The topic of lying and deceit (if one can separate the two) is one that has been in my mind since Alexander Pruss blogged it about it a few months ago (see this Google search). Pruss has some good things to say, but I think he has neglected some fundamental aspects of the subject.
First, let me say that I disagree with Pruss that “There is good reason to think lying is always wrong.” Well, I disagree with the caveat that he and I may be talking about two different things (that is, he may be construing the term more narrowly than I). For the purposes here, I will use lying to refer to the act of expressing (generally verbally but perhaps even through non-verbal means) a statement or sentiment that the speaker either knows is false or knows she does not have reason to assert as true. (I am incorporating non-verbal communication here because they very often have semantic content that is translated by the other party; a shrug in response to “Do you know what time it is?” would be a lie if the non-verbal party had in fact just glanced at a timepiece he knew to be accurate.) This is to be contrasted slightly with deceit, which is a lie told with a specific goal in mind. These seem like linguistically appropriate terms, although I freely admit that they may have weaknesses in usage.
One reason I wish to separate these two concepts is that, pace Pruss, lying does not seem to be entirely unjustified on Kantian grounds, although I would argue that deceit is. Indeed, I think there are counterexamples where lying can be useful (in a sense that does not treat the individual as a means to an end) or even a consensual act between individuals.
The most obvious example of a consensual lie for me is fiction. When I pick up a Faulker novel (say, The Sound and the Fury), I am distinctly aware of the fact that Yoknapatawpha County is not a real place, and neither are the characters in the novel. Moreover, I am reasonably certain that Faulkner knew that the Compsons are not real people and that the events described in the novel did not actually occur. By all rights, this ought to be called a lie, but one normally does not see fiction condemned as immoral.
A few points seem to bring out the distinction here: first, we cannot say that Faulkner intended the readers of The Sound and the Fury (or any of his books set in the fictional county) to think that Yoknapatawpha County is a real place (although of course Mississippi, the state it is purportedly in, does exist) or that any of the events are true. There is an expectation that the work as a whole is not true, and the reader’s consent justifies this. But consent is not sufficient to ward off the criticism of using the reader as a means to an end; this is clearly not true of fiction, which generally has other purposes that are of benefit to the reader (entertainment, self-awareness, positive emotional instigation, etc.).
My main thesis here is that deceit has an element of subversion: that is, lying is bad when it undercuts or subverts something generally accepted as positive. Fiction has a positive end in itself, and we do not generally consider “reading only things that are true” to be something worthwhile to subvert. On the other hand, seemingly harmless lies (like telling someone you’re doing fine when in fact you are distressed) can subvert ordinary conversation, and this subversion usually violates the Kantian considerations Pruss mentions. In general, the cliché that honesty is the best policy may hold, but it does not appear to be universally so.