Evaluating the feasibility of SI's plan — LessWrong