One is not quite sure what to make of Hard Truths, though it is certainly interesting (at least through chapter seven) and incisive. Certainly good: the points about engineering one's way to hard-and-fast lines (working over both the concepts and objects simultaneously). In fact Arthur Fine was just here talking about science studies and constructivist sociology of knowledge in ways that, it seems, would be up Millgram's alley; he does mention Foucault and (Arnold) Davidson somewhat early on to distinguish himself from them, but, though he calls on Latour in ch. 7, it's not really a theme of his discussion thus far.
Reading that use of Latour (applied to precisificationist approaches to vagueness) calls to mind a famous injunction from The Mythical Man-Month. A footnote of Millgram's:
Interestingly, Latour pins Aramis' ultimate failure on the unwillingness of those involved to violate a condition that Fine, 1975/1996, pp. 127, 129, takes to be a sine qua non of the precisiﬁcationist approach: that truth values remain stable under further precisiﬁcation. It was the engineers' and project managers' insistence on sticking with the deﬁning features of their vague ideal ('nominal Aramis') that made the political compromises necessary for Aramis’s survival impossible. The lesson Latour draws from Aramis's failure is that the workability of any realistically large project involving precisiﬁcation of this kind depends on one's ability to give up the truths ﬁxed by one's initial, still-very-vague description. (See pp. 48, 99–101, 108f., 119f., 281, 295.)
The injunction from Brooks being: "Where a new system concept or new technology is used, one has to build a system to throw away, for even the best planning is not so omniscient as to get it right the first time. Hence plan to throw one away; you will, anyhow." (One can say much the same thing about dissertations, or any other endeavor in which, in prosecuting the project, one is also learning its boundaries, how best to pursue it, etc.; Fine (Arthur, not Kit), in fact, seemed fond of quoting Dewey to the effect that we learn in our investigations how to investigate, and we needn't simply be making what we already more-or-less thought more precise.) Google reveals a corollary attested only at that precise URL to the effect that it has to be a sincere effort, too, you can't go in to your first attempt to hash things out with the project of making a toy that you can discard.
A slightly more substantive update made the day after the above was written: In chapter nine Millgram says this:
Fourth and finally, once you allow partial truth, you no longer have the option of treating truth as a primitive. When you characterize a claim as true enough, or true in a way, or almost entirely true … you need to be able to explain what you mean by that. These explanations, we have seen, proceed case-by-case, can themselves involve a great deal of subtlety and nuance, and as we are seeing, they are the occasion for a great deal of clarificatory theorizing. Whether or not full truth is what we understand the best, complacency is not an attitude we can reasonably adopt toward partial truth.
Since Millgram has previously referred to Aristotle for the claim (which hardly needs such a weighty authority to back it up) that the ways of missing the mark are many, but there's only one way to hit it, this bald assertion that allowing partial truth—which is, after all, often characterized as that banner under which the various fallings short of the mark are united—means denying the primitiveness of truth, at least, depending on what one means by "primitive" here. It can still be at least more fundamental than partial truth, which is we seem to understand primarily with reference to hard truth, the way that Aristotle proposes understanding failed or otherwise partial exercises of a capacity:
Which we may interpret (following Kern in Quellen des Wissens, which is the source of the remark a couple posts infra about barn facades and Megarians, which I plan on eventually returning to in some way) as the claim that reference to a rational capacity explains the successful exercise immediately, but must be supplemented, in order to explain a foundered exercise, by reference to some particular thing that got in the way (how to construe this in light of 1048a15–24 is not terribly clear to me in itself nor, and this is the point to which I may yet recur, in Kern's exposition), that particular thing only really being understandable precisely as something that is unfavorable for the exercise of the capacity.
[T]he same rational formula explains a thing and its privation, only not in the same way; and in a sense it applies to both, but in a sense it applies rather to the positive fact. … such sciences must deal with contraries, but with one in virtue of their own nature and with the other not in virtue of their nature; for the rational formula applies to one object in virtue of that object's nature, and to the other, in a sense, accidentally. For it is by denial and removal that it explains the contrary. (Metaphysics θ 1046b8–14)
This is, in fact, just the way one could take Millgram's example of factory seconds:
Because irregulars deviate in indefinitely many ways from the specification, there was no point in replacing the 'theory' with a taxonomy of defects; it would not have made sense for the Levi's outlet to have a shelf for the jeans with the nonstandard zippers, and another for the jeans sewn with off-color thread, etc. However, when you say that an item is irregular, which is tantamount to saying that its official specification is almost but not entirely true, you are not suggesting that the item does not exist. After all, you are in the factory outlet precisely because the irregulars are there on the shelves.
They're there, and they're Levi's, all right, but what they are in particular is imperfect Levi's, each imperfect in its own unique way; the proper account of them is "Levi's, but …". The proper account of the stuff that gets sold in the regular stores (factory firsts?) is just "Levi's"; that is, not "Levi's, not but (infinitely long disjunction goes here)"—the existence of partial jeans doesn't mean we can't take non-partial jeans as the basic case.
I just noticed that the immediately following section addresses "Naïve Action Theory" and the "strictly incredible" consequence of the view presented there that one is faced with an infinite regress of increasingly smaller intentional actions—a consequence that Thompson at least flirts with (calling it a "suspicion" and then a "conjecture"), that Rödl I suspect endorses, and that Lavin has argued for explicitly (though not, unfortunately, in print yet). I don't think it actually is a consequence of Thompson's relections, though, or at least, Thompson structures things in a way that makes that consequence seem unavoidable, but it is in fact avoidable; briefly, the sort of answers Thompson is prepared to accept to what we might call the inward-directed "how?" question (the "why?" question being "outward" in the sense of looking beyond the present action to some end or other action it subserves) is constrained from the outset to other answers, and I think he even considers this an advantage over the "why?" question where, it seems, we have to accept the not-quite-non-answers "I just feel like it" and "oh, no reason, really." But the consequence of this constraint is just the consequence Millgram correctly notes is incredible. Lavin has it that the constraint is necessary lest we fall into an objectionable metaphysics of action, but that isn't so; that is, the metaphysics he wants to avoid is objectionable, but we can jettison the constraint and still avoid it. Jettisoning the constraint allows us to accept the following as an answer to the "how?" question: "I just do, see, like this". (Anyone who wants to read approximately 17,000 words on this topic is in luck!)
Of course, even if we allow that an action can be intentional under a description that doesn't allow for its being redescribed partwise in such a way that the descriptions of the parts also give descriptions under which the part-proceedings are intentional in themselves, you might think, we're still stuck with the consequence of a regress in what is happening (and this I know Rödl accepts), which may also be incredible; it's less obviously incredible, at least, at least insofar as we stick with the natural attitude and consider how things are presented in experience. We can acknowledge that when it gets down to the unclefts we may be forced to think of things differently. (In fact this isn't unlike Millgram's response in the case of the consequence he does consider. One of the reasons I said above that "one is not quite sure what to make of Hard Truths") is that if one can recur to partial truth, one may do so prematurely; in this case, something like that has happened, I think. I mean: it is true that Thompson's picture as presented by Millgram is mostly right. But it can be made a lot more right if we examine it closely and remove one of the presuppositions, something that can be done with perfect justice. (I also, though this is a much more local complaint, don't really think that one has to, or should, construe Thompson's method as Davidsonian.).)