- Moral values are real properties of objects
- Moral reasons are creations of moral agents
However, if reasons must be reasons for someone, then it's hard to see a difference between a world in which there were no real values and a world in which values were all created by agents. For facts must be taken by agents as reasons in order to be reasons; therefore, agents can take even non-factive claims as reasons. To continue the previous example, if it isn't actually raining, I could still take the (alleged, but false) "fact" that it is raining to be a reason to take my umbrella -- I've simply made a mistake about what the weather is. Why, then, should we take any reasons-claim as involving a real feature? It seems that reasons-claims about subjective features (I think, wrongly, that it's raining) are indistinguishable from reasons-claims about objective features (it actually is raining). Which would mean that we should eliminate objective features from our ontology. (Note that, although my interest is primarily in moral reasons, this rather disastrous result is not confined to morality. Anything that could be taken as a reason could be debunked through this argument.)
I think there might be a way out, though, if we consider the idea of function. Take a simple physical example. We individuate pens from other objects (paper, couches, puppies, etc.) on the basis of their function. That is, a pen is an object that is used to write in ink. If it doesn't write in ink, it may have been a pen, it may become a pen, but it is not a pen. If real moral values serve a similar function, and subjective moral values could not, then it would seem that we can accept both (1) and (2). That is, if real moral values serve this function, agents may have to take them as reasons in order for them to become reasons, but this does not debunk them -- it's a fairly simple "mentalizing" of a real feature of the world. (Indeed, we could probably do a similar sort of thing for other putatively objective sources of all kinds of reasons.)
The usual suggestion for the function of morality is social coordination. (This isn't a great candidate, as it doesn't pull morality off from law or courtesy, but I'll let that go for now as it's a seperate problem.) So, would social coordination be a function that only real values could serve and subjective values could not? My instinctive answer is "yes". If values are all subjective, then it would take a minor miracle to explain (a) the anthropological data showing that most people (the overwhelming majority of people, in fact) agree on fundamental values (e.g., charity is good, murder is bad) and (b), ignoring (a), how people who were creating their own values would ever come to the needed agreements amongst each other in order to coordinate social activity.
That is, by (a), most people do, in fact, agree on fundamental values. If everyone is creating their own values, though, then this agreement is truly strange. It is highly coincidental that people living in different times and regions of the world would come to agreement on fundamental values. Even if they had contact with each other (a highly questionable assumption), given the way people are, it's equally as plausible that there would be instinctive disagreement rather than immediate agreement. Yet, immediate agreement is what we actually see. If there are no real values, this cannot be accounted for.
So, that's one mark in the favour of (1) as nicely compatible with (2). The other comes from (b). Even if we suppose that there is some way to account for this agreement without theorizing the existence of real values, the fact of social coordination is difficult to account for on the basis of subjective values. (Strictly, (b) and (a) aren't that different from each other. Both arguments turn on the difficulty of explaining the necessary agreements in values.) If everyone is coming up with their own values, then, one would expect, these should diverge at some points, often quite widely. But, if this is so, why would people ever (i) actually get together in coordinated social activity and (ii) ever succeed in coordinated social activity? With regard to (i), for people to get together to coordinate socially, it seems that they have to be valuing the same outcomes -- else how can they share the same purpose in a collective activity? (Certainly, there might be some minimal cooperation of the "I'll do X for you, if you do Y for me" kind, but I'm talking about large-scale activities, such as building a house with a group of people, rather than one-on-one economic-style transactions.) With regard to (ii), the success of coordinated social activity seems to rely on shared purposes. If these purposes derive -- and, as reasons, they should -- from values, and the values are subjective creations of individual agents, then it would be astonishing if purposes were shared. Hence, it would be astonishing if coordinated social activities ever worked. But, they do.
So, it follows either that, since (2) is true, (1) must be true as well (for the contradictory leads to inexplicable circumstances); or, since inexplicable circumstances arise if we assume (2), compelling us to accept (1), we should reject (2). As (2) seems to express a conceptual truth, I think we must accept (1).