In morals, as in logic, you can't explain something by appealing to something else unless the chain terminates in an axiom.
The question "why is it bad to rape and murder?" can be rephrased as, "how can we determine if a thing is bad, in the case of rape and murder?"
The answer "rape and murder are bad by definition" may be unsatisfying, but at least it's a workable way: everything on the list is bad, everything else is not. But the answer "because they make others sad" assumes you can determine making others sad is bad. You substitute one question for another, and unless we keep asking why, we won't have answered the original question.
Okay, then interpret my answer as "rape and murder are bad because they make others sad, and making others sad is bad by definition".
Link to Blog Post: "Extremism in Thought Experiments is No Vice"
_____
_____
This is a LW discussion post for Yvain's blog posts at Slate Star Codex, as per tog's suggestion:
Scott/Yvain's permission to repost on LW was granted (from facebook):