This New Idea Could Explain Complexity
ฝัง
- เผยแพร่เมื่อ 28 ก.ย. 2024
- Check out courses about science, computer science, or math on Brilliant! First 30 days are free and 20% off the annual premium subscription when you use our link ➜ brilliant.org/....
The universe creates complexity out of simplicity, but despite many attempts at understanding how, scientists still have not figured it out. We do know that complexity relies on the emergence of new features and laws, but then again we don't understand emergence either. The first step must be to clearly define what we are talking about and to measure it. A group of scientists now put forward a way to do exactly this. Let’s have a look.
Paper here: arxiv.org/abs/...
Correction to what I say at 04:07 "You will still get the correct prediction". I meant to convey that the prediction doesn't get worse if you average "lumped" classes rather than the full set. Either way, it will be a probabilistic prediction so it's correct only in a statistical sense either way.
🤓 Check out my new quiz app ➜ quizwithit.com/
💌 Support me on Donorbox ➜ donorbox.org/swtg
📝 Transcripts and written news on Substack ➜ sciencewtg.sub...
👉 Transcript with links to references on Patreon ➜ / sabine
📩 Free weekly science newsletter ➜ sabinehossenfe...
👂 Audio only podcast ➜ open.spotify.c...
🔗 Join this channel to get access to perks ➜
/ @sabinehossenfelder
🖼️ On instagram ➜ / sciencewtg
#science #sciencenews #math #complexity #maths
I gave this some thought when I was sequencing DNA in 1994 using the God awful method "dideoxy chain termination sequencing" method. Very painful, but the point is. Identical twins are the same (ignoring the complexity) and yet you find a large number of emergent properties between them that are unexplainably different. While there are aspects we can certainly lump together, there are strange permutations that are unexpected.
That's very interesting, hadn't thought about this. Thanks for sharing!
We are all the same (ignoring the complexity)!
I was still using sanger sequencing in 2017 :(
@dr: nature vs nurture is a long standing topic. Monozygotic twins are the ideal study objects for such research.
@@dtrcs9518 Was your face as sunburned as mine?
what a great explanation of such a fascinating subject. Our cat is called Kant, now he gets a new name, we call him Emergence. He is lumped with dogs
Are you relating to his behavioral aspect?
@@Dr.M.VincentCurley you Kant do that.
I was told that a good cat name should have a hissing sound (like psps) because that is grabbing their attention, or something like that. "Emergence" would better, if this is true. (But maybe your cat is different and likes these names, who knows?)
If your intuition is to name your cat Kant, then it's too late!! You've already done it!!
@carl: hope you‘re not living in an english spoken region. If you call your cat like that, the neighbours could be irritated 😄
EDIT: I can’t get rid of that picture in my head now and keep giggling 😄
"Usually these people are computer scientists." Shots fired.
As a software engineer, I laughed out loud
@@corbono As a software engineer, I laugh at your laughing out loud.
@@anthonylosego As someone who took an introductory webdev course, I lol'd at your lol'ing bc their lol.
@@anthonylosego as a software engineer i ask: could you explain? Didn't get the joke (well her's, not yours)
Most software engineers know diddly squat about anything except their own interminable code that defies comprehension to normal Man. I can't speak for how Women see things.
A research group led by Dr. Hector Zenil offers a deeper view on the concept of 'emergence' (in science and philosophy), pushing beyond traditional information theory approaches like Shannon entropy. Their work (eg., the field of Algorithmic Information Dynamics, see, for instance: “Emergence and algorithmic information dynamics of systems and observers”, published by the Royal Society) tackles the limitations of using stochastic processes to explain causality, emphasizing the need for a more computational and algorithmic approach. This insights into causal emergence provides a fresh perspective that moves past simple correlations.
Yes, that is total bullshit. Nature does not compute anything, on any level.
3:46 I get the idea, but I hate this example, because I have yet to see a traffic flow prediction that was actually correct. They always predict far more traffic than in reality.
In my experience, there are two reasons for this:
First, traffic engineers almost exclusively plan for moving as many cars as possible, rather than as many people as possible, so they overplan for cars to the detriment to every other form of transportation.
Second, traffic engineers are paid to build roads. They are not paid to _not_ build roads. So every projection always results in (surprise!) proof that they need to build more roads.
Incidentally this highlights an interesting difference between "traffic engineers" and "transportation engineers" and I talked about that with Build the Lanes on an episode of the Urbanist Agenda podcast.
Such a great (albeit non sequitur) comment to see on a Sabine vid!
One of my biggest pet peeves when driving is that roads are built exclusively for people who already know where they are going. Which makes driving in the city a miserable experience. Need to park somewhere? better hope you're lucky enough to be on the right side of the road and can spot the almost indistinguishable parking turning before you pass it. If not, add another 15 to 45 minutes to your trip so you can try again.
Not just bikes!!!!
If they always predict more traffic than there really is, why is traffic always congested? Shouldn't it be the other way around?
I don’t undestand the details, but a clear understanding emerges.
LUMPABILITY = integration or summation of states, dimensions -- a syntropic process!
Complexity is dual to simplicity.
Micro is dual to macro.
Increasing the number of dimensions or states is an entropic process -- differentiation or reductionism.
Decreasing the number of states or dimensions is a syntropic process -- integration or holism, LUMPABILITY.
Increasing (divergence, entropy) is dual decreasing (syntropy, convergence).
Syntropy (prediction) is dual to increasing entropy -- the 4th law of thermodynamics!
Teleological physics (syntropy) is dual to non teleological physics (entropy) -- physics is dual.
Integration is dual to differentiation.
Reductionism is dual to holism.
Homology (syntropy) is dual to co-homology (entropy).
"Always two there are" -- Yoda.
@@hyperduality2838 if two depend entirely on one another and cannot have an existence in isolation without the counterpart, could they not be described as two parts of one? There is no male without female or female without male. How do two things give rise to each other when they both are dependent on the other which they give rise to in order to exist in the first place
@@hyperduality2838I am surprised nobody mentions about statistical mechanics. Lumpability seems to be just dimensionality reduction for space and ensemble averaging for time. Some degrees of freedom/dimensions matter a lot less at larger length scales (thus can be marginalised out), and microscopic observables can be average in time (assuming that the ensemble is not perturbed too frequently).
@@__-op4qm Lumpability is just a clumsy way of saying syntropy (convergence).
Your mind associates or integrates information -- integrated information theory.
Syntactic information is dual to semantic information -- information is dual.
Syntax (objective, absolute) is dual to semantics (subjective, relative) -- languages or communication.
If mathematics is a language then it is dual.
Syntropy (prediction) is dual to increasing entropy -- the 4th law of thermodynamics!
Teleological physics (syntropy) is dual to non teleological physics (entropy) -- physics is dual.
Categories (form, syntax) are dual to sets (substance, semantics) -- Category theory.
"Only the Sith think in terms of absolutes!" -- Obi Wan Kenobi.
"Always two there are" -- Yoda.
All messages in a communication system are predicted into existence according to Shannon's information theorem -- a syntropic process, teleological.
Making predictions to track targets, goals and objectives is a syntropic process, teleological.
In homology you start with hypervolumes and reduce the number of dimensions:-
Hypervolumes become volumes become planes or surfaces become lines become points or zero dimensions and this is clearly a dimension reduction process -- syntropic.
Points are dual to lines -- the principle of duality in geometry.
In co-homology you do the opposite or opposame:-
Points becomes lines become planes become volumes become hypervolumes -- increasing dimensions or states is an entropic process.
Convergence (syntropy, homology) is dual to divergence (entropy, co-homology or dual homology).
Sine is dual to cosine or dual sine -- the word co means mutual and implies duality.
Vectors (contravariant) are dual to co-vectors (covariant) -- dual bases.
Riemann geometry or curvature is dual -- upper indices are dual to lower indices (tensors).
Gravitation is equivalent or dual (isomorphic) to acceleration -- Einstein's happiest thought, the principle of equivalence (duality).
Space is dual to time -- Einstein.
@@__-op4qm "The brain is a prediction machine" -- Karl Friston, neuroscientist.
Your brain integrates information to make predictions -- a syntropic process, teleological.
Certainty (predictability, syntropy) is dual to uncertainty (unpredictability, entropy) -- the Heisenberg certainty/uncertainty principle.
Hence there is a 4th law of thermodynamics.
Synergy is dual to energy -- energy is dual.
Synchronic points/lines are dual to enchronic points/lines.
Highly lumpable is such a good insult
to your mom?
There are much worst than a mere insult:
"Mrs, your test results show your breast is highly lumpable"
It's not, it applies only to a group of things.
@@Caellyan Yeah, it says that they are easily added to a big group: don't think for themselves or such.
@@Windows__2000 Lumpenproletariat?
Lumpability? I would have gone with "Glomular."
I would have gone with ‘morsular’
Corpuscular
tubular idea
Lumpability. Useful for making complicated things simpler. Secret superpower for mathematicians, you could say! (My AI say's)
That'll mix with the term 'Glomarular'. Glomarular Filtrate is basically urine. So, then it'll become 'piss parameter'.
4:00 - having conducted a few traffic studies in the US... "You still get the correct prediction" is giving us a LOT of credit we probably don't deserve.
But then, the fact that your individual odds of getting stuck at any traffic light (not involving a train, anyway) for more than 120 seconds are basically zero means that statistics kinda work. Which is reassuring.
And yes, rather than use actual stats terminology, traffic engineers and city planners call "the value that tells you how meaningful your grouping selections have been" "Lumpability". The reason for this is that they often have to explain the results of their research to town councils and city board selectpersons who barely understand the concept of "road".
This reminds of software architecture. The underlying purpose of code encapsulation is causal closure to reduce complexity. Similarly the lumpability of data reduces the test complexity to manageable levels
My thoughts exactly, it sounds like it's describing layers of an architecture that are cleanly separated by an abstraction.
Emergence, then, is a "decoupling" of two distinct "scales"?
This is what engineers do as well I think. An example I can think of is using transistors to create logic gates to create structures like adders/multipliers (for an alu) and control units to build a CPU, or transistors going to things like the different stages of an op-amp to piece together the op-amp itself so it can be used to build things like voltage comparators or a difference amplifier or whatever other configuration you want. And these things can be used to create larger circuits.
Yes... layers of abstraction work effectively in the human realm because the natural world already does something similar.
Not the same kind of "complexity", Complexity is an essential aspect of the objects so described. Removing any complexity removes the nature of the object or the entire object itself.
Cat Memes just emerge from the OSI model.
Wolfram's A New Kind of Science uses some of these ideas as a fundamental property, particularly that complexity arises out of simple rules, as with Cellular Automata. What's fascinating is that from simple rules, you can derive BOTH the math of Relativity and Quantum Mechanics, and why they are the way they are is completely explained and understandable. Again, from very simple rules. Simple rules can even produce what seems to be "randomness" -- unpredictability is what's called computational irreducibility. It's impossible to predict the output in the future without running the rules in sequence. It's still early, and we don't have experiments to make predictions and provide evidence that it's "what's really going on" yet, but it's fascinating stuff.
Yes, well said & accurate! That Simple-rule generated "randomness" was used as the cryptographic basis for a generation of NATO radio communication, starting in the late 1980s. Darpa has been ALL OVER this research.
Wolfram is definitely not the first person to look at the problem in depth. I was reading entire books dedicated to the subject in the 90s and Ilya Prigogine did a lot of the original ground work in the area back in the 1960s. I applaud Wolfram's efforts in the area but I don't feel that his research has created much in the way of actual insights beyond simple categorization. If this new paper holds up, it will be one of the first major advances in the subject I've seen in a long time.
Hi, if I understand Wolfram also use Cellular Automata ....to try explain the Universe using simple rules....
@@danheidel True, "complexity science" was not invented by Wolfram, but he has made important contributions.
@@danheidel FYI, wolfram's CA work was in the 1980s. He knew about Prigogine's work, of course, and built on it.
FYI, Wolfram's work had HUGE effects but he is forbidden to mention it for National Security reasons. He's party to NDA from Darpa. The juicy good stuff we aren't allowed to know about. I know, though ...
Who doesn't want a t-shirt with "Highly Lumpable" emblazoned on it? 😆
With an Ompa-Lumpas:
"Ompa-lumpa-dee-di-do. Lumpability is what we do. "
with the L crossed and an H scribbled in?
Lumpability sounds a lot like computational reducibility.
LUMPABILITY = integration or summation of states, dimensions -- a syntropic process!
Complexity is dual to simplicity.
Micro is dual to macro.
Increasing the number of dimensions or states is an entropic process -- differentiation or reductionism.
Decreasing the number of states or dimensions is a syntropic process -- integration or holism, LUMPABILITY.
Increasing (divergence, entropy) is dual decreasing (syntropy, convergence).
Syntropy (prediction) is dual to increasing entropy -- the 4th law of thermodynamics!
Teleological physics (syntropy) is dual to non teleological physics (entropy) -- physics is dual.
Integration is dual to differentiation.
Reductionism is dual to holism.
Homology (syntropy) is dual to co-homology (entropy).
"Always two there are" -- Yoda.
In addition to Wolfram's work, the Santa Fe Institute has done great work in Complexity and Emergence. My favorite novel about Emergence is "Lila" by William Pirsig, the same guy/genius who wrote "Zen and the Art of Motorcycle Maintenance".
Interesting! thanks a lot
I'm still undecided on whether to buy "A New Kind of Science" - it's over a decade old now. Maybe I should just buy a more recent book by SW.
In Lila, though, we were told that a Creator God was necessitated by the second law of thermodynamics. His utter misunderstanding of science disgusted me mightily.
@@michaelstreeter3125 It is worth it if you can find a deal on a second hand copy. It exhaustively and methodically goes through every rule (within the constraints of his rule set). I found it interesting to see the visualizations of each rule in a way that is super easy to grok. It also offers a peak into chaos theory because it is so methodical.
@@evelynsinclair4937 Huh, I must have read the other "Lila" by Pirsig. I have no idea how you came up with your conclusion. "Lila" is more focused on metaphysics than science, and outlines a concrete instantiation of emergence . A number of people I know hated the book. I think it is brilliant.
So much to learn, so little time. Thank you for lumping together such high density information.
LUMPABILITY = integration or summation of states, dimensions -- a syntropic process!
Complexity is dual to simplicity.
Micro is dual to macro.
Increasing the number of dimensions or states is an entropic process -- differentiation or reductionism.
Decreasing the number of states or dimensions is a syntropic process -- integration or holism, LUMPABILITY.
Increasing (divergence, entropy) is dual decreasing (syntropy, convergence).
Syntropy (prediction) is dual to increasing entropy -- the 4th law of thermodynamics!
Teleological physics (syntropy) is dual to non teleological physics (entropy) -- physics is dual.
Integration is dual to differentiation.
Reductionism is dual to holism.
Homology (syntropy) is dual to co-homology (entropy).
"Always two there are" -- Yoda.
Thank you Sabine. This is one of the more interesting and timely podcasts to my particular situation - helping build alternative / supplemental schools for young children in Japan.
In the broadest sense, education as a process of social maturation is an emergent process, though typical institutionalization of that process tends to restrict human potential. My research area as a college instructor was in tapping into the students' intrinsic motivation as a social primate by replacing end of semesters tests and papers with what I termed an "Event-Driven Curriculum" ... student presentations in front of a real audience as a fractal of what we professors do in academic conferences.
By chance (synchronicity?), only a few days ago, I came across a link in substack's "Naked Emperor" to an on-line article in Quanta Magazine named "The New Math of How Large-Scale Order Emerges" ... a great supplement to this video. Thank you again Sabine!
So happy to see that! I’ve been fascinated by threshold effect as general principle forever but never found literature on it… so it’s emergence! Good to know
And emergent properties
Great review of a seminal article. It applies directly to my current research. I will look into how it might be used "by the numbers." That is to say, using the exact ideas and doing rigorous application to the concepts in my work. TY Sabine.
This is rooted in Wolfram's cellular automata work. For anyone interested, he published the book, "A New Kind of Science" with a follow-up in 2023, 20 years after the first book was published. It's an excellent book; my degree is in computer science and I've been fascinated by cellular automata since I first discovered them in the early 80's so stumbling across Wolfram's book made my day.
Does he explain how cellular mechanisms can form when there is no evolutionary pressure to create building blocks of said mechanism that does not yet exist because evolution has no foresight?
Super interesting, thank you for showing this topic. I once, long ago, made computer programs for business applications. The most valuable lesson I learned there is, that the program should have the same structure as the problem you are tackling. This way, the program would be more robust, less prone to mysterious errors, which proved to be true. This gave quite some “freedom”, not sure how to put it, and maintaining the program was rather easy as long as you kept a keen eye on the also changing structure of the problem. Somehow your video rings a bell, I’ll have a look at it a few times.
freedom = maintainability?
low technical debt?
This makes perfect sense as what you are stating in basis is that the “problem” is understood. At this point there is no longer a problem as such, but rather a known characteristic that may be worked with and offset for. This is what your program is doing - providing a naturally fitting response 🕹️
It looks to me these ideas are Stephen Wolfram’s ideas. He published the cellular automata long ago and most recently he and his team talk about observer theory. How this new paper is different from Wolfram’s team ideas apart from the mathematical particularities? I found observer theory the most convincing theory of everything.
I was waiting for Wolfram to be mentioned in the video 😮
Wolfram is utterly naive, regarding computation, information,, and meaning. The three central counterarguments are:
1. it is impossible to xtract meaning from something
2. the claim of fundamentally computational disregards Heisenbergs principle
3. Information is not received, but a result of perception
That theory is neither convincing nor leading anywhere. particularly it is not explaining the mechanism by which complexity arises
Yep, I posted the same thing. This is straight out of Wolfram's A New Kind of Science, except not as advanced. Still, I'm happy to see that the ideas are spreading. People should realize that ANKOS explains both Relativity and QM in terms of simple rules, and if nothing else, it's remarkable that we have ANY theory that can completely resolve those two into the same simple model, without having to invoke some sort of "AND THEN MAGIC HAPPENS", as we do with so much physics ("and then the particle splits at some random time, for magical reasons").
@@monnoo8221 Sure he is. He was also the leader of multiple highly successful Darpa projects to generate Quantum Neural Computing and AI. His books, which are HAMSTRUNG by Non Disclosure Agreements, only say a FRACTION of what he wanted to say. Dapra would not let him publish the key insights & implications for 'national security' reasons.
@@cuthbertallgood7781 Sure is. I'm shocked this knowledge has been so slow spreading.
It's crazy that this paper doesn't mention Stephen Wolfram anywhere...
You beat me to it.
They even use Mathematica for the graphics. They are certainly familiar with Stephen Wolfram :/
Ya, but Wolfram rarely mentions the concepts he's drawing from either. Maybe it's a tit-for-tat happening now?
Much of Mario Bunge's work in the second half of the last century revolves around material systems and emergent properties resulting from their rules of interaction. In general, in complex systems (if possible) it is not trivial to deduce emergent properties from component properties or interaction rules.
An example is the structures that arise in the flock flight of starlings in which each individual averages the flight orientation of its approximately 5-7 nearest neighbors. A simplification of this to be able to work statistically on a computer with flock phenomena is the Vicsek model.
“Emergence” is a transition (conditional) to a higher-level programming language (for example, from assembler to C, etc.)
In fact, translation into machine codes still occurs, but it is hidden from the user.
Did mathematicians really create a formal description to "decide on cohort before running experiments"?
Ya our ontology is starting to run together as the fields over specialize and don’t talk to each other.
Did mathematicians ever make experiments? 😉
Ask The Romans.
@@DonDee123 I would say proving something is an experiment but it can only have definite outcomes yes or no. There is no significance testing required. Other experimental methods EMERGE from mathematical concepts, tho, lol. So, I guess it's good, that you don't need a significance associated to if the sqrt(2) is irrational are not.
@@MsSonali1980 Math has its axions, which can’t be proven. And everything else is derived from them. Math is the language of science, needed to analyse and interpret data from experiments. But math itself makes no experiment.
The way that I’ve seen emergence framed as “separation of scales” seems problematic to me. Ultimately, more simplifying models can be used effectively at larger scales because with increasing scale, less resolution tends to be necessary for usably high accuracy of analysis. But that doesn’t mean the scale “below” disappears, only that our resolution of imaging lacks perfect precision but remains workable.
But in some cases a low resolution of imaging remains workable in others not. I guess what they are trying to do is use this fact for a definition of emergence. Of course now you need to define what 'workable' means.
No one is saying the scale below disappears. But in some cases, it becomes irrelevant. All this will make more sense if you study complexity science.
You can think of the emergence between scales as the result of two statistical effects. Combinatorial effects create new behaviors at large scales while the law of large numbers flattens out the details of the level below. Both of these effects are only on average if they exist in a given setting
@@DeLambada As in, some system phenomena can be considered more “emergent” than others based on how effectively those systems can be described without any reference to lower, underlying scales of objects? So, “emergence” is more of an adjective than a noun? Which ultimately depends a lot on the system and the complexity of the constructs used for modeling. But, like, all of this boils down to the choices we use in describing systems and what level of precision / accuracy we want, which feels subjective in a way that it feels like we’re missing something on how these ideas properly connect to objective knowledge and science.
Well, to be fair, in most applications you don't need to take electron spin or ionization to make a program work, or as Sabine pointed out, in order to model orbital mechanics for time scales longer than human civilization.
Even a thing like tic tac toe works with only marks on some surface. It's whole and complete without taking any scale into account, just certain rules and bounds with very precise limits. Despite this it's still the product of the extremely complex quantum to meter scale object that is a human with a mind.
The beautiful, shimmering patterns in flocks of flying starlings are a glaring way to get the concept of emergence across. The patterns are the result of a few simple rules each bird follows in regards to other starlings flying close to it. A single starling by itself would give no indication whatsoever of the patterns that flocks exhibit -- the patterns are emergent. Here's a National Geographic video of them in action:
th-cam.com/video/V4f_1_r80RY/w-d-xo.html
Absolutely!
Great example!
In the traffic models someone could be having an EMERGENCY, but that does not matter for EMERGENT behaviour
In artificial neural networks, the activation dynamics of non-trainable variables is strongly coupled to the learning dynamics of trainable variables. During the activation pass, the boundary neurons (e.g., input neurons) are mapped to the bulk neurons (e.g., hidden neurons), and during the learning pass, both bulk and boundary neurons are mapped to changes in trainable variables (e.g., weights and biases). For example, in feed-forward neural networks, forward propagation is the activation pass and backward propagation is the learning pass. We show that a composition of the two maps establishes a duality map between a subspace of non-trainable boundary variables (e.g., dataset) and a tangent subspace of trainable variables (i.e., learning). In general, the dataset-learning duality is a complex non-linear map between high-dimensional spaces, but in a learning equilibrium, the problem can be linearized and reduced to many weakly coupled one-dimensional problems. We use the duality to study the emergence of criticality, or the power-law distributions of fluctuations of the trainable variables. In particular, we show that criticality can emerge in the learning system even from the dataset in a non-critical state, and that the power-law distribution can be modified by changing either the activation function or the loss function.
An every day´s pleasure, to follow Sabine´s thoughts. Astonishing how much content she compresses in five minutes. This one I have to watch more than one time. 😊 (This is a bit like "Mehr als nur Atome")
I have had a hypothesis that gravity is emergent. Rather than a fundamental force with an associated particle, it's a perceived force much like a centrifugal force. However, being a lowly youtube, layman commenter, I don't have the resources to test such a hypothesis. Also, I could be way off in my understanding of particle physics as well...🤷♀
Lmao
@@rayparent1 I would appreciate being corrected, as I said, I am a layman. More productive than just laughing.
@@thisisbriannaandrea Nothing to laugh about, some others had similar thoughts too. But of course there´s a lot of research on quantum gravity.
@@thisisbriannaandrea most people that talk like you in my experience think they know better. I laugh because you talk about testing without being intimately aware of the mathematical structure of our theories. Which sparks my crackpot detector.
Maybe ive spent to much time talking to flat earthers and you were genuine. If so this is an open area of research
@rayparent1 I agree! I don't know better which is why I stated it as hypothesis given my limited understanding of particle physics. I don't claim it to be definitive. And unlike flerfers, I am open to new information, evidence and experimentation. And will revise or toss out my hypothesis based on new information.
New???!!! That’s Stephen Wolfram “New Kind of Science” book that was published 20 years ago!! 😢😢😢
Wonderful! I need more :) This emergence is a reason that some scientists don't support superdeterminism, claiming that sum is different than its parts (recently Kurzgesagt made a video about it). It is also a heart of entropy and I have a hunch that going deeper into this phenomena would give us answers to many scientific questions. and many philosophical ones. Thanks for the video, Sabine :)
Individual drivers tend towards the average driver as the amount of drivers increase. You can calculate information about a canister of gas without knowing something about every single particle inside the canister, precisely because you have so many of them that the statistical information caused by numbers dominates whatever individual differences these particles might have.
But this is precisely the emergence of the next layer. Every layer tends to be the limit (mathematical definition) of the summation of something in the lower layer.
Sounds like this paper has connections to the ideas in Wolfram's book "The Second Law."
indeed, but what happened to rule 30? Not detected by the new algorithm?
The Second Law was explained in a perfect way by Arieh Ben Naim in his book series about Entropy ( 1) Information Theory, 2) A Farewell to Entropy, 3) Entropy Demystified, 4) Information, Entropy, Life and the Universe)
I would go further to say that LLM processing power allow s for reduction in lumpability. And every decrease in lumpability corresponds with an increase in predictive power. Delumping means moving from word lumps to granular 1 dimensional scales. Then those scales can be combined in increasingly complex ways to allow for multidimensional analyses that increase predictive power AND computational efficiency. Multidimensional gradient scales > lumpability.
Keep in mind computer programming also happens in lumps. Most notably the binary lump. 0 or 1 instead of 0 through 1. True or false instead of 0 through 1 true and 0 through 1 false.
Ya gots to get your own show -- you just lumped too much in two paragraphs, and my head feels bumped.
@@pnf197 😅👍
@@Thomas-yl8lb LLMs say: " 'Fart' goes well with 'brain' "
Are you having it?
I agree, and very good insight here.
@@pnf197 Doing a show is a lot of effort. I can respect the heck out of Sabine for doing it for us, but I couldn’t do it myself. 🤣
I’m not big on these more recent videos not ending with a strong conclusion and transitioning directly into a sponsor segment. We need the takeaway message to be stated in a way that we recognize it as the takeaway message, not as a transitional sentence into a sponsor message.
1) pay for content if you don’t want advertising - of course not! I’d also want knowledge & understanding to be free. Unfortunately our societies don’t value that. They value endless consumption & greed for some & grueling drudgery of work - if lucky, for the rest of us.
2) the conclusion was clearly stated twice in the beginning. Make the effort to rewind if you’ve already forgotten it.
What I can't wrap my head around is the relationship between scale and emergence. if we have layers of systems built on each other, that implies that scale in this regard is inherently discrete, no? What dictates that boundary layer? like that transition between states of matter, or that balancing of order and chaos on a knife's edge in chaos theory... There's something eerily fascinating about that transition to order, that moment of self-organization in such eerily beautiful and coordinated ways.
You should watch Wolfram's New Kind of Science series he has on TH-cam.
He shows, with proofs of exhaustion, that rules typically emulate the behaviour of other rules. That when a rule is computing something at one scale, you can course grain it (blur it, or zoom out to a larger scale) and see it computing something that seems completely different (and you can test that for yourself by running cellular automata like game of life, and blurring it to see what behaviour it emulates. Hint : It no longer looks like game of life) In extension, it is this ability for rules to emulate one another that implies Computational Equivelence : That all systems are capable of emulating each other, and the maximal ceiling of complexity for this emulation, is that of a Turing Universal Machine.
Here is an example : Take an image of white noise. You can look at this image up close and see that its made of little random black and white pixels. But if you "zoomed out" or blurred the image (called course graining) then the noise becomes a single uniform color.
This is a visual statement that Randomness and Uniformity are the same thing...equivalent to each other...as a function of scale. This being true, is an interesting fact. The same kind of thing with light : White light is made out of all wavelengths of light. a single color, being comprised of a near infinite number of colors, is again, another interesting fact that scale expresses equivalences between what are two opposite things (randomness and uniformity, the continuous and discrete, infinity and one-ness)
Take another set of transformations : If we have in front of us an unshaded cube, from your angle, looking at this cube head-on, you see a square, but from my perspective which is looking at the same cube from a 45 degree angle...i see a rectangle. Both of us seeing different shapes...is an interesting fact. How can it be that we see different shapes when describing what is precisely the same object. Building off of this type of example, you can take a random distribution of dots and fill the cube. In most directions where you rotate this cube, it just appears random to us, but rotate it exactly 90 degrees and almost like magic, you find that it aligns in a patterned lattice. Again, it's another interesting fact, that this object can "emulate" the behavior of some other object, (randomness, to a pattern) under a group transformation (rotation)...a mere change in perspective...
And that's the clue : What dictates the "boundary layer" is perspective...the observer. Why does this system do that thing at that scale, and this other thing at this other scale? It's because WE impose a perspective on the system at a particular scale. At a smaller scale...we wouldn't say it has these properties...where at others, we might say that it does (because at a different scale it's emulating the behavior of something else) for instance a molecule sized bacteria can not hear sound, or understand what it is...because sound is the result of many molecules moving up and down in a certain pattern. Therefor this bacteria, would not have a theory for sound. It would have a completely different description of "physics" as a creature of that size.
Again...this is an interesting fact, that we impose a lot of the properties of systems to be what they are, as a result of trying to describe it. And this makes sense in the work of Stephan Wolfram, where all these systems are equivalent (like a single abstract object), and we simply "sample" or perceive this complex object in a particular way, because of the characteristics of our scale, our senses etc.
I hope this kinda long paragraph, addresses in part your question. Cheers,
@@NightmareCourtPictures thanks for the response. I can see how your examples relate to scale- but it seems to me they are more about describing the effects that perspective has on our ability to quantize this than the strangness inherent in emergent properties that seem to have no specific or linerar connection to its constituient scales- you can say that the idea of sound is irrelevant to something at the microscopic scale, but what about something like flow- it's an emergent behavior that doesn't feel like it has some sort of natural gradual transition- at some point there isn't flow, and then there is, no? or consciousness perhaps as a better example- it seems there's a specific threshold at which the sum becomes more than the parts. I think your answer explains part of this through perspective, but I feel like there's something else going on too, as if there's a metaphysical component we haven't quite figured out.
@@drewjaques7437 sound is very much like “flow” which I assume you mean like the direction of waves in water. Sound waves…water waves…basically the same construct: water is constituted by collections of molecules. If you were the size of a molecule you wouldn’t be able to perceive the flow of water.
Better way to visualize this: imagine your at a bowling alley. People are rolling balls generally speaking in the same direction. We humans wouldn’t describe this as flow…but in aggregate bowling balls in a bowling alley is flowing, in a single direction. We don’t normally describe bowling balls being rolled down a bowling alley as a liquid flow because the bowling balls is roughly our size and we don’t perceive enough of the aggregate behavior to think of it as a flow. So meaningfully, the description of flow among a collection of discrete objects is a reflection of our perception, a perception imposed by our size relative to the bowling balls.
In the wolfram model, the underlying metaphysics for why it is like this is because systems follow rules. The rules of the bowling alley: roll the ball and hit the pins. The resulting dynamics of many bowling balls rolling down an alley results in a larger scale dynamics that you can consider as something like a laminar flow. In extension all systems are following rules…and all those rules create behaviors that can then be extracted as physics depending on the perspective of the observer of those dynamics. If your a large observer you describe the balls in laminar flow…as a normal human observer you describe it as a bowling alley…as a tiny observer you describe it as molecules moving around randomly. Notice how at each scale the properties of the same system take on radically different behaviors, again based on who is doing the observation.
@@NightmareCourtPictures Ok, went and watched some wolfram vidoes. I see what yall are saying, but there still seems to be a flaw in our understanding for things like emergent behaviors that are radically different from the constituent behaviors. He talks at the edges of this in a few places, how simple rules can create complex patterns, but the problem of self definition and equivalent operations seems wrong instinctually- he talks about how our mathematics are really just a bunch of scaffolding add-ons to the maths of geometry and algebra, and i think he's right in that the divisions of specialities in sciences, the fields and tracks are really limiting our ability to create axiomatic rules that work better. The limit on computational models and similar issues like our inability to analyze/understand our own consciousness seems at odds with the idea of the simple to complex rule- as if there's something missing. So yea i can see the argument that in most cases where the scalar differences seem to result in systems/mechanics is really a matter of perspective, but that doesn't really address the root of some of these other issues. For example, let's ignore consciousness. What about matter at the nanoscopic scale? An electron isn't a physical object, it has absolutely nothing in common with our understanding of everyday matter- it doesn't have a size or a shape, it's a probabilistic wave. I don't see how we get from that to macroscale naturally.
Speaking as a research mathematician (PhD 1979), this is the clearest, and most correct and complete explanation of emergence not requiring advanced mathematics I have ever encountered. (Of course, we expect that from Sabina.) I was also a computer science professor, and I find it interesting that systems analysts working on very large computer programs have to "offload" some of the "bug detection" onto the end users, because they are emergent.
"Lumpability" is my new favorite word.
Also, Stephen Wolfram (the Mathematica guy) got a start on the cellular automata side of this in his 2002 book, _A New Kind of Science._ it's a fascinating read, and now available for free on line as a PDF. (I spent like $60 on it some years ago, but who knew?)
Complexity science is older than that.
I have a well founded maths & physics background with specialization in IT and sincerely, I have read it (A new kind of science) and I've got nothing new, it is extremely boring and : OK, complexity could emerge from simple rules, Boltzmann knew it! He showed how Entropy could emerge from the atomistic structure level, Claude Shannon's Theory of Information was the key to understand it. So what with Wolfram? Now they have discovered a new term "lumpability" when IT has the more proper "encapsulation" to hide complexities when you don't need them
@@DanielMartinez-ss5co Boltzman operated in adiabatic systems, Wolfram put the problem in a wide context of any system.
@@DanielMartinez-ss5co I spent my entire 40+ year career in I.T. and software development, so I'm qualified to say that surely your credentials prove that Wolfram is full of shit.
Right.
"Boring" is a personal assessment, not a universal one. I found the book intelligent and engaging, but my physics background is merely that of more than 30 years as a passionate layman, so what do I know?
Gee, you win.
@@jan.kowalski You should read Arieh Ben Naim books about Entropy, not Wolfram that is “Words, Words, Words”
Great topic! Thanks for making me aware of this paper. This is something I have worked on and I'll take a close look at this paper.
Fascinating. According to Wikipedia (I know, I know…), the term “lumpability” has been around since 1976.
It originates from the area of modeling and simulation of systems, a field I briefly taught to future teachers. However, we used methods and terminology from the former Czechoslovakia, which was “a bit” different, so I never heard this term or any similar ones before.
I guess one learns something new every day…
I defer to Wolfram on anything in this domain…
I'm sure one of the pictures was Rule 30 ie a non-repeating aka random emergence out of a few simple rules.
I'm a little surprised that an even simpler system was not postulated than 25 physical units (itself almost certainly a limitation of current knowledge) and starting with a more logic/rule based starting position.
@@commentarytalk1446 Yes, that was Rule 30.
More on this please!
(I’m not a computer scientist, but I think computation is more fundamental than particle physics)
John Conway's Game of Life is an incredible tool that illustrates this concept. He wrote a lot about emergence and synergy and self-organizing systems like cellular automata. The Game of Life is a cellular automata "game".
Be nice, Sabine! It's not"lumpable", traffic flow just has big bones, okay???
So, they've just discovered Conway's game of Life? Bless...
Wolframs ' a new kind of science ' came out about 30 years ago and is by far the most thorough investigation into cellular automata
I bet you they are already familiar with this
@@xmathmanx Yes. Also, Wolfram was a team leader in assorted Darpa projects to generate new Quantum Computation & AI technologies. Seems like those were successful. NKS is an active thing. Perhaps someone needs to write another book on this topic.
@@energyscholar I never heard of wolframs association with DARPA and it doesn't sound likely to me
@@xmathmanx Of course you've not heard about it! It's covered by NDA! I'm sharing inside information right now.
His first Darpa project was in the late 1980s, when one of his CA rules were used as the cryptographic basis for next-generation NATO radio communication.
Later he lead a Darpa project starting in late 1990 to build a new technology to crack Public Key Cryptography. Other project scientists included Stuart Kauffman, Murray Gel-Mann, & Brosl Hasslacher. Gel-Mann had to withdraw for health reasons and was replaced by a young Aussie super-genius. This project SUCCEEDED beyond wildest expectations and was HEAVILY CLASSIFIED. It's created a new form of Quantum Neural Computing that has lead directly to the current breakthroughs in AI.
This is history-making stuff that's not documented publicly.
I could go on but will stop with those two.
I've had the privilege of interviewing quite a few people connected with these projects. Much of the work took place at the New Mexico Center for Complexity.
Lol, "Doesn't sound likely to me". Funny!
His particular special talent, beyond innate brilliance, is how he enhances the effective intelligence of those with whom he works.
Huh, cyber sec researcher here, and this is of great interest for what I get up to.
With the tcp/ip stack we’ve taken this idea of separation of ideas as an explicit rule of network architecture, ie you don’t need the hardware layer to understand the application layer, or vise versa
It’s interesting since we haven’t updated it in a while too, with emergent cloud computing and edge functions, it would be an interesting branch of research to design emergent computing concepts beyond our current explicit stack
In the WAD file format, textures are stored in so-called "lumps". WAD stands for "Where is All the Data?". (Back then as now, textures are the most disk space-consuming assets in a video game)
5:45 Wilma Flintstone would agree
I see what you did there...
I've always like the idea that even the fundamental particles and forces emerge out of a much deeper multidimensional cellular automaton in a manner similar to John Conway's "Game of Life". The extremely simple rules for how a grid of on and off pixels evolves from one state to the next is endlessly fascinating.
It's not just that working on one level of emergence renders low levels irrelevant -- it would in fact be _counterproductive_ if not impossible to try and regulate the flow of traffic in a city or perform a heart bypass in terms of particle physics. Levels of emergence that are not sufficently relevant to the task at hand become meaningless obstructions.
I am not sure that we have a lot of evidence for emergence as a principle. I think small building blocks could cause "complexity" to emerge, but it could also be reversed from all we know.
The key question is, when you plot different resolutions like in the video, ftozen at one point in time, and then start evolving them, are all resolutions evolving simultaneously or not? If they are, it's hard to put a direction that must be implied by causality.
I don't know what you mean by a resolution evolving, sorry, can you explain?
The separate resolutions aren't different realities that are evolving in unison, it's all one reality that evolves, we just have different descriptions as a matter of convenience. I don't think there's a good case for downward causation, which seems to be what you're implying. If moving water molecules are one description, and waves are another description, I think it's clear that all that really exists is the atoms, and waves are a higher order concept that we use because motion of waves is what we perceive, but I think it's a mistake to say that because the atoms and the waves evolve simultaneously, that causation can go the other way and the waves could cause the atoms.
@@generichuman_ yep, make it flat, make it into dus, and then try explaining. usual positivist crap
@@SabineHossenfelder Yes, I might have hyper-focused on a very specific part of the video, sorry :D. I meant if we look at the representation at 1:37, which was very interesting to me, we see different resolutions of "the world", so to say. These are still representations of the same "world" and in the same space, and (I think) same time - i.e. if we somehow start evolving the "world" over time according to some laws, all of the resolutions should evolve simultaneously.
So if that is true, it seems hard to say if the direction of causality goes from deepest resolution upward, ex from particle physics to the solar system, or the other way around. It seems like it could be the other way too, for all I can tell.
Though it might be only me, I'm a data scientist, not physicist. :D
@@aleksandarjovanovic9080Surely, if you are a data scientist, the laws of physics are just data compression from data about the universe :-) I think you said "particle physics" when you really meant "particles". Physics is the study of reality, not reality itself.
For fear of sounding like an actor, going back to your videos on free will, I've had this idea for a while that consciousness is a result of localized areas of complexity. The metaphor I use when explaining it to other people is you think of a string as the universe, then you tie a knot in it, that knot is a mind. It's still part of the string, but it's also a distinct object. Simpler creatures are less complex, so they have a lower level of consciousness, until you get down to creatures that are basically just organic machines. I figure that rather than living creatures having some kind of divine pleroma that gives them thought, it might just be an inherent property of the universe.
I'm going to go out on a limb and say that this was actually your most important video. I think there's a good chance that this video's impact, at least in it's second order effects, will probably have the biggest impact of all the videos you've created.
I bet this was a relief for a hobbiest ❤
Wow, this sounds closer to Wolfram Physics... Before we know it, Sabine will be advocating we replace String Theory with Wolfram Physics. I am just teasing, of course, or am I? 😏 Seriously though, I would love to hear Sabine pontificate on Wolfram Physics: the good, the bad, and the ugly...
It's called complexity science. Wolfram has contributed to it.
Well she made fun of him in the song video "theories of everthing"...
@@Thomas-gk42 Sabine making fun of someone... who would have thought that possible? 🤔
@@EricKolotyluk hihi..
It certainly makes sense that regardless of the scale, all scales are governed by to the same relationship to energy, and, regardless of the scale, particles seek out combinations for relative stability at the cost of individuality and thus grow ever larger and more complex in an ongoing struggle with entropy.
This is 101 of computer sciences (CS). Really nice of her to use it as an example, since the whole field of CS is 100% based on this. In programming specifically, we use high-low level languages to describe the level of this abstraction (not lumbability). For example we call Python and JAVA high level (of abstraction) programming language and C and Assembly low level (of abstraction) programming languages. The low and high refer only to the level of abstraction, not to the quality, power or difficulty of the languages (like many new and non-programmers often think)
However, I don't see how anything in this is in anyway new or could give us a leg up, since it is a very well understood and "abused" phenomena. For example, neural network AIs are based on exactly this emergent phenomena that we have used since 1943
I have studied the Physical and Natural Sciences for over 60 years & can't remember ever having enjoyed presentations more. Your brief offerings are more like being immersed in a thought provoking song than a dry didactic lecture. Nicely done as always Bee.
I paused the video half way through and tried to imagine where it would end. So if the idea is to mathematically relate the micro, standard and macro, then that deals with transposing elements and forces from one to another -- 4 possibilities at first, later 6 perhaps. The things that immediately came to mind were teleportation (and entanglement) and faster than light travel (and wormholes). When like elements or processes or forces [I'm not sure where I'm going with this] get grouped they get scaled up to the next level; when broken down you end up at the next smaller level. Assuming you're an expert at some part of this, there must be nearly infinite practical applications (i.e.: some things move faster than light [or otherwise ignore that speed limit], most don't). Thank you, Sabine, for a thought-provoking segment. The above may not be new, but it's new to me, and pretty cool.
I'd use this kind of analysis to define what is an "observer".
For example, measuring the spin of a particule causes a great change on your behavior as a complex system, then you are observing that spin.
If it does not, then that spin is part of your complex system and counts as your own internal workings or even "noise".
But I am just saying things here.
Interesting idea!
I looked up about lumpability, this word has been around in pure math at least since 1984 (Pamela G. Coxson, Lumpability and Observability of Linear Systems, Journal of Mathematical Analysis and Applications 99).
Well, the way I think of it, each level of details can be explained by probabilistic estimation of the previous level.
I’m a physicist doing a master degree in computer science exactly because of this kind of ideas. You can’t imagine how exited all this recent development is 😍😍😍
LUMPABILITY = integration or summation of states, dimensions -- a syntropic process!
Complexity is dual to simplicity.
Micro is dual to macro.
Increasing the number of dimensions or states is an entropic process -- differentiation or reductionism.
Decreasing the number of states or dimensions is a syntropic process -- integration or holism, LUMPABILITY.
Increasing (divergence, entropy) is dual decreasing (syntropy, convergence).
Syntropy (prediction) is dual to increasing entropy -- the 4th law of thermodynamics!
Teleological physics (syntropy) is dual to non teleological physics (entropy) -- physics is dual.
Integration is dual to differentiation.
Reductionism is dual to holism.
Homology (syntropy) is dual to co-homology (entropy).
"Always two there are" -- Yoda.
What theoretical connections can be found (?) between > hierarchical emergence< as mentioned above and the work of Karl Ludwig von Bertalanffy 1901-1972 , Frederic Vester 1925-2003 and others, which is called Systemtheory .
Kinda reminds me of Isac Asimovs Foundation, in it a mathematician found a way to predict the future - not the future of a single human - but the future of 'society'. the basic idea is that while single persons may behave unpredicably, on a larger scale a whole society moves based on rules and therfore can be predicted.
The last description around automata should mention the author of this concept, Stephen Wolfram, who wrote a very interesting first book, A New Kind of Science.
I feel like at higher levels of complexity it’s harder to trace the ways structures emerge. I’m thinking about the way universal cultural norms emerge based on the typical ways that people think. There are some people whose minds are less suited to act as pieces of that greater structure, maybe a business or a religion, sometimes because they are more aware of the minute details at a lower level of complexity or of the abstract structures emerging over the webs our society is weaving at this moment. That atypical connectedness between levels of complexity reminds me a bit of like… evolution because there was the normal way that the sequence was “supposed” to be replicated (and at the higher epigenetic level, regulated), but the imperfection of the process is part of what causes speciation and diversity, allowing for more complexity. But, maybe there’s more to it than the genetics comparison (and actually maybe there is more to genetics than we understand) and, instead it’s nice to think that there could be structures themselves emerging from (or connected to) the more detailed and more abstract thinkers, and not just think of them in relation to the obviois structures of society.
I’ve spent a lot of time trying to model the impact of radiation effects in semiconductors on spacecraft systems. This is a good example of complexity and scalability, because the radiation cause, an energetic ion or photon, collides with the atoms in a semiconductor crystal. This in turn results in liberated charge, which can change the state of a bit, or create an unwanted pulse in continuous signal, which can affect propagate through software, which can affect more hardware, such as a star trackers, which might cause a satellite to lose track of where it is in space. Of course we do everything we can to arrest the propagation of these effects, but prediction is not always possible because of the complexity of the system and all the scales involved. I think I’m going to read this paper and take a look at Wolfram’s book.
on bigger scales, our own world could be perceived as quantum
This may sound weird but lumpability sounds a bit like when tiny waves cancel out as very small timescales and reveal large emergent swells below the noise.
The question then is... is the swell emerging from the little waves (probably), or is it an independent process revealed when you ignore the tiny waves (which would be a bit nuts).
Most of what this video touches on has been around for a long time. Wolfram published "A New Kind of Science" in 2002 (the origin of the diagram in this video's thumbnail), and critics pointed out that the book was a variant on older ideas, and not as new and original as Wolfram claimed. The origins of Chaos Theory, Complexity Theory, the Butterfly Effect, and Emergence date back to the 19th century. Cellular Automata (e.g. thumbnail diagram) date back to the 1940's, and went mainstream in the 1970's with Conway's game of Life. The concept of Lumpability dates back to the 1960's.
And the critical flaw remains, sense can't come from nonsense. Garbage in garbage out.
Now that we can objectively measure lumpability, we’re one step closer to proving or disproving the conjecture that, for any given type of behavior (or piece of furniture) ‘you can like it or you can lump it’.
I think being able to say "lumpable" or "lumpability" now will definitely enrich discussions. FWIW electrons are very lumpable with each others and also with protons within the causal closure of electromagnetics.
Wolfram did good work on the cellular automata. He did this awhile ago. "A New Kind of Science" is the book.
Studying only elementary particles and their interactions, we can't have any idea of a cumulative effect that pulls them together when there are in a huge number, and that we call gravity. Indeed gravity isn't included in the standard model, and attempts to do so all failed. This and plausibly many other reasons show that extrapolation is not valid in science. Physics is an experimental science, and complex systems as well as simple ones can only be studied experimentally. This is equally true for all other disciplines, and especially about life. We can't claim to understand life in terms of elementary particles and their interaction, with or without emergence. Even if we have 13 decimals of accuracy in the infinitely small, it can amount to total inaccuracy in the infinitely large.
Man it sure feels like Douglas R Hofstadter touched upon this in 1979, at least came close. He talked about levels, popping and pushing up and down between them.
In the computer simulation of two-dimensional fluid mechanics we can model the flow as a cloud of point vortices and add some Brownian motion to represent viscosity, as proposed by Alexandre Chorin. In addition we could add a rule that the vortices are quantised, with the parameter of quantisation being equal to the parameter of Brownian motion (the dimensions are the same). So here we have a computer simulation of wave-particle duality which runs in polynomial time. Surely it can be done?
Unfortunately grown-up quantum mechanics for many-particle systems requires a configuration space of at least as many dimensions as there are particles, and normally three times as many. Any computer simulation of the collapse of the wave function is likely to require a computer which can cope with exponential-time algorithms in order to describe the behaviour of any detector. Your reference to cellular automata does not distinguish between polynomial time and exponential time. I think you could be in for disappointment.
No reason why we shouldn't pretend that we have such a computer and think about how we can programme it. I would propose adding tachyonic Brownian motion, but I am sure there are other ideas. The simplest systems involving a collapse of the wave function are likely to be quite complicated and beyond our abilities. Two molecules of nitrogen tri-iodide are one such system. There's simply no way anybody can show that I am either right or wrong until we have that computer. Frustrating !
Emergence is just a change of description, not a change of state.
Molecules don't stop being a collection of interacting quarks and leptons, it's just that on molecular scales the quark and lepton picture is just an impractical description.
In the halcyon days of discussions on Quantum Gravity and String Theory, there was optimism that a Theory of Everything could be postulated, but the problem emerged that gravity, the force which is so essential in shaping the universe at the astrological scale, completely vanishes in the quantum domain as though we were suddenly talking about two different and unrelated universes. But by some miracle, both Quantum Mechanics and General Relativity accurately describe the same universe on their own terms.
This problem exists across every area of science in the paradox that we have a coherent image of how we think the natural world works, but it's a patchwork quilt of different fields any two of which which may or may not intersect with each other. The theory of everything was supposed to take this mosaic and fuse all the bits and peices together. Extensive attempts to do this have created the impression that this simply isn't possible and there is in fact something keeping all of the different fields seperate aswell as continuous with each other.
Emergence is the new model to explain how Quantum Physics and General Relativity describe the same universe, despite all apperance to the contrary.
Just been to a business training, an instructor told us to analyse population changes and movements to predict future business opportunities. Very similar concept of today’s topic, in my opinion.
5:38 is my favorite statement Sabine has ever made 😂
As far as computer algorithms go, I think that to understand the flow of a program you really must understand computer architecture. Micro architecture, not so much, but arguably yes you do have to understand the computer lock, stock and barrel in order to make the most out of the programs that you write. With Python it isn't as true, but the more powerful the language is the more you need to know about what it is being compiled on, and how.
It is hard not to think that something in the nature of components predisposes them to complexity and that those features of the building blocks are as yet unknown. Examining a brick should yield the probability of a building (if you subtract any human purposes or goals that led to the brick in the first place).
I'm just a lump. But I think when we speak about space and reality, the rules/laws themselves are emergent. This is because what we really have are patterns. Patterns that manifest consistently, we call laws....But what if they change?
This is quite fascinating. It's like trying to understand something that appears to exist for the sole reason of not being understood.
As presented, the "levels of lumping" occur on an axis of physical+numerical scale. As a slightly (ex-) comp-sci guy, the word "Abstraction" is strong in my head. In that context, there need not be a single (universal) "upwards/bigger/better" direction, there may be several, each leading towards a different kind of specialisation (e.g. class of observation or use-case). Like different Views on a database - especially if they are complex (e.g. state-machine driven/interpreted) ones. Neurons vs neural network also comes to mind (though still a kind of "bigger"). Then there's Mandelbrot's various creations (e.g. his famous figures) and of course fractals (rules => types => dimensionalities).
Applying this (more generic) perspective of abstraction back to physics, what _other_ kinds of lumping are recognised or postulated in such observational philosophies?
Like do lumping-particles always have to be physically/geographically close (relative to other such particles), as with planets, or could they just all be in random locations but "on the same wavelength" in some fashion. For example different matter with different absorption spectra in an energetic nebula. Human communications comes to mind (internet and otherwise). Viruses (alone or interacting with others or other stuff) could be another. Same for mankind's historical and current _memes_ (of the pre-social-media kind).
To be honest, as someone who thinks about code, not specificly the computer, and grew up with coding since my teens, and being able to understand even hard parts of my field by now, I think in the world of programming, the solid world view of math meets the flow of logic and the intuition of art. This makes it natural, that "emergence" emerges in that field. The hard part is, that coming from other fields, usually programming is simply seen as something formal, which is why there are so many different types of programmers. So yes, I believe those "computer scientists" are right, as we are only still at the beginning of understanding that science itself, and often forget, that the world of virtual logic is by itself interesting, if we only think about the limitations of our current hardware.
It is also quite interesting, that many of the programming paradigms and patterns actually arise from limitations or rules of the underlying programming language or hardware itself.
And in the end, simulating large systems, like writing a simulator for gravity, and trying to make that simulation realistic but at the same time fast enough to be only part of your program, gives you great respect of the numbers involved. I am also kind of amazed, that none of what you said in this video surprises me after 30 years of working with simulations or games
Is there a relationship between “emergence” and entropy?
I always figured emergence comes from organization, it's about the arrangement of the lower scale instead of its properties.
I'm not so much interested in complexity (yet, who knows once I actually learn about it), but I'm fascinated by emergent behavior. That is, unpredictable behavior that arises from a complex system. Like consciousness arising from trillions of connections constantly exchanging electrical and chemical signals, thus generating the entirety of your lifetime's experiences. All while working in tandem with an almost equally complex system (your body) which acts as a proxy with which to receive information that is turned into our experiences.
Great show. This is a sweet spot for you.
0:59 Simple to less simple? Following what plan ? Following what program?
Wolfram might have been mentioned.
There's a notion that since the Universe is relative, the scale itself is relative and there's just no smallest / largest scale - it's a continuum and that includes the laws most appropriate for each of the scales.
The authors consider a (discrete) stochastic process and study whether there is a coarse graining of states that leads to a _deterministic_ dynamic. This is easiest to decide in the case of a Markov process and reduced to this case by replacing the process with the process of increasing histories. The interesting take is the following: you can have a nontrivial deterministic macroscopic evolution driven by a stochastic microscopic one. This may explain the existence of deterministic laws of physics in the macroscopic domain when there are no such laws on the subatomic level. However this all depends on the process of coarse graining which is _subjective_ (what distinctions do you think are important?) So the negation free will rest entirely on how coarse you willing to see the world.
Not entirely. Emergence is a process where you loose or transfer out some information. This has to have a thermodynamic effects. On different scales even our current world (in "out" scale") could be seen as "quantum".
What's the difference between "lumpability" or "causal closure" and generalization?
A form of scaling emergence is the connection of electrons to gravity:
emev=.51099895631
Gn=6.674202636*(10^-11)
..the scaling logarithm in base ten leads to this effect:
log(logGn+12)=emev^2
The topics you're discussing are as interesting as ever, however I miss those longer in-depth videos.
3:30 shots fired
I am not sure what exactly the problem is. There are several ways to make properties disappear (for example randomness, small forces & energies, mechanical limitations, boundaries). And new properties can appear either because some disappear or because you created new mechanical actions.
What would be nice is a measurement like entropy.
The oft-asserted "independence" of higher level "laws" seems to me simply an artifact of higher error tolerances at the given scale. These laws are always approximative, and the phenomena they model could in theory be modeled better by more fundamental laws; it's just that doing so would be computationally onerous and the difference in accuracy would be trivial for most of intended applications. But there are always points and places at which the higher-level "laws" break down. We can model fluids as continuous media using the Navier-Stokes equations for many purposes, but not when the mean free path length is large or the representative scale sufficiently small. This is because, of course, fluids are not *actually* continuous media. The higher level "laws" make ontological commitments that ultimately aren't true.