That’s shaping up to be a really interesting club. If the world wasn’t currently on fire, or if I enjoyed this sort of thing, I would’ve considered taking the time to write up some essays as an application for joining it, something about efficient communication (as opposed to effective-if-you-try-real-hard) and pragmatic navigation of disagreement (that cultivates progress towards changing one’s mind without unsustainable urgency). But I might be on track anyway (edit: the reference is to the fact that the linked comment was surprisingly heavily karma-downvoted; now it’s back to the positives, and some disagreement-vote is reasonable).
It does seem that Lesswrong is becoming more “archipelagic” lately, which… I mean… I guess that’s what it now says on the tin is what it is explicitly aiming for? So it seems hard to complain <3
That said, I wish the mods and perhaps Duncan himself would spend more time thinking about the CAP theorem, and the importance and value of healing “partitions” quickly and thoroughly, rather than taking the risk of letting minor confusing disagreements explode into all out forks.
partitions are good, as long as information flow between them continues, and they are able to have useful disagreements. Those are not trivial desiderata, but they’re achievable, and it’s worth it to allow and encourage there to be multiscale grouping. We should be trying to reduce graph orientability; every node should make their own decisions, and yet via each node observing the graph and making themselves more like outliers in connectivity pattern, no node should stay an outlier in connectivity statistics. we need scale free soft groupings with no outgrouping.
I think, ideally, for anything that really matters, I’d selfishly prefer to just be in consensus with flawless reasoners, by sharing the same key observations, and correctly deriving the same important conclusions?
The whole definition of a partition here is “information NOT flowing because it CANNOT easily flow because cuts either occurred accidentally or were added on purpose(?!)” and… that’s a barrier to sharing observations, and a barrier to getting into consensus sort of by definition?
It makes it harder for all nodes to swiftly make the same good promises based on adequate knowledge of the global state of the world because it makes it harder for facts to flow from where someone observed them to where someone could usefully apply the knowledge.
If an information flow blockage persists for long enough then commitments can be accidentally be made on either side of the blockage such that the two commitments cannot be both satisfied (and trades that could have better profited more people if they’d been better informed don’t happen, and so on), and in general people get less of what they counted on or would have wanted, and plans have to be re-planned, and it is generally just sad.
I’m in favor of privacy, if that’s what you’re talking about? But I don’t see how reply bans advance the normal and valid goals of privacy. If someone is spreading lashon hara, I think reply bans would tend to make it worse, if anything?
Maybe you’re saying something super clever about “multiscale grouping”? When I google that term I just find stuff that… might be old school machine vision algorithms? Maybe there is a metaphor here, but I don’t see it yet.
I think, ideally, for anything that really matters, I’d selfishly prefer to just be in consensus with flawless reasoners, by sharing the same key observations, and correctly deriving the same important conclusions?
But you’re embedded in physics, and can rely on the fact that you will never be a flawless reasoner. You’re made of neurons, none of which are flawless reasoners, but they are able to work together to be a single agent by nature of keeping each other informed about what your opinion is as that opinion gets refined. Your neurons operate at or near criticality, so any neuron could potentially cause an update that propagates through the whole brain; neurons’ uncertainty about whether other neurons will provide an insightful contribution, combined with consensus network that refines away errors in ways that diffuse towards your self, is what allows free will to fall out of a deterministic system: your neurons inform each other of your personality, and you move your environment towards yourself.
In a social network, overly dense connectivity can break edge-of-chaos, criticality-seeking behavior, by resulting in a network that accept updates from people with too little processing. This is especially severe when there’s any sort of hierarchy, especially when that hierarchy is related to a hierarchy of control or dominance.
I propose that, if there are conflicts about approach to reasoning, information flow should continue, and if things go well, the partition should be one that results in the networks staying overlapped but separating partially.
(I do not intend to be at all metaphorical. I am intending to make claims that these patterns are literally the same, not mere metaphor. If they are not literally the same, my claim is wrong, and discovering it will teach me new things.)
I can link some lectures I’ve watched recently about this. Eg, I liked this one on “what is complexity”, which goes over how complex systems science is about the process of understanding what laws can be stated universally about large systems that are neither simple due to high entropy nor simple due to low entropy. It is not highly relevant such that it is worth it if that’s not new to you, but if it is new to you, it may be important background knowledge.
Also, keep in mind that there’s a good chance I’m straight up just not as smart or educated as most people on here; I compensate for that the same sort of way as current models do—I’ve seen a lot more stuff shallowly than most people study deeply. (but a real PhD would actually be good at stuff I merely fangirl about.)
on reread, seems like I may have missed components of your reply in my reply. I’m about to sleep; if you reply to me with emphasis on which parts I missed I’ll reply tomorrow
Note for the record:
I can no longer reply to comments addressed to me (or referring to me, etc.) on the post “Basics of Rationalist Discourse”, because its author has banned me from commenting on any of his posts.
That’s shaping up to be a really interesting club. If the world wasn’t currently on fire, or if I enjoyed this sort of thing, I would’ve considered taking the time to write up some essays as an application for joining it, something about efficient communication (as opposed to effective-if-you-try-real-hard) and pragmatic navigation of disagreement (that cultivates progress towards changing one’s mind without unsustainable urgency). But I might be on track anyway (edit: the reference is to the fact that the linked comment was surprisingly heavily karma-downvoted; now it’s back to the positives, and some disagreement-vote is reasonable).
It does seem that Lesswrong is becoming more “archipelagic” lately, which… I mean… I guess that’s what it now says on the tin is what it is explicitly aiming for? So it seems hard to complain <3
That said, I wish the mods and perhaps Duncan himself would spend more time thinking about the CAP theorem, and the importance and value of healing “partitions” quickly and thoroughly, rather than taking the risk of letting minor confusing disagreements explode into all out forks.
partitions are good, as long as information flow between them continues, and they are able to have useful disagreements. Those are not trivial desiderata, but they’re achievable, and it’s worth it to allow and encourage there to be multiscale grouping. We should be trying to reduce graph orientability; every node should make their own decisions, and yet via each node observing the graph and making themselves more like outliers in connectivity pattern, no node should stay an outlier in connectivity statistics. we need scale free soft groupings with no outgrouping.
https://www.microsolidarity.cc/
I think, ideally, for anything that really matters, I’d selfishly prefer to just be in consensus with flawless reasoners, by sharing the same key observations, and correctly deriving the same important conclusions?
The whole definition of a partition here is “information NOT flowing because it CANNOT easily flow because cuts either occurred accidentally or were added on purpose(?!)” and… that’s a barrier to sharing observations, and a barrier to getting into consensus sort of by definition?
It makes it harder for all nodes to swiftly make the same good promises based on adequate knowledge of the global state of the world because it makes it harder for facts to flow from where someone observed them to where someone could usefully apply the knowledge.
If an information flow blockage persists for long enough then commitments can be accidentally be made on either side of the blockage such that the two commitments cannot be both satisfied (and trades that could have better profited more people if they’d been better informed don’t happen, and so on), and in general people get less of what they counted on or would have wanted, and plans have to be re-planned, and it is generally just sad.
I’m in favor of privacy, if that’s what you’re talking about? But I don’t see how reply bans advance the normal and valid goals of privacy. If someone is spreading lashon hara, I think reply bans would tend to make it worse, if anything?
Maybe you’re saying something super clever about “multiscale grouping”? When I google that term I just find stuff that… might be old school machine vision algorithms? Maybe there is a metaphor here, but I don’t see it yet.
But you’re embedded in physics, and can rely on the fact that you will never be a flawless reasoner. You’re made of neurons, none of which are flawless reasoners, but they are able to work together to be a single agent by nature of keeping each other informed about what your opinion is as that opinion gets refined. Your neurons operate at or near criticality, so any neuron could potentially cause an update that propagates through the whole brain; neurons’ uncertainty about whether other neurons will provide an insightful contribution, combined with consensus network that refines away errors in ways that diffuse towards your self, is what allows free will to fall out of a deterministic system: your neurons inform each other of your personality, and you move your environment towards yourself.
In a social network, overly dense connectivity can break edge-of-chaos, criticality-seeking behavior, by resulting in a network that accept updates from people with too little processing. This is especially severe when there’s any sort of hierarchy, especially when that hierarchy is related to a hierarchy of control or dominance.
I propose that, if there are conflicts about approach to reasoning, information flow should continue, and if things go well, the partition should be one that results in the networks staying overlapped but separating partially.
(I do not intend to be at all metaphorical. I am intending to make claims that these patterns are literally the same, not mere metaphor. If they are not literally the same, my claim is wrong, and discovering it will teach me new things.)
I can link some lectures I’ve watched recently about this. Eg, I liked this one on “what is complexity”, which goes over how complex systems science is about the process of understanding what laws can be stated universally about large systems that are neither simple due to high entropy nor simple due to low entropy. It is not highly relevant such that it is worth it if that’s not new to you, but if it is new to you, it may be important background knowledge.
Also, keep in mind that there’s a good chance I’m straight up just not as smart or educated as most people on here; I compensate for that the same sort of way as current models do—I’ve seen a lot more stuff shallowly than most people study deeply. (but a real PhD would actually be good at stuff I merely fangirl about.)
on reread, seems like I may have missed components of your reply in my reply. I’m about to sleep; if you reply to me with emphasis on which parts I missed I’ll reply tomorrow