If the amount of noise is *very*small, and the desired amount of sharpening is small, the order of operations usually won't make much of a difference, but this is rarely the case. For more realistic scenarios, Hoogle is absolutely correct that sharpening should almost always be done after NR (noise reduction). If you are interested, there are good mathematical reasons for this and we could discuss them.
Paul's caveat is related to another aspect of sharpening: Sharpening is commonly divided into three types, and people skilled at this always do them in the following order:
1. Input sharpening - to correct for the various blurring processes associated with the input device, ie, the camera, the scanner, etc.
2. Artistic sharpening / blurring - usually used to selectively sharpen and/or blur various areas in the image to achieve the vision of the artist.
3. Output sharpening - to correct for various blurring processes associated with either the output device or preparation of the image for particular output devices or destinations (eg, down-rez'ing before posting on the web).
Probably 90% or more of all photographers do only step #3 or do a single operation to approximate a the three step routine. I suspect this is what you are asking about. This is always done last.
Paul sounds like he was talking about artistic sharpening, and this can be done almost anywhere in the workflow.
T