Suppose true belief is intrinsically good or at least that false belief is intrinsically bad.
It's the middle of the summer, and I agree to go with a friend for a hike. I always wear long pants, no matter the weather, except when swimming. My friend has a very good inductive argument that I will wear long pants, and thus forms the belief that I will. If I do wear long pants, then my friend has a true belief. If I wear shorts, then my friend has a false belief. Since I have reason to bestow goods and prevent bads to friends (actually, on everyone), it follows that my friend's prediction gives me reason to wear long pants. But somehow I find this counterintuitive. I feel a pull to say: The direction of fit in belief is world-to-mind, and while there is reason to make mind fit the world, there does not seem to be reason to make the world fit the mind.
Maybe you can get out of this puzzle by saying that true belief isn't valuable; only knowledge is. Fine. But if I do wear long pants, doesn't that bring about that my friend had knowledge that I would? After all, my friend had inductive evidence that in other cases (like observations of a deer's behavior) we would count as sufficient for knowledge.
Since I do have the intuition that true belief, and if not true belief then knowledge, is intrinsically valuable, I have to conclude that we do have reason to make the world fit the mind. In particular, this means that we have reason to act in predictable ways, and to do what people expect that we will do (even if they don't expect it of us to do it in the normatively charged sense of "expect"). In particular, this suggests that the distinction between promises and predictions cannot be drawn simply along direction-of-fit lines.