competitive advantage.

It's obviously difficult to 'prove' this experimentally on living organisms, we haven't got even a few hundred years in which to make observations. But computer simulations have been carried out in which, by following a few simple rules as outlined above, it can be shown that UNPLANNED complexity can grow from a simple starting point.

Someone else pointed out that the fossil record seems to show periods of stability, with relatively rapid speciation between. This model, much supported by the late Stephen Jay Gould, is quite understandable, if you view the periods of change as environmental changes, forcing organisms well adapted to the conditions previously, to test out the possibilities of their gene pool against the new conditions.