How SIAI could publish in mainstream cognitive science journals — LessWrong