I just fell across some confused remarks from KBJ:
One can value human life intrinsically and nonabsolutely. That is, one can value it for its own sake (because of the kind of thing it is), but be willing to trade it for some greater good. Since valuation is a matter of degree, one can value human life very much without valuing it absolutely. For example, I might be unwilling to kill an innocent human being in order to save 10 innocent human beings but willing to kill an innocent human being in order to save 1,000 or 1,000,000 innocent human beings. Being willing to sacrifice an innocent human being does not make my valuation of innocent human beings extrinsic; it makes it nonabsolute.
In the example discussed, the value of human life is not being outweighed by some other value. Rather, it is being outweighed by a greater quantity of the same value, namely, the welfare of innocent people. For a crass analogy: if an investor is willing to spend some money now in order to earn more tomorrow, this plainly is no indication of his placing less weight on the value of money. Indeed, such behaviour is precisely what we would expect from one who places supreme value on money. Likewise, a willingness for utilitarian exchanges like those described in the quote is precisely what one would expect to see in someone who places supreme value on human life.
Insofar as deontologists are unwilling to make such exchanges, they show themselves to not value human life "absolutely".* On the contrary, they appear to barely value human life at all, since they care less about the million lives than they do about getting their hands dirty. It seems that what the deontologist really values absolutely is his own performance or avoidance of certain act-types (e.g. killing).
* = That is, if "absolutely" in this context means "above all other values", as suggested by KBJ's framing of it as a matter of "weight" rather than "structure". Or perhaps it means that lives have infinite value. But, at best, that would simply make the various scenarios incommensurable. It couldn't provide a basis for prefering the death of millions to the killing of one.
Alternatively, I wonder whether "absolute value" might involve a category error. The 'absolute' in moral absolutism typically refers to inviolable side-constraints on action. As explained in the linked post, these are not aimed at realizing some value; they are simply constraints on what any moral agent may do. I have doubts about whether this is a coherent picture, though, so I will stick with value-talk throughout this post.
It's not as if the deontologist even values the general prevention of evil acts like killing. He wouldn't get his hands dirty in order to prevent a million other killing-acts performed by others. No. His values are quirkier than that. It's only the value of his own actions that looms large in his moral evaluations. (Incidentally, this leads to standard deontology being collectively self-defeating, as Parfit has shown.)
As you might be able to tell, such priorities strike me as kind of moral fetishism. We should recognize that what matters is human welfare. Sanctimonious "moral absolutists" don't. They instead think that what matters is to keep their own hands clean, no matter the suffering of others. (Moderate deontologists, such as KBJ, are not so bad. They think their moral purity is more important than the death or suffering of a few others, but if the stakes were high enough they would eventually relent and commit terrible acts in order to prevent much worse.)
Having said all that, my indirect utilitarianism would probably lead me to largely agree with them in practice. It's just as a matter of theory that the values of non-consequentialists are warped.