Say (in the extreme case) you are Einstein in the moment where he realized that E=mc², and you think so far ahead that you can imagine the nuclear bomb. You don’t know if it is possible to build, but you decide to keep your work secret. As the field of physics advances, other will come across the same insight. How can these isolated, silent individuals find the others without going public about it? Is there some kind of shelling point they can construct without knowing of each other?
Edit: Just to be clear and avoid confusing; I don’t have any such potentially dangerous insight. I was thinking in a more general way about AI policy. Advocating for an agreement that no ones creates an AI with property X, would just be an invitation to try building an AI with property X. Just like you wouldn’t read aloud a list with bad words in preschool. That got me thinking that there should be a way to check with others, before publishing.
They can’t. Literally any change in behavior is (a bit of) Bayesean evidence for their beliefs. More importantly, if they believe it’s that dangerous and should be left unknown for as long as possible, they probably should not try to find each other. Just burn your work and make it impossible to disclose (deceptively, by acquiring syphilis or something).
In more sane belief-sets, there are probably no discoveries that are both foreseeably dangerous and possible to delay by more than a few years. Someone in this situation should focus on making trusted people aware earlier than untrusted ones will discover it independently.
As far as I know, there is unfortunately no system for this. I think what people typically do is contact MIRI leadership, but I don’t know MIRI leadership to have particularly put silent people in touch with other silent people as a result.
Thank you.
The best (but still complicated) idea I have as a general solution (beside contacting MIRI) is to set up a website explicitly as a “shelling point for infohazard communication” and allow people to publish public keys and encrypted messages there. When you think you have an infohazard, you generate a key using a standardized method and your idea as seed. This would allow everyone with the same idea to publish messages that only they can read. E.g. Einstein would make a key from the string “Energy is mass times the speed of light squared.” and variations thereof (using different languages). And leave contact information as a message.
I don’t know if there is any decentralized and encrypted messenger protocol that would allow for that. With that, the website would only have to contain the instructions to avoid the legal consequences of hosting.
The necessary raw tools (zero knowledge proofs) may already exist, so it’s just up to someone intrepid entrepreneur and engineer(s) to productize it.