If one is still using WinOS (and there are some of us who still do ) you can use ABW profiles and soft proof B/W images. The principal advantage of the ABW driver is that it gives you an increased DMax and uses lesser amounts of color inks than printing B/W via the normal driver. The advantage of using profiles is to improve the linearity of the B/W response from the printer. There is a lot in the LuLa archives and elsewhere on this. For those using MacOS it's a different story and unless one wants to do a fair amount of trial and error work on images, it's probably best to use the normal driver. That being said, one can prepare a profile in the normal way and include a good B/W stepwedge set of patches which should help smooth out the printer response (I use 51 patches when doing my normal ICC profiles for just this purpose). The only way of preparing ABW profiles is to use Roy Harrington's QTR system under WinOS.
I have not been able to compare results obtained printing with QTR profiles under ABW because I don't have access to that option. However, drawing on issues raised from the Ilford paper review, I have done further comparison work with the tonal settings in the ABW driver, and compared them with "Photoshop Manages Color" using the normal driver and Ilford's profile for GMS paper. Let us recall I am using Mac OSX 10.6.8 and an Epson 4900, with Epson driver version 8.64, the most recent for this combination of OSX and printer.
I made four prints of the Northlight Images B&W printer test target (1) Photoshop Manages with the GMS profile, (2) ABW-Normal; (3) ABW dark and (4) ABW Darker.
Firstly, contrary to what I thought, I discovered that every time the Epson driver is re-opened and ABW selected, the tone setting defaults to Darker. Hence the result we reported in the recent article on the Ilford paper is with the Darker setting (I hadn't noticed this nicety at the time). OK, so now we have results for all of them but Darkest, which wasn't necessary in terms of what I was looking for. I was interested in seeing differences of maximum black and tonal neutrality between the four settings, both visually and by the numbers. Outcomes are truly only moderately interesting - i.e. I find the differences subtle and the arguments for one or the other of these approaches somewhat academic. But for what it's worth, here goes:
For maximum black, L* is 4.9 for (1), 2.06 for (2), 2.05 for (3) and 2.93 for (4). I found this last result counter-intuitive so I repeated the reading and it was within 0.03 the same. Practically speaking however, when you look at these prints side by side close-up under a Solux bulb you see no difference in the blackness of maximum black.
For 50% gray, none of them reported 50% gray. (1) is 51.8, (2) is 55.9, (3) is 52.6 and (4) is 49.22. The really interesting observation from this result is that "50% gray" is considerably brighter in (2) than in (4), and the impact of this result is visible in the prints. I think it explains why the shadow detail is visually better reproduced in the ABW "normal" setting than in the ABW "Darker" setting once you emerge from the very lowest end of the tonal range in orderf to see any differentiation of anything under any setting. Between (1) and (2), I would give a slight edge to the deep shadow detail rendition of (2), and this is consistent with the numbers for L* in the gray readings (51.8 vs 55.9), but seeing the visual difference here requires a trained eye if you hadn't been pre-advised.
In respect of neutrality, none of them are truly neutral, a* and b* values never being zero, but some come pretty close. On the whole, not surprisingly, Photoshop Manages Color is the least neutral with a* and b* values of 0.54 and 1.17 respectively. It must be said - these are small differences, and it isn't until you look at this print compared to the others that the slight difference in hue becomes visible - this print is ever so slightly warmer to the eye, as the number confirm. All the results with the ABW driver are below zero, indicating a slightly cooler hue bias, again which only becomes noticeable compared with the warmer one. None of these results look intentionally "toned" when you see them in isolation. Print (2), ABW-Normal won the neutrality prize with a* and b* values of -0.26 and -0.77. The others were off by slightly larger absolute amounts, but the differences are trivial. hence the real winner from all four, if you like maximum shadow detail and neutrality is the ABW driver at the normal tone setting - I remind, for this combination of OS, printer and paper.
Then I compared what I saw on my display with each of the four prints, and given how my display is calibrated and profiled, (NEC PA271W, BasicColor 4, luminance at 110 cd/m3), the tonal alignment is closest to print (1) - Photoshop Manages Color, but also very close to # (2) - ABW Normal, because results (1) and (2) are really close for overall luminance above the very bottom of the tone scale. This coheres with Eric Myrvaagnes' finding #3 in his post above.
So going back to my first response to Dave, the OP, differences between the approaches are there but subtle, and much depends on the starting point you prefer, once you've run the tests. I would defer to you Alan on the numerical perspective for the linearity factor, as I cannot replicate your work on this point. I would say from a visual perspective however, that tonal representations between the display and the prints are very close in any of these results, save for what is going on at the very low end of tonal scale, as mentioned above.