SUMMARY: When processing my first week of shots with a new 24-105/L lens on my Canon 30D, I found the color and brightness of the images to be superb, but noticed that it required a lot of unsharp masking in PS to generate a sharp image. This led me to do newspaper-on-the-wall tests, as detailed here.
I then got a replacement lens from B&H Photo; newspaper and red bowl shots, detailed below, showed the new lens to be more sharp than my original, compared to a "control" 24-85 lens. But far more importantly, my field usage of the replacement lens on a trip to Japan has shown it to be phenomenal and very sharp, with little or no unsharp mask necessary:
I believe the extensive field results are a better test than the newspaper studies, for there are possible artifacts in the latter which I discuss below.
DETAILS: In the galleries here, the 24-105/L was tested side-by-side with a 24-85mm Canon USM lens. All was done on a tripod with a remote release and the IS off. I tried shooting a newspaper on the wall as well as shots of various items around the house. Note that for the newspaper shots, the corner images look a bit softer than the central ones, but this could be an artifact of the assay. It is difficult to be sure that the camera is exactly perpendicular to the wall. Since the focus point was in the center, any variation in camera angle from 90 degrees will result in the corners being out of the original plane of focus. The effect will be more pronounced at wide apertures.
Verdict: In the newspaper-on-the-wall assay, the replacement 24-105/L lens is at least as sharp as the 24-85, and in some settings definitely sharper, making it much sharper than the original lens. In shooting a red bowl, the new lens now focused much better than the original one and gives smoother background and more detail than the 24-85 lens.
Moral of the story: Canon's quality control leaves something to be desired, but if you persist, you can get a very good lens.