Hume's Law

I'm working on a paper about Hume's Law, the principle that "no 'is' from an 'ought'." I think there's a lot to be said for it -- I think it was an important insight, and that a proper understanding of it will yield important insights into normativity. I also think I've never seen a decent formulation of it that wasn't false. Here's the gist -- the intuitive part -- the part I think is *right*: suppose I want to prove a moral statement. (We're working with an intuitive definition of "moral statement", here -- "Murder is wrong," or "Murder is right," or "Freddie ought to keep her promise," etc. are all pretty clearly examples of moral statements.) Hume says I'm doing something wrong if I think I can prove a moral statement by logic without relying on some moral premise. Example: suppose I try to run the following argument:

1. If I go to bed now, I wonÂ’t finish my project before lunch tomorrow.
2. I promised Emily that I would finish my project before lunch tomorrow.
3. Therefore, I ought not to go to bed now.

That argument isn't valid -- (3) doesn't follow from (1) and (2). We're missing a "bridge" premise (maybe something like "I ought to always keep my promises"). There is a 'gap' between the moral and the non-moral, and logic isn't going to get us across it. Yesterday I made a comment on an interesting blog post by Jeremy Pierce which invoked Hume's Law. He said: One sexual act creates a permanent bond between these animals. That doesn't always happen with humans, but I'm wondering whether this is scientific evidence that it should. I replied: It's not obvious to me that there's ever (or that it's ever conceptually possible for there to be) scientific evidence establishing any should-conclusion. (This is roughly Hume's Law.) He responded (I'm quoting excerpts here, apologies if I miss important context): I think Hume's perspective on this issue is actually pretty silly once you think about it. ... The problem is that Hume says there's never a fact that can give rise to a moral truth. Utilitarianism has facts about what creates happiness. Social contract theories will have facts about what rational people would agree to. Ethical egoists have facts about what's in your personal self-interest. All these theories then say that those facts determine what's morally right and wrong. But I think Jeremy's missing the point here. Of course he's right that a utilitarian (take, for example, me) believes that the facts of the matter determine its ethical properties. But utilitarianism -- or any other normative moral theory -- is a substantive moral claim. No facts, by themselves, can demonstrate something to be right or wrong. They only can once we add in, as a premise, our moral theory. It's just like when we added in the premise about how we ought to keep our promises in the example above. So I think there's something very importantly right about Hume's Law. But it's notoriously difficult to formalize just what it is. Here's a first take: "No valid argument has only non-ethical premises and an ethical conclusion." But here are several counterexamples:

• Everything Jeremy says is true. Jeremy says murder is wrong. Therefore, murder is wrong.
• Dan is a vegetarian. Therefore, either Dan is a vegetarian or we ought to condemn him for hypocrisy.
• Fred cannot save Jordie. Therefore, Fred does not have an obligation to save Jordie.

My project in my paper is to sort through these, and identify in what sense exactly Hume's Law is correct. I really believe there is something interestingly true about it. Thoughts?