Optimal Strategies for Reducing Existential Risk — LessWrong