One of the goals of LW is to be a forum for the development of rationality, or the application of a particular mindset to oneself and the world. I think this involves a combination of philosophical commitments and practical skills; looking at how that was balanced in a similar situation in the past helps clarify what’s happening in the present and what some of the counterfactuals are.
In many ways, psychological phenomena runs into the problems of early astronomy. Different people in different parts of the world see slightly different skies and have different weather; you might only have one observation of where things were on any particular night. Knowledge that a particular tool or technique works for one person might have little bearing on whether or not it works for others, which suggests a focus on local observations.
That said, LW is about a very computational and mathematical view of the world. It reminds me of LaPlace’s Demon, the sense that something could have the universe’s true dynamics and run all the numbers and predict things exactly, and there’s something spiritually solid about yearning for that standard. (It isn’t even shaken by quantum indeterminacy, which is neatly resolved by many-world’s choice of “branch both ways”, such that in the broader multiverse both exist, and you’re just uncertain over which branch you end up in. Solomonoff induction and Tegmark’s multiverse extend it even further, such that even uncertainty over the universe’s true dynamics doesn’t upset this view!)
This feels importantly different from, idk, a working mathematician’s view of math? Hypothesizing that somewhere out there are all the true propositions and all the false propositions is pretty different from the set of known-true-propositions, and the methodology for determining what goes in that set. Or a working software engineer’s understanding of computation, in which runtime considerations are important instead of ignorable.
And so the particularist’s observation of a specific interval of spacetime and the universalist’s theory of everything clash.
LW as futurist hub
Beyond that, I think LW is a forum with lots of people with a particular outlook (which I’ll call ‘futurist’). We’re living in the hinge of history, AI is important, the humanist neo-Enlightenment is good, many other assumptions that I won’t bother to list out.
This, in some ways, reminds me of the relationship between gentlemen-as-a-class and England. (Maybe in a future post I’ll elaborate on how I think about classes and their relationship to society; here I’ll just mention the association.) When I think about LW as an institution or futurists-as-a-class, many of the same practical social considerations come into play: how do we keep the conversation going? How do we engage in the practical business of living?
And, moreover, what do our roles and templates look like? What is our bricolage from the materials at hand in our culture?
To talk about the universalism vs particularism issue, one of the issues with universalism is that it’s trying to solve either provenly hard problems, in the sense that it requires exponential or worse efficiency for an algorithm to do it, or we suspect that it’s really hard to do, and we just haven’t proved it.
One of the best examples here is learning efficiently from data, and there’s a line in a paper that talks about one of the issues for universalism in practice:
Any polynomial-time algorithm for finding a hypothesis consistent with the data would
imply a polynomial-time algorithm for breaking widely-used cryptosystems such as RSA!
And this is considered unlikely by complexity theorists and cryptographers.
This seems similar to the issue mentioned below:
This feels importantly different from, idk, a working mathematician’s view of math? Hypothesizing that somewhere out there are all the true propositions and all the false propositions is pretty different from the set of known-true-propositions, and the methodology for determining what goes in that set. Or a working software engineer’s understanding of computation, in which runtime considerations are important instead of ignorable.
The problem for mathematics in general is at least in the RE complexity class, which is wildly intractable, and even propositional logic is NP-complete in satisfiability. These are really hard problems in the general case, so hard that without radical assumptions about physics will likely remain intractable in the general case.
Universalism is a currently impractical doctrine, but one that is sometimes useful.
LW as rationality hub
One of the goals of LW is to be a forum for the development of rationality, or the application of a particular mindset to oneself and the world. I think this involves a combination of philosophical commitments and practical skills; looking at how that was balanced in a similar situation in the past helps clarify what’s happening in the present and what some of the counterfactuals are.
In many ways, psychological phenomena runs into the problems of early astronomy. Different people in different parts of the world see slightly different skies and have different weather; you might only have one observation of where things were on any particular night. Knowledge that a particular tool or technique works for one person might have little bearing on whether or not it works for others, which suggests a focus on local observations.
That said, LW is about a very computational and mathematical view of the world. It reminds me of LaPlace’s Demon, the sense that something could have the universe’s true dynamics and run all the numbers and predict things exactly, and there’s something spiritually solid about yearning for that standard. (It isn’t even shaken by quantum indeterminacy, which is neatly resolved by many-world’s choice of “branch both ways”, such that in the broader multiverse both exist, and you’re just uncertain over which branch you end up in. Solomonoff induction and Tegmark’s multiverse extend it even further, such that even uncertainty over the universe’s true dynamics doesn’t upset this view!)
This feels importantly different from, idk, a working mathematician’s view of math? Hypothesizing that somewhere out there are all the true propositions and all the false propositions is pretty different from the set of known-true-propositions, and the methodology for determining what goes in that set. Or a working software engineer’s understanding of computation, in which runtime considerations are important instead of ignorable.
And so the particularist’s observation of a specific interval of spacetime and the universalist’s theory of everything clash.
LW as futurist hub
Beyond that, I think LW is a forum with lots of people with a particular outlook (which I’ll call ‘futurist’). We’re living in the hinge of history, AI is important, the humanist neo-Enlightenment is good, many other assumptions that I won’t bother to list out.
This, in some ways, reminds me of the relationship between gentlemen-as-a-class and England. (Maybe in a future post I’ll elaborate on how I think about classes and their relationship to society; here I’ll just mention the association.) When I think about LW as an institution or futurists-as-a-class, many of the same practical social considerations come into play: how do we keep the conversation going? How do we engage in the practical business of living?
And, moreover, what do our roles and templates look like? What is our bricolage from the materials at hand in our culture?
To talk about the universalism vs particularism issue, one of the issues with universalism is that it’s trying to solve either provenly hard problems, in the sense that it requires exponential or worse efficiency for an algorithm to do it, or we suspect that it’s really hard to do, and we just haven’t proved it.
One of the best examples here is learning efficiently from data, and there’s a line in a paper that talks about one of the issues for universalism in practice:
And this is considered unlikely by complexity theorists and cryptographers.
This seems similar to the issue mentioned below:
The problem for mathematics in general is at least in the RE complexity class, which is wildly intractable, and even propositional logic is NP-complete in satisfiability. These are really hard problems in the general case, so hard that without radical assumptions about physics will likely remain intractable in the general case.
Universalism is a currently impractical doctrine, but one that is sometimes useful.