That makes sense. Philosophically, I like the idea of device independence in the monitor calibration, with dependence on the particular printer/paper combination taken care of by the soft proof mechanism. But apparently you do not find this sufficiently accurate?
The problem with paper white simulation when softproofing is that all palettes or tools or ANYTHING on your monitor that is pure white will seriously affect your perception of the softproof (the softproof will always look "colored" if there is anything "whiter than white" on your screen). This
alone is reason enough to adjust your display to visually match paper white (either way which approach you are choosing to start with).
The basic concept is really striking: adjust your monitor to look exactly as your paper ... and there will be nothing that is irritating or distracting.
Once you are there the only remaining issue is the adaption to the blackpoint. When you are working with a paper with a very low contrast you won't turn down the luminance too much (doesn't make sense anyway)... you have to lift the blackpoint. And it takes some time and also some test-printing to get used to a "black" that you initially would at best accept as a "grey".
What Andrew is talking about is a
perfect match of monitor and print. And if this is what you are really aiming at you have to tweak your monitor for each paper individually. There is no better way. And it requires some work and effort.
But if you can accept small (sometimes maybe only minute) differences... than you can just as well work with an "averaged" monitor calibration: Adjust your monitor to match a "reference" paper (coated paper used in quaility catalogues for example... or simply your favorite photo-paper) and "estimate" how much "warmer" or "cooler" a certain other paper is. If you do so softproofing without "paper white simulation" (but with "black ink" simulation) is the easier (more fail-safe) way in any case.