This was a very thought provoking article. I had to laugh because we are thinking along the same lines: I am waiting for Canon to announce their upgrade of the current 5D, as I decided my next body would be a full frame and it's not as out of reach as some of the real pro equipment. I like the results a couple photographers were getting with it and it would work well with my current lenses. Meanwhile, a friend points out the Rebel xsi coming out and I'm thinking. . .whoa, I don't know when the 5D upgrade will be, at least until next October at the earliest. The xsi has over 12 megapixels. . .and a LOT of other bells and whistles that interest me. That's a lot of camera for the money, and if I get it next month. . .I can wait until the 5D replacement is over a year on the market and the initial price comes down or wait for a sale. Do you think the images from the xsi will be vastly better than the previous xt and xti Rebels?
Thanks for you opinion,
-Kim

Hello Kim, I do think the upgrade to the 5D (7D)? will inherently be capable of better (lower noise) images then the XSi, simply by means of physical reality. Larger pixel sites do create less noise.* However the gain is proportional to square root of area and thus not that significant. No doubt, the 7D will offer other great advances. Many people would consider the full frame sensor an advantage (focal length of lenses).
On the other hand, the 7D (5D upgrade) will have enormous requirements on the lenses, which only the most expensive Canon L Series glass will meet.
Because the image circle required is much larger for the larger sensor, the lenses need to have excellent quality all the way out to the edges.
Even a cheaper lens will have good sharpness around the center of the frame.
The XSi has a crop sensor. Imagine a Letter Format cropped out of a Tabloid Print. The result looks like a digital zoom or if someone just takes a 1Ds M3 (22 Megapixel) image and crops out 12 Megapixels.
At first glance, this is not a favorable thing to do. However, the smaller sensor is cheaper to manufacture and when you consider that you will be able to get away with standard lenses, you will either save tons of money or have money for other things.
As a travel photographer, I find it just as important to have money for travel expenses (flights are getting expensive).
If you decide to opt for the full frame models, you may have to budget in some expensive lenses as well.
I am thinking of getting an XSi now (for its 14-bit ADC resolution) and a 50D later (when the 40D upgrade arrives). I enjoy lightweight equipment and if I need very high resolution, I simply expose multiple frames and stitch them to Ultra High Resolution Mosaics.
As a pure landscape artist, the extra resolution and clean images may be worth the money and if you take images for advertisement or magazines (portraits), you are probably better served with the 7D. For portrait work, edge sharpness is not nearly as important as it is to the landscape photographer.
Interestingly, I think that the larger sensor is not going to do much for you in terms of the diffraction limits I was talking about above. Because a larger sensor requires you to stop down more on your lens to achieve the exact same Dept of Field then a camera with small sensor, you will hit the limits at about the same DOF (with a smaller aperture though). The benefit is on the other end of the scale, where you can create a more shallow Depth of Field. The laws of optics determine the limit in terms of resolution vs. Depth of Field.
Somehow I feel reminded of Heisenbergs uncertainty principle of quantum mechanics. Translated into photographic terms I could postulate: "The sharper I need my images to be, the less Depth of Field I can create and the more Depth of Field I need, the more sharpness I am going to lose". Of course thats just true at apertures smaller then a certain range. For my 20D thats about f/13 - f/16. It ties directly into a previous Tutorial about Hyperfocal Distance.

*As an engineer, I design microchips on CMOS (complementary metal oxide semiconductor) - the same material sensors are made of. If there is some interest, I can elaborate on that bold statement about the noise in great depth.

One can understand the reasons for the megapixel race. They're almost duplicating the race for the highest DPI scanner, sort of.
I find these kind of races, very boring. I would much rather have them improve on the sensor, especially better ISO sensitivity. Like the new Nikon D3 and Nikon .. please lower the price on that one!. And while they are at it could they also increase the dynamic range of the sensor

At what point do you think there will be little need for photographers to worry about pixel count? 8, 14 into the 20's and higher.... then digital backs around 39 and higher.... there must be a point where we can focus on plasticity rather than size and sharpness? Quality first, pixel count second?

I think we have to start looking at pixel density instead of Pixel Count. A digital back with 39 Megapixels will have much larger pixels than a small point and shoot with 8 Megapixels.
I have been working on an article about Diffraction that talks a bit more about the pixel size and why it matters. I will put it up shortly.
Quality is always more important then pixel count. An Enlargement from a D30 (3 Megapixel DSLR) will look better than a large print from a 8 Megapixel Point and Shoot.
The reason is that we rarely ever look at a very large picture with our nose almost on the canvas, but the cleaner noise free look of the D30 will look far superior from a short viewing distance away, which is more natural anyways.
BTW, I bought the 450D, but mainly for its cheap price and the 14-bit image depth. Not to mention the 12 Megapixels were important too but not the only argument for this camera.
I don't think there will ever be enough Megapixels for some people while others care more about quality of their work than technical details. Between these two extremes, everyone will find their place.