that giving money to SIAI was the right thing to do.
Warning alarms ringing in my head. Phrasing it instead as “one potentially good thing to do” might be less psychologically appealing but it’s also a shit ton more accurate, and I totally follow Eliezer in emphasizing that our comparative advantage is in relentlessly seeking the truth, not in being “clever” with our self-deception. See Rain’s comment in this thread. Donating to SingInst is probably a better call than buying shiny new boots, but the rabbit hole is ever deep and full of surprises.
That may be true for you, or even for most people on LW. In my case, though, explicit conscious reasoning about how to maximize utility (1) underperformed taking advice from people smarter than myself, and (2) failed to result in useful action.
I know, but I feel like maybe we’re missing a third option. Like, if we further bootstrapped our explicit conscious reasoning then we wouldn’t have to rely on cheap tricks. It is impossible to go too meta.
Warning alarms ringing in my head. Phrasing it instead as “one potentially good thing to do” might be less psychologically appealing but it’s also a shit ton more accurate, and I totally follow Eliezer in emphasizing that our comparative advantage is in relentlessly seeking the truth, not in being “clever” with our self-deception. See Rain’s comment in this thread. Donating to SingInst is probably a better call than buying shiny new boots, but the rabbit hole is ever deep and full of surprises.
That may be true for you, or even for most people on LW. In my case, though, explicit conscious reasoning about how to maximize utility (1) underperformed taking advice from people smarter than myself, and (2) failed to result in useful action.
I know, but I feel like maybe we’re missing a third option. Like, if we further bootstrapped our explicit conscious reasoning then we wouldn’t have to rely on cheap tricks. It is impossible to go too meta.
Let me know if you figure out how to do that. In the meantime, I’m going to continue using the least bad known option.