I see where you're coming from, but this is not a situation most of us will encounter. I would venture that most here are printing on inkjet-receptor-coated media
Well, yeah, like I said, the HP offering will be really helpful in a lot of situations. I reckon it will cover 80% of the printing needs out there, say in the "good enough" color range. Although I do believe that the following scenario may not be uncommon: printshops will try to jam some of that heavy weight colorcopier paper used for greeting cards through the HP and I'd be surprised if something useful comes out the other end.
On the upside, maybe one of the manufacturers will recognize this market gap of a heavy weight double sided paper type that does allow inkjet greeting cards.
40%M input = 40%M on paper? Sorry, I'm not a linearization expert, but this has been my assumption in the past. As the HP engineer said, "linear response in colorimetric density as a function of input digital counts."
Conventional densitometry doesn't apply to inkjet printing at all. That is the entire core of the problem. Conventional densitometry requires 2 important ingredients:
1. Definition of a small range of most absorbing frequencies that will be taken as the reference (in reality a filter function over the entire frequency range is used),
2. Definition of the maximum absorption. (maxD).
Now, initially you would think that the maximum absorption is simply found at 100% ink deposit. You would be surprised in how many cases this is not true, even on matte papers. Even if this is the case, how does the printer know that the ink is behaving correctly? The most absorption may well occur while the ink is running...
Maybe HP has an additional test pattern like line-pairs that allows them to test this.
But the definition of the filter function also isn't as easy as it may seem. The maximum primaries of inkjets usually have a very different contorted behavior relative to the intermediated steps than for example press prints. Compare a decent Euroscale profile to any inkjet profile. This contortion is 3D through the colorcube, so testing for any combination of chroma and lightness will not mean anything.
So when you have some kind of definition of 1 and 2, you can end up with a definition of "linear", but the chances are pretty good that there is no relation to any visual linearity, and, more importantly, having "linearised" the primaries is absolutely no guarantee that the combination colors will suddenly behave properly, or predictably. This contrary to press printing, which is why they have gotten away with dot-gain control as the primary "color management" parameter for so long in that arena...