11 ms·

When I was a kid, I was fascinated by the word "emergence". In pop science, it was spoken in reverent tones. Never mind that they never gave a definition of it

by knzhou 6y ago

When I was a kid, I was fascinated by the word "emergence". In pop science, it was spoken in reverent tones. Never mind that they never gave a definition of it that didn't sound trivial -- the word was dripping with meaning and the promise of hidden wisdom. I went to college and studied physics. I learned how the behavior of many atoms leads to thermodynamics and the properties of materials. I learned how elementary quantum mechanics leads to atomic physics and chemistry. I learned how the strong force gives rise to nuclei. I learned how the Standard Model produces everything on this list. I learned how, at each layer, the correspondence was fiendishly subtle, but the work of thousands of physicists over decades had built the bridges. At the end of it, I was still dissatisfied, because I hadn't learned what emergence was. I'd see crackpots presenting their personal theories of everything, declaring that they knew what no physicist ever could, because their model had emergence. They knew about it, and we didn't. But I still didn't know what it was. Then I dug deeper and it became obvious. Emergence is a beautiful idea that has always been present, in some form or another, in every field of science. It is so well-developed in some fields, such as particle physics, that it has become a suite of quantitative methods with incredible accuracy. I had been learning it the entire time, and now use it every day without a second thought. Meanwhile, "emergence" is a brand name. It's a specific word passed down over generations of pop science used to convey mystique. It's a tool some charlatans use to pretend they know more than they do. No offense meant to the article, but I often hear people say they're most excited about some substanceless theory because it "has emergence", and this is part of why.

mistermann 6y ago
What did you conclude though...is there no such thing, or do you think there is but no one has figured it out, yet?
bonoboTP 6y ago
I agree it's wooey in many cases, and is used as a near synonym of miracle or magic. But I think a good summary is that emergence happens when relations between things on a finer level of analysis become the things themselves on a coarser, higher level AND that these new things also have rich and complex interactions among them, making the levels roughly equally interesting, complicated and useful. Overall the bottom level is "enough" in theory but rising up the levels allows us to gain orders of magnitude more "practicality". Sometimes by common sense we stumble upon a mid-level concept and then when we learn about the lower level explanation, we proclaim that the original phenomenon was "emergent". Now it's a great (but I think not very fruitful) philosophical debate to decide which level has how much "realness", is the higher level just a convenient fiction, an illusion, a practical description and there is a lowest, rock bottom level which the universe/God really "cares about", or are all levels just as real and the whole hierarchy/ladder is just our conception,and not only can rising up the abstraction be an artificial action, but maybe descending lower can also be a illusory/non-natural and a result of our search for practicality and the rock bottom isn't really that real and the universe doesn't know to care about it. Just as in math the naive explanation is we have fundamental axioms and we derive the higher level theorems from it. But in reality what happens is we know what theorems we want (to make things interesting or empirically useful) and then look for ways to structure the axioms to give rise to the "real" theorems. Or, what view of a signal is more real and fundamental, the time-domain or the Fourier-domain? Both can be equally real or equally fundamental, and can give rise to equally interesting Analyse. Or perhaps not, and the time domain is what the universe cares about and has in its "source code" (which is probably a misplaced metaphor though) and Fourier is just for convenience... Other times when making the relations into thing (reifying them) and vice versa we don't really move higher or lower, we stay roughly on the same level. An example is duality in graph theory (faces to nodes, nodes to faces) or in optimization (exchanging the role of variables and equations). And at the risk of sounding wooey, this is similar to the duality, non-duality ideas in Eastern philosophy, looking at the same stuff but interpreting "the gaps" as "the things" and vice versa. And then saying neither is more real, both are real and neither are, together they are real, separately they aren't etc. Maybe the levels are not a strict hierarchy, maybe they loop back, maybe consciousness is indeed such a feedback loop across different levels (a la Hofstadter), then you get to resonances in feedback loops and if you push it far it gets quite wooey. But it's still very differently wooey than some quantum healer fixing the resonances in your liver through the TV.
- Thorentis 6y ago
  Isn't this what abstraction is? A computer looks magical if you don't understand binary logic gates and transistors etc. Even though I understand it all, the things my computer can do seems magical. I think this is what people mean when they say "emergent" but for some reason don't use more widely understood terms like "high levels of abstraction". Perhaps this hints at us not even understanding the lower levels of many areas well enough yet.
  - bonoboTP 6y ago
    I expanded my comment because an important thing is to discuss whether the different levels are actually a property of the universe or just a property of our description? Is math discovered or invented. What is part of the map and what of the territory. I think people use emergence more when it's viewed and suggested as independent from our description, it really emerges on its own and we just discover and describe this. And "abstraction" is used when we view our role greater, we invented an abstraction to easier manage the complexity. So emergence is more incidental, accidental, serendipitous and unexpected, while abstraction is intentional and purposeful and designed. Now which is which and whether the distinction itself touches on something real or is just a different view of the same thing, is another great topic for discussion.
- walleeee 6y ago
  > But I think a good summary is that emergence happens when relations between things on a finer level of analysis become the things themselves on a coarser, higher level AND that these new things also have rich and complex interactions among them, making the levels roughly equally interesting, complicated and useful. Overall the bottom level is "enough" in theory but rising up the levels allows us to gain orders of magnitude more "practicality". Hofstadter discusses this in his book I Am a Strange Loop: there's a line that's something like "sometimes the bottom level - despite being entirely responsible for the effect in question - is nonetheless totally irrelevant to it"
tasuki 6y ago
You might enjoy https://www.readthesequences.com/The-Futility-Of-Emergence https://www.readthesequences.com/The-Futility-Of-Emergence
- wizzwizz4 6y ago
  I prefer reading these on LessWrong, because people often point out when Eliezer's said something unjustified. (Often he provides the justification in the comments, but on a couple of occasions there hasn't been a justification – he's just been flat-out wrong.) This one's pretty good, though. https://www.lesswrong.com/posts/8QzZKw9WHRxjR4948/the-futility-of-emergence https://www.lesswrong.com/posts/8QzZKw9WHRxjR4948/the-futili...
  - wizzwizz4 6y ago
    Turns out that https://www.readthesequences.com/ https://www.readthesequences.com/ contains the edits for the book version, so if you're reading them all through it'll probably be more coherent on that website. Dilemma!

by knzhou 6y ago

mistermann 6y ago

What did you conclude though...is there no such thing, or do you think there is but no one has figured it out, yet?

bonoboTP 6y ago

I agree it's wooey in many cases, and is used as a near synonym of miracle or magic. But I think a good summary is that emergence happens when relations between things on a finer level of analysis become the things themselves on a coarser, higher level AND that these new things also have rich and complex interactions among them, making the levels roughly equally interesting, complicated and useful. Overall the bottom level is "enough" in theory but rising up the levels allows us to gain orders of magnitude more "practicality". Sometimes by common sense we stumble upon a mid-level concept and then when we learn about the lower level explanation, we proclaim that the original phenomenon was "emergent". Now it's a great (but I think not very fruitful) philosophical debate to decide which level has how much "realness", is the higher level just a convenient fiction, an illusion, a practical description and there is a lowest, rock bottom level which the universe/God really "cares about", or are all levels just as real and the whole hierarchy/ladder is just our conception,and not only can rising up the abstraction be an artificial action, but maybe descending lower can also be a illusory/non-natural and a result of our search for practicality and the rock bottom isn't really that real and the universe doesn't know to care about it. Just as in math the naive explanation is we have fundamental axioms and we derive the higher level theorems from it. But in reality what happens is we know what theorems we want (to make things interesting or empirically useful) and then look for ways to structure the axioms to give rise to the "real" theorems. Or, what view of a signal is more real and fundamental, the time-domain or the Fourier-domain? Both can be equally real or equally fundamental, and can give rise to equally interesting Analyse. Or perhaps not, and the time domain is what the universe cares about and has in its "source code" (which is probably a misplaced metaphor though) and Fourier is just for convenience... Other times when making the relations into thing (reifying them) and vice versa we don't really move higher or lower, we stay roughly on the same level. An example is duality in graph theory (faces to nodes, nodes to faces) or in optimization (exchanging the role of variables and equations). And at the risk of sounding wooey, this is similar to the duality, non-duality ideas in Eastern philosophy, looking at the same stuff but interpreting "the gaps" as "the things" and vice versa. And then saying neither is more real, both are real and neither are, together they are real, separately they aren't etc. Maybe the levels are not a strict hierarchy, maybe they loop back, maybe consciousness is indeed such a feedback loop across different levels (a la Hofstadter), then you get to resonances in feedback loops and if you push it far it gets quite wooey. But it's still very differently wooey than some quantum healer fixing the resonances in your liver through the TV.

Thorentis 6y ago
Isn't this what abstraction is? A computer looks magical if you don't understand binary logic gates and transistors etc. Even though I understand it all, the things my computer can do seems magical. I think this is what people mean when they say "emergent" but for some reason don't use more widely understood terms like "high levels of abstraction". Perhaps this hints at us not even understanding the lower levels of many areas well enough yet.
- bonoboTP 6y ago
  I expanded my comment because an important thing is to discuss whether the different levels are actually a property of the universe or just a property of our description? Is math discovered or invented. What is part of the map and what of the territory. I think people use emergence more when it's viewed and suggested as independent from our description, it really emerges on its own and we just discover and describe this. And "abstraction" is used when we view our role greater, we invented an abstraction to easier manage the complexity. So emergence is more incidental, accidental, serendipitous and unexpected, while abstraction is intentional and purposeful and designed. Now which is which and whether the distinction itself touches on something real or is just a different view of the same thing, is another great topic for discussion.
walleeee 6y ago
> But I think a good summary is that emergence happens when relations between things on a finer level of analysis become the things themselves on a coarser, higher level AND that these new things also have rich and complex interactions among them, making the levels roughly equally interesting, complicated and useful. Overall the bottom level is "enough" in theory but rising up the levels allows us to gain orders of magnitude more "practicality". Hofstadter discusses this in his book I Am a Strange Loop: there's a line that's something like "sometimes the bottom level - despite being entirely responsible for the effect in question - is nonetheless totally irrelevant to it"

tasuki 6y ago

You might enjoy https://www.readthesequences.com/The-Futility-Of-Emergence https://www.readthesequences.com/The-Futility-Of-Emergence

wizzwizz4 6y ago
I prefer reading these on LessWrong, because people often point out when Eliezer's said something unjustified. (Often he provides the justification in the comments, but on a couple of occasions there hasn't been a justification – he's just been flat-out wrong.) This one's pretty good, though. https://www.lesswrong.com/posts/8QzZKw9WHRxjR4948/the-futility-of-emergence https://www.lesswrong.com/posts/8QzZKw9WHRxjR4948/the-futili...
- wizzwizz4 6y ago
  Turns out that https://www.readthesequences.com/ https://www.readthesequences.com/ contains the edits for the book version, so if you're reading them all through it'll probably be more coherent on that website. Dilemma!