I often sand to 1200 grit or more on projects with a rubbed oil finish. I dry sand to 400 - 600, and then wet sand through the higher grits. I think it makes a huge difference in the small details - for example, cross grain flecks in quarter-sawn cherry or maple. If you wet sand to that fine a grit, you are going to get more color variation because the natural variations in the wood are going to show more clearly. Cherry is (in)famous for that. If you stop at 220, the additional roughness in the surface over a 1000-2000 grit finish is going to mask some of the color variation. Personally, I like to take advantage of the color variation, and I try to choose wood carefully for grain and color to get what I want. Furthermore, it's not necessarily important to see all that detail in every piece, so I don't sand to that high a grit all the time.
If I'm putting on a hard finish, I sand to 400-600 and call it a day, because wet sanding a hard finish may affect the look of the finish, but it's not doing anything to the wood.
On a more philosophical note, there was an article a while back in Scientific American on the aesthetics of Jackson Pollock's paintings - the guy who splattered paint on the canvas. In a blind experiment, people were shown a Jackson Pollock painting and a respectable wannabee Jackson Pollock painting, and asked to choose which one they preferred. Regardless of whether liked Pollock's style or not, there was a very clear preference for Pollock over the wannabee - and this was true over a selection of Pollock and wannabee paintings. Since it seemed like the style didn't really lend itself to that kind of preference ("just" splattering paint on canvas, after all), the authors tried to analyse factors that might account for the preference. One of the things they analyzed was the fractal complexity of the paintings. They found that Pollock's paintings were an order of magnitude (i.e. over 10x) more complex, more detailed, than the wannabees. Their conclusion was the Pollock painting engaged the viewer's brain in more unconscious analytical pattern analysis, which translated into a more pleasing aesthetic at the conscious level (again, regardless of whether you really like Pollack to begin with). No one at the time the article was written knew what techniques Pollock used to get more detail - he worked behind closed doors.
Now think of that in terms of wood grain. As you sand through successively higher grits, you bring out more detail. The detail is patterned due to the way that wood grows, but it's not geometrically regular, so the fractal complexity metric would seem to fit. If you apply the conclusions of the Pollock article, the increased detail of the grain should result in a more pleasing piece even if the viewer is not really aware of the grain patterns. That's good -if- the focus of the piece is the wood. So that's the idea that I'm playing with by sanding to higher grits, and I think it plays out pretty well.
That's not saying that every piece will look better if it's sanded to higher grits. You just have to decide how to balance the detail of the wood vs. the lines of the design vs. the utility vs. whatever other design factors you have, and work appropriately.
