Readers of this blog might be a little surprised -- aren't I a mathy, nerdy kind of guy? Indeed, and this is precisely the reason why I consider the "all ethics can be derived from science" position to be so bizarre. After all, anyone familiar with abstract mathematics knows that our choice of axioms -- our set of starting assumptions -- is critical. A seemingly innocent assumption, such as the Axiom of Choice, can have wildly unintuitive consequences, like the Banach-Tarski paradox and a well-order of the real numbers. Under the classical set theory axioms, a basic question like the Continuum Hypothesis is provably irresolvable. You can't get anywhere in formal mathematics until you have a very clear set of starting axioms. You have to make some basic choices about the structure of mathematics before you can make any logical derivations, and these choices matter.
It's the same with morality. First, you have to set down some basic axioms, and this process will inevitably involve subjective human judgment. Perhaps you think you can circumvent this by calling upon evolutionary biology or some other science, but even then you're axiomatizing the assumption that evolutionary biology should matter in moral philosophy, which is hardly a core logical principle. After you choose your axioms for moral evaluation, of course, then you should use science and logic to draw normative inferences, but you can't claim that you "started" with basic logic. That just doesn't make any sense.