Being able to discard your own beliefs when they turn out to be wrong is like going to the gym. It's not easy, and not everybody can stick with it, and it hurts at first, but then you're stronger. It's especially critical to support claims with evidence when you're providing any kind of healthcare, because of the kinds of decisions your beliefs are affecting; not to do so is unethical. If you read this and you think I've missed evidence for acupuncture (in the form of peer-reviewed studies at the same level of rigor as new medications - an equal standard) please point me to it. I will be glad to update my belief, and in public at that.
You would think proponents of alternative medical treatments (both practitioners and patients) who don't have a solid evidence base would be in favor of obtaining more evidence - because that would move these treatments into the mainstream, and help a lot more people. Yet, for some reason, very often there is a strong resistance to doing this, even among self-described rationalist alternative practitioners. And the rest of us can't be blamed for being suspicious of why this might be. (That said, when alternative medicine practitioners actively generate more evidence and then recommend modifying or discarding practices as a result, they're doing it right, and assuming they find treatments that work, they're benefiting future patients.) But there are mounting studies showing that most alternative treatments are really placebos for most of their applications - and one of these is acupuncture.
There IS some evidence that acupuncture helps with some musculoskeletal issues, but it would be interesting to do a study of this outcome, comparing against amateur massage and increased exercise (I'd bet money on equivalence here.) Think about that for a second. For most of acupuncture's applications, it's the same as giving someone a sugar pill and saying "this will make your back feel better". And one of the curious arguments I've been hearing recently about this is that "if placebo makes you feel better, then it's okay".
That's a very, very slippery slope. For one thing, a person's comfort with the ethics of using placebo seems to depend entirely on whether that person likes the one who's providing the placebo. For example - if I start telling patients that if they buy my my magic stick, their back pain will improve - is that ethical? No? How is that different from the acupuncture evidence? (Don't be too quick to say it's because there's no evidence for my stick. You're right, there's no evidence either way. But there is active evidence of no effect for acupuncture!)
(Curiously, these conversations, even with rationalists who happen to like alternative treatments, become very like talking to religious folks. They ask you to prove negatives - "Have you looked everywhere for evidence of efficacy" - and apply different standards of evidence. Drug doesn't work in studies? It's a fake. [Fill in alternative treatment] doesn't work in studies? Western medicine can't evaluate this, or the establishment is against it, or just ignore it.)
Usually it seems people's "ickiness" with my magic stick placebo comes from two places: either 1) I don't actually believe in the magic stick; or 2) I'm the one who "made up" the magic stick. As for #1, let's say I don't believe in antibiotics, but I give them to someone based on the hospital's management standards for some infection. Does my personal disbelief in them affect how they work? (If you answered "yes", I didn't realize I was that powerful. I'm going to start personally disbelieving that Jack in the Box is bad for me.) As for #2, well, someone had to be the first person to make it up! It seems difficult to explain that it's weird and icky for me to make up a new placebo, but if someone did the same thing a thousand years ago and we've been blindly following them since then, somehow that becomes okay. Fine then, I'll start charging peopole to give them advice out of Bald's Leechbook. It's also worth asking what progress would look like in these fields. In a hundred years, based on new evidence, allopathic medicine will have more effective treatments and will certainly have discarded ones that we still haven't figured out are B.S. Will acupuncture have made similar progress?
The Kahun Gynecological Papyrus. It's 38 centuries old. The ancients used it, so our modern Western standards of "evidence" don't apply. As long as it makes you feel better, what does it matter? Now hold still while I fumigate your eyes with goose leg fat and you'll feel all better. Oh all of a sudden you have a problem with traditional placebos?
As an aside, if a branch of medicine spends more time explaining its "tradition" than its treatment outcomes, you should be suspicious. (There are many fields that do this; I'm not just picking on acupuncture here.) For example, would you really care about the ancient culture of the people that invented your HIV medicine? As a general heuristic, every minute they spend telling you about the founders or civilization it comes from is a point off its credibility. And yes, this absolutely applies to mainstream allopathic medicine. Whenever you challenge something with poor evidence but that is entrenched in practice, you're likely to hear about somebody at Hopkins or Mass General who did it, so it must be good, right? No dice pal, show me the money! See? Same standard!
If it seems like I get on a bit of a high horse about alternative medicine it's that, forgive me, I hold any kind of healthcare claims to a very high ethical standard. Part of that ethical standard is that the patient can give informed consent for their treatment, and if you're telling them things that aren't true (that's what a placebo is), you are absolutely destroying that core value of consent. So if I hurt a few people's feelings in defense of this value, I guess I'm okay with that.
To be clear again at the end: we should use those medical treatments for which there is evidence of efficacy and safety, whatever their origin ("alternative", traditional, a corporate laboratory, an academic center, etc.) - with an equal standard of evidence, whatever their origin.
 There is Class C evidence that cardiovascular exercise can also have benefits in depression equivalent to SSRIs like Prozac. There's even a conversion: 5 miles equals 10 mg. (Class C evidence means weight of expert opinion based on experience, rather than controlled studies.) This can and does influence practice. See? Same standards.
 And speaking of exercise improving musculoskeletal complaints, in a study of age- and weight-matched older runners vs. non-runners, the runners had fewer knee problems. So there, non-runners!