In
Balaji Srinivasan's interview on
Justin Kan's Quest podcast, he talks about blockchain as heralding a shift from 1 root user to N root users, and I noticed that this dovetails with the NNTD, and I want to elaborate on that.
One of the core principles of the NNTD is that of
symmetry. Essentially, insofar as there exists ways of relating to the world and to others that are more coordinate-with-able in some general sense, and insofar as there are attractor basins around those, how do we discern who else is further along in what ways, starting from a position of ignorance where we can't take for granted that someone knows how to do it and we just need to do it like them?
(that was an awkward sentence. anyway)
Hmm, I'd like to build this up almost as a logical argument. Maybe all of this is obvious already but I think it'll be satisfying to do.
This is quite easy to prove by offering a simple extreme example: someone who kills people who disagree with them will be less able to think effectively with others than someone who listens to and tries to understand the points of view of others.
This is not just a matter of compatibility. Ceteris paribus, two people who both kill disagreers but who have other disagreements can't coordinate as well as two people who have the same other disagreements but more capacity to listen.
how do we
find what all of these are (both theoretically and instantiatedly)
while also not taking on naive versions of such traits that allow better collaboration with other naive folks but creates security vulnerabilities against others?
This could be naivety in some immediate sense (eg assuming nobody will ever try to cheat you makes you vulnerable to con artists)
or it could be something that creates an feeling of satisfying cothinking but the quality of output of that cothinking is uncompetitive against existing agents (of similar individual intelligence).
"essentially always" is of course a huge bar that's not worth trying to clear, in that ofc any set of particular approaches will be more workable for some contexts than others and for any particular strategy there's a counterstrategy or perspective from which it is precisely unworkable. But we might say something like "essentially always allows you to collaborate better with others who are also trying to solve the puzzle of how to collaborate better with others (etc) even if they're coming from a completely different approach"
In essence, what's being said here is that there's an underlying truth to what makes collective intelligence work, rather than it simply being a matter of arbitrary shared language (although this is also a necessary component, including simply in the literal sense of vocabulary and grammar and shared cultural literacy).
So if we've concluded (or assume arguendo) such truth exists, how do we find it amidst the cacophony of competing claims about what it means to communicate well, act with integrity, or behave ethically? In other words, what does a rigorous science studying human relating look like, given that we can't get outside of our own perspectives and we all have blindspots? What does discovery in such a field look like? What does knowledge transfer look like? (Does knowledge transfer also transfer blindspots?)
(I guess [citation needed] on "we all have blindspots". idk who would question that but I just want to acknowledge that I threw in an unqualified new claim there)
Given that science is fundamentally descriptive, not prescriptive, we are concerned with phenomena as they are in fact occurring. Physics is the study of particles as they behave, not as they "should". Biology is the study of life as it behaves, not as it "should".
Then, if we assume that it's possible to have way higher levels of collective intelligence than we do now, by coordinating human thinking together in various ways...
(This assumption seems justified by both the fact that the burden of proof of something being impossible lies on whoever says it can't be done, and by the fact that we are so beyond fucked as a species if this is not true that we will either figure this out or die trying and so anyone trying to figure it out must necessarily assume we can do it. (I guess some might argue that friendly AI could solve things instead but from my perspective most of the same basic principles of trust-building and coordination apply there and are probably HARDER to get right there than in the human case. I'm hugely in favor of people working on this though!))
If it's possible to have way higher levels of collective intelligence, we should then assume that we will get there (unless we die first) from an almost Molochian perspective - since it would outcompete current levels on most playing fields.
But ah, speaking of Moloch... I've been using a frame this whole time that assumes this greater collective intelligence would be a good thing. What reason(s) have I to think that?
Even if the collective intelligence is in some sense simply self-interested, any intelligence on the planet right now would want to solve threats to its existence—AGI, nuclear risk, etc... which current intelligences may not be capable of at their scale
In general, a whole that is being supported can support its individuals way better than a whole that's being undermined
As far as I can tell, coercion is the main thing blocking greater collective intelligence