How much should we live up to other people's expectations?

I've had an especially interesting series of discussions with professors over the last few months around this topic as it relates to learning (and grading). This thread of conversation has been going on ever since a conversation with Gill (badly paraphrased) about ECS, way back when I was actually complaining about it in what was probably my sophomore year. I was griping about how ECS didn't have any metrics (yes, the dreaded m-word; I also questioned this!) which meant you never knew what you were supposed to do. "Well, " Gill explained, "we want you to stumble around and decide what you want to learn, and find out how to learn it." "But it's so inefficient!" I said. "Exactly."

Then this semester, I was talking to Ozgur about the vagueness of his comments during our design project reviews. "It's not enough feedback," I said. "We don't know what we did that was good or bad." "In the real world," Ozgur replied, "people might not tell you that. We give you lots of feedback in your earlier design classes*, and I wanted you to see how you would do without it. You need to decide for yourselves if something is good or bad." Adjust to feedback, but don't depend on it; it might not come. It's a variant of waiting for someone else to tell you what to do.

*which is open to debate, but I've found that folks in general are usually good about giving you really good, detailed critiques if you ask them in person later... I don't do this as often as I should.

If you're a teacher, how can you strike the balance between having your students do what you think is good and having them do what they think is good? Is it possible to imprint your own values too strongly onto them and prevent them from becoming their own person?

If you're a student, how can you strike the balance between doing what you think is good and doing what other people think is good? How can you tell whether what you want is what you want, or what you want because other people want you to do it?