I just finished viewing a 12+ hour training series on the process of printing. (http://www.luminous-landscape.com/videos/tutorials/camera_to_print_and_screen.shtml
) It was very comprehensive and I highly recommend it, even for experienced printers. The training is done by Jeff Schewe and Michael Reichmann. Jeff actually helped write the sharpening algorithms licensed for use in Lightroom 3 and 4. He was also the author of the original article that discussed the benefits of enlarging images in steps. It was because of this article that Adobe's Thomas Knoll reworked the resizing algorithms in Photoshop, making this process unnecessary. (If memory serves, this happened around the time the first CS Suite was introduced so it's been unnecessary for a while now.) Jeff is sponsored by EPSON and has been heavily involved in their printing workshops as well. Of course, Reichmann operates the popular Luminous Landscape website.
All of the topics raised in this thread were discussed in depth. I'll give the broad strokes of what was said.
1. Resizing in steps is no longer necessary as explained above.
2. Some input sharpening at 100% (at the original resolution) is always beneficial. It compensates for the softening that occurs in every digital image due to the demosaicing process. This is not to be confused with output sharpening which only happens at the time of printing or saving for display on screen.
The amount and type of output sharpening depends on the final use of the image. Will it be on screen? Will it be in print? If in print, on what type of surface? All of this affects output sharpening. If you are printing or exporting out of LR, then LR adjusts the sharpening for you to compensate for all these factors—one of the benefits of using LR to print or export.
If resizing in PS, sharpening along the way is also no longer necessary. The resizing algorithms take care of this as well. Three different algorithms (bicubic, bicubic sharper, bicubic smoother) allow you to vary the results to taste.
3. Printing at a preset DPI is completely unnecessary and resampling from, say 290 dpi or 310 dpi to get to a perfect 300 (or whatever) can actually degrade image quality (because of the resampling). HOWEVER, upsampling can be beneficial. Today's higher end printers can take advantage of high DPI files to print additional detail.
The old rule of thumb was that if the print was between 180-480 dpi then do nothing. Just send it to the printer. The new rule is that if the DPI of the final print is less than around 300, upsample it by 50%. If it is above 300, upsample to get to 720dpi (EPSON) or 600dpi (Canon, HP). The difference is subtle to be sure but it can be seen upon close inspection. Again, this can be done effortlessly when printing out of LR. (As an interesting side note, it used to be impossible to print out of LR at a DPI higher than 480. Schewe is the person who convinced Adobe's Eric Chan to get this limitation removed.)
4. The goal of these techniques is to maximize image quality regardless of viewing distance. At a reasonable viewing distance most of this stuff borders on imperceptible, particularly on matte art paper. However, it remains current best practice.
Going back to this training series, one of the things I enjoyed about it is the "inside baseball" talk that you get because Jeff is so closely involved with the engineers at Adobe that develop Photoshop and Lightroom. If you're interested in maximizing the quality of printed output (as well as minimizing the differences between screen and print) it is a great resource.