I agree that when AGI R&D starts to 2x or 5x due to AI automating much of the process, that’s when we need the slowdown/pause)
If you start stopping proliferation when you’re a year away from some runaway thing, then everyone has the tech that’s one year away from the thing. That makes it more impossible that no one will do the remaining research, compared to if the tech everyone has is 5 or 20 years away from the thing.
I think it’s a high bar due to the nearest unblocked strategy problem and alienness.
If you start stopping proliferation when you’re a year away from some runaway thing, then everyone has the tech that’s one year away from the thing. That makes it more impossible that no one will do the remaining research, compared to if the tech everyone has is 5 or 20 years away from the thing.