We can distinguish between two kinds of value-subjectivism. You can have explanatory value-subjectivism, which simply means that in explaining someone’s actions, you appeal to their evaluations, not yours — just as in explaining someone’s actions you appeal to their beliefs and not yours. If you see someone walking out on a bridge, and you know the bridge is unsafe and is likely to collapse, but they don’t know that, then in interpreting why they’re doing what they’re doing you shouldn’t attribute to them your belief that the bridge is unsafe if they don’t have that belief. If you try to explain their action by appealing to your belief that the bridge is unsafe, your explanation isn’t going to be any good.
So likewise, if you’re explaining their actions you also have to appeal to their values. Suppose that you hate vanilla ice cream, and you see someone trying to get some. What they’re doing would make no sense if you assumed that they share your value. Instead, your evaluation of their taste in ice cream doesn’t make any difference to explaining — whether they’re right or wrong to like vanilla ice cream, nevertheless the fact that they like it is what explains their going after it."