Tuesday, September 06, 2016

Sorry, the universe wasn’t made for you

Last month, game reviewers were all over No Man’s Sky, a new space adventure launched to much press attention. Unlike previous video games, this one calculates players’ environments from scratch rather than revealing hand-crafted landscapes and creatures. The calculations populate No Man’s Sky’s virtual universe with about 1019 planets, all with different flora and fauna – at least that’s what we’re told, not like anyone actually checked. That seems a giganourmous number but is still less than there’s planets in the actual universe, estimated at roughly 1024.



User’s expectations of No Man’s Sky were high – and were highly disappointed. All the different planets, it turns out, still get a little repetitive with their limited set of options and features. It’s hard to code a universe as surprising as reality and run it on processors that occupy only a tiny fraction of that reality.

Theoretical physicists, meanwhile, have the opposite problem: The fictive universes they calculate are more surprising than they’d like them to be.

Having failed on their quest for a theory of everything, in the area of quantum gravity many theoretical physicists now accept that a unique theory can’t be derived from first principles. Instead, they believe, additional requirements must be used to select the theory that actually describes the universe we observe. That, of course, is what we’ve always done to develop theories – the additional requirements being empirical adequacy.

The new twist is that many of these physicists think the missing observational input is the existence of life in our universe. I hope you just raised an eyebrow or two because physicists don’t normally have much business with “life.” And indeed, they usually only speak about preconditions of life, such as atoms and molecules. But that the sought-after theory must be rich enough to give rise to complex structures has become the most popular selection principle.

Known as “anthropic principle” this argument allows physicists to discard all theories that can’t produce sentient observers on the rationale that we don’t inhabit a universe that lacks them. One could of course instead just discard all theories with parameters that don’t match the measured values, but that would be so last century.

The anthropic principle is often brought up in combination with the multiverse, but logically it’s a separate argument. The anthropic principle – that our theories must be compatible with the existence of life in our universe – is an observational requirement that can lead to constraints on the parameters of a theory. This requirement must be fulfilled whether or not universes for different parameters actually exist. In the multiverse, however, the anthropic principle is supposedly the only criterion by which to select the theory for our universe, at least in terms of probability so that we are likely to find ourselves here. Hence the two are often discussed together.

Anthropic selection had a promising start with Weinberg’s prescient estimate for the cosmological constant. But the anthropic princple hasn’t solved the problem it was meant to solve, because it does not single out one unique theory either. This has been known at least since a decade, but the myth that our universe is “finetuned for life” still hasn’t died.

The general argument against the success of anthropic selection is that all evidence for the finetuning of our theories explores only a tiny space of all possible combinations of parameters. A typical argument for finetuning goes like this: If parameter X was only a tiny bit larger or smaller than the observed value, then atoms couldn’t exist or all stars would collapse or something similarly detrimental to the formation of large molecules. Hence, parameter X must have a certain value to high precision. However, these arguments for finetuning – of which there exist many – don’t take into account simultaneous changes in several parameters and are therefore inconclusive.

Importantly, besides this general argument there also exist explicit counterexamples. In the 2006 paper A Universe Without Weak Interactions, Harnik, Kribs, and Perez discussed a universe that seems capable of complex chemistry and yet has fundamental particles entirely different from our own. More recently, Abraham Loeb from Harvard argued that primitive forms of life might have been possible already in the early universe under circumstances very different from today’s. And a recent paper (ht Jacob Aron) adds another example:

    Stellar Helium Burning in Other Universes: A solution to the triple alpha fine-tuning problem
    By Fred C. Adams and Evan Grohs
    1608.04690 [astro-ph.CO]

In this work the authors show that some combinations of fundamental constants would actually make it easier for stars to form Carbon, an element often assumed to be essential for the development of life.

This is a fun paper because it extends on the work by Fred Hoyle, who was the first to use the anthropic principle to make a prediction (though some historians question whether that was his actual motivation). He understood that it’s difficult for stars to form heavy elements because the chain is broken in the first steps by Beryllium. Beryllium has atomic number 4, but the version that’s created in stellar nuclear fusion from Helium (with atomic number 2) is unstable and therefore can’t be used to build even heavier nuclei.

Hoyle suggested that the chain of nuclear fusion avoids Beryllium and instead goes from three Helium nuclei straight to carbon (with atomic number 6). Known as the triple-alpha process (because Helium nuclei are also referred to as alpha-particles), the chances of this happening are slim – unless the Helium merger hits a resonance of the Carbon nucleus. Which it does if the parameters are “just right.” Hoyle hence concluded that such a resonance must exist, and that was later experimentally confirmed.

Adams and Groh now point out that there are other sets of parameters altogether in which case Beryllium is just stable and the Carbon resonance doesn’t have to be finely tuned. In their paper, they do not deal with the fundamental constants that we normally use in the standard model – they instead discuss nuclear structure which has constants that are derived from the standard model constants, but are quite complicated functions thereof (if known at all). Still, they have basically invented a fictional universe that seems at least as capable of producing life as ours.

This study is hence another demonstration that a chemistry complex enough to support life can arise under circumstances that are not anything like the ones we experience today.

I find it amusing that many physicists believe the evolution of complexity is the exception rather than the rule. Maybe it’s because they mostly deal with simple systems, at equilibrium or close by equilibrium, with few particles, or with many particles of the same type – systems that the existing math can deal with.

It makes me wonder how many more fictional universes physicists will invent and write papers about before they bury the idea that anthropic selection can single out a unique theory. Fewer, I hope, than there are planets in No Man’s Sky.

38 comments:

akidbelle said...

Hi Sabine, thanks for the post; I think maybe we can address real issues.

Basically, I understand the argument for fine-tuning as follows:
1) We believe that the equations we use are appropriate to predict the observables - and we checked it.
2) Those equations come from a theory that can eat (almost?) any set of parameters.
3) We believe that those theories are fundamental (at least close enough).

=> Therefore we believe that the parameters of our theory are fine-tuned.
=> Therefore anyone saying the opposite is a crackpot (normal sociology).

In a similar manner, when the sorcerer of the tribe invokes some of the spirits, everyone around is frightened; we proved that those who are not frightened are burned by the tribe. Therefore the spirit exists and magically burns those infamous who are not frightened.

It seems to me that fine tuning (if not parameters freedom) is just a theory of ignorance, resulting from the very fact that gravitation has been stuck for 100 years now and the SM for about 40 (or maybe ore). Where am I wrong?

Note that I am trying to be a little aggressive to make sure I am understood; no offense meant to anyone.

Thanks,
J.

KC Lee said...



A helpful reminder to anthropos is that not only "Sorry, the universe wasn't made for you", it did just fine for tens of billions of years, without any humans around.

Related to the above, one way to view our "measurement problem" is that it could be self-inflicted, We turn two perfectly natural quantum systems interacting with each other (in environmental decoherence), into a human laboratory-induced quandary.

And if we are serious about classical physics emerging from quantum physics, the single first principle we need is that quantum physics itself tells us the universe is random.

If randomness = naturalness, out go all the "Naturalness", "Fine tuning" etc. problems.

The biggest obstacle to considering the above is, perhaps, the size of the human ego?

Just some random thoughts, on yet another excellent post.



Matthew Rapaport said...

True enough though none of it does anything to explain the actual values as they have come out.

KC Lee said...



Just in case Matthew's comment happens to relate to "fine tuning" etc., all actual values we have observed could be the result of random occurrence (if, a big if, one accepts the first principle that the universe is random).

It answers the fundamental question "Why is the universe this way, and not that way?"

This first principle could be way too simple. So simple that it sounds like avoiding the issues. Well understood. Still, these could be issues that humans have made up, in a sense.

Then again, perhaps it is more appropriate to please Rev. Occam, at the risk of putting a good many paychecks, at risk? Best,


KC

Uncle Al said...

"the missing observational input is the existence of life" Cads + fools = priests. "Weakless Universe" and standard physics exclude net matter over antimatter; dark matter is mysterious.

Spacetime is trace chiral anisotropic toward fermion quarks, then intrinsic chiral maldistributions (Sakharov conditions). Dark matter is Notherian leakage of angular momentum conservation. Opposite shoes non-identically embed within chiral spacetime. They follow non-identical minimum action vacuum free fall trajectories.

"Ο Θεός είναι γεωμέτρης": Geometric Eötvös experiment: Oppose single crystal (self-similar mass distribution) test masses of right-handed space group P3(1)21 and left-handed P3(2)21 alpha-quartz. 40 grams total contrasts 6.68×10^22 pairs of opposite shoes (0.113 nm³ enantiomorphic unit cells). Look.

KC Lee said...



"I find it amusing that many physicists believe the evolution of complexity is the exception rather than the rule."

Perhaps another reason could be:

As physicists, we know that entropy always increases (complexities decrease), with time.
But our very civilization is the result of the opposite. Even our being is the result of anti-entropy. Best,


KC



TheBigHenry said...

Matthew Rapaport,

Your comment seems true enough to me. But it also seems to me to be a non-sequitur.

Does Sabine's post presume to explain "actual values"? I think not, but I wouldn't presume to speak for her.

Sabine Hossenfelder said...

KC,

"As physicists, we know that entropy always increases (complexities decrease), with time."

I highly doubt this is correct and I am surprised to hear that you believe we "know" this. You have to be very careful in specifying the entropy of what you are referring to. We clearly know that the entropy of a system can locally decrease even while the entropy of the whole system goes up. This has nothing to do with "anti-entropy." Best,

B.

Sabine Hossenfelder said...

akidbelle,

I can't make sense of your three steps. The argument for finetuning comes about the way I explained. You take one of the parameters, ask what happens if it's somewhat larger or smaller, and then you find that something doesn't work anymore, presumably preventing chemistry as we know it.

Best,

B.

Sabine Hossenfelder said...

Matthew,

Right. And what I'm saying is that drawing upon the anthropic principle isn't going to change anything about this. Best,

B.

JimV said...

I've been making a layman's version of your argument (against fine-tuning) for a little while, but without much specificity, just speculation, e.g., that in a universe with different constants, protons might not be stable but some other baryons or mesons might be, allowing atoms of a different kind. From now on I'll just link to this post. Thank you.

KC Lee said...


Bee,


Thank you for pointing out the obvious error in my statement about entropy!

Not only does the statement apply only to a closed system, of which we have no firm proof our universe is one, it greatly simplifies situations (in addition to your kind reminder) where subsystems (in a many-body system for example) may or may not serve as its own "bath" etc etc.

"Anti-entropy" is from being lulled into waxing philosophical. Sorry!

I will remember to be careful. Best,


KC

akidbelle said...

Sabine,

in the 3 lines, the important word is "we believe". All this is based on the "belief" that known physics is fundamental and close to completion - history repeats, you know about the next generation and the next decimal to check... now it is the next mathematics that includes or extends the ones that work.

I can also explain differently. My argument is that fine-tuning is all about the belief in parameters "freedom". Do you know anything else really free? I don't.

But the existence of free parameters is a property of theory - not of nature. Proof is that you can chose many other sets of parameters to imagine a different universe and the theory will work safely. But you cannot check it experimentally. It has the same scientific values as the spirits of the sorcerer. You cannot observe it directly but you can see its effects in sociology.

Now assume exactly the opposite: only one set of parameters, symmetries, and particles listing can exist because it is constrained for some self-consistence reason (I don't even need to know how).

This assumption has exactly the same value as the opposite, just because we don't know. But its immediate consequences can be checked in the real world: Since physicists cannot even address the real problem, they publish many fairy tales.

Thanks for asking.
J.

Uncle Al said...

@KC Lee "we know that entropy always increases" Empirically wrong. All oscillating chemical reactions spontaneously reverse entropy. Large scales, too - the Belousov-Zhabotinsky reaction. Energetic systems with positive feedback (e.g., life) can be very naughty. Ergodic systems can cycle (closed trajectories in phase space; strange attractors).

https://www.youtube.com/watch?v=3JAqrRnKFHo
(Add ceric ammonium sulfate and mix in a beaker: bulk flipping red and green some 30 times, the Traffic Light)
https://www.youtube.com/watch?v=-DCkPN_FgOQ
Hand of God version.
https://www.youtube.com/watch?v=Z1kg2Yw0xgw
Liesegang rings re banded agate.

Physics in principle contains all chemistry. Make aspirin from benzene and propylene (one molecule each), and any needed inorganic reagents.

PhilG said...

Who claimed that "anthropic selection can single out a unique theory"? I don't see that as an element of anthropic reasoning or fine tuning arguments at all. If you want to destroy the anthropic argument this way you have to show that most of the space of possible parameters supports complex life. It certainly is not enough to show that there are some other points that support it.

Sabine Hossenfelder said...

PhilG,

I don't know who you think wants to "destroy the anthropic argument". The anthropic argument - that we live in a universe hospitable for life and our theories must be able to accommodate this - is of course correct. What I have said is wrong is the idea that our universe is finetuned for life. You don't have to show that most of the space of parameters support complex life, because most of that parameter space we'd never observe anyway - that's exactly what the anthropic principle does for you, getting rid of most of parameter space. You'd have to show that among the places that support life, we're not in a likely one. And to show that this isn't correct, a handful of counterexamples are sufficient.

And, yes, you are right, strictly speaking this is incorrect because you can't make a statement about probability without a measure. Invent a measure and you can make all and everything likely or unlikely. But this just means that the whole question is ill-posed to begin with. Either way, it's a wrong argument. Best,

B.

Phillip Helbig said...

"The general argument against the success of anthropic selection is that all evidence for the finetuning of our theories explores only a tiny space of all possible combinations of parameters. A typical argument for finetuning goes like this: If parameter X was only a tiny bit larger or smaller than the observed value, then atoms couldn’t exist or all stars would collapse or something similarly detrimental to the formation of large molecules. Hence, parameter X must have a certain value to high precision. However, these arguments for finetuning – of which there exist many – don’t take into account simultaneous changes in several parameters and are therefore inconclusive."

I've been saying this for years and am surprised that many people overlook this.

Phillip Helbig said...

A helpful reminder to anthropos is that not only "Sorry, the universe wasn't made for you", it did just fine for tens of billions of years, without any humans around.

This is indeed a good argument against overstating humans' influence (e.g. "someone must have collapsed the wave function of the universe"), but it doesn't rebut the anthropic principle because the anthropic principle (in any sensible form) doesn't say that the universe was made for humans (or intelligent life, or life) nor that humans (or life, or intelligent life) must arise.

Sabine Hossenfelder said...

akidbelle,

"Now assume exactly the opposite: only one set of parameters, symmetries, and particles listing can exist because it is constrained for some self-consistence reason (I don't even need to know how). "

That's not logically possible. It would mean that you can derive a theory based on self-consistency alone, but self-consistence isn't enough. You always need to draw upon observational input - physics does not equal math. In principle drawing upon the existence of life is just another observational input. I just find it bizarre that some physicists for some reason believe it's a better input as just taking the damn measured values of the free parameters. Best,

B.

Jo Andres said...

Can't we be more naively pragmatic (ignoring extended philosophical discussions about what the anthropic principle actually is)? Might the anthropic argument be true or not, we will never finally know. There ever might be a law behind finetuning rather then a (in some sense arbitrary) selection from random - what the anthropic argument basically is. The physicists of the late 19th century could have laid back and argue that the coincidence of 1/SQRT(EPSILON Zero times MU Zero) with the speed of light was just random and go for a cigar. Obviously, the Standard Model and all speculated extensions are so unsatisfying that we are far from the need to catch for a straw, as the anthropic argument is. Theory selection by the anthropic argument kills motivation to keep on searching for more.

Sabine Hossenfelder said...

Jo,

The anthropic principle is arguably correct, there's no doubt about this and no discussion about this. And I just told you that theory selection by the anthropic argument does *not* relieve you from having to search for other explanations, exactly because that selection does not single out a specific theory. Best,

B.

Theophanes Raptis said...

Imagine the horror though, if it was actually made for "some-body"/something else entirely!

http://www-history.mcs.st-and.ac.uk/Biographies/Fantappie.html

Arun said...

"One could of course instead just discard all theories with parameters that don’t match the measured values, but that would be so last century."

:) :) :) :)

Uncle Al said...

At the moment of particle-antiparticle 180° impact there are photons (lepton collision) or photons plus neutrinos (hadron collision) disassembling at lightspeed. Why should any other particle appear?

Detonate high explosive. Why should anything but small molecules (H2O, CO2, CO, NOx, N2) appear? Pop TNT/RDX underwater, obtain up to 90% carbon yield as detonation nanodiamond (~5 nm diameter).

http://www.nature.com/nnano/journal/v7/n1/fig_tab/nnano.2011.209_F1.html

Perhaps we should care less about the filling and more about the background, especially for the Big Bang.

Jo Andres said...

Think I got the point, I am just not sure if your post means in a strict and general manner that arguing with the anthropic principle is a waste of time and we should stop it at least in science. It is falsifiable, but just as a placeholder for better explanations and does not make verifiable predictions

KC Lee said...



Uncle Al,


Thank you for sharing three interesting videos on chemical reactions. Watching the first one on Belousov-Zhabotinsky already evoked the notion of creationism. The title of the second one reinforced that thought. Actually watching that video suggests, tongue-in-cheek (!), the Creator may suffer from Parkinsonism.

In appreciation of your 3 chemistry videos, please allow me to share 3 physics papers: two on partial thermalization [calling attention again to my careless mistake which Bee and you kindly pointed out] and the third one on chaos (classical) coinciding with increased entanglement (QM), over time.

Hope you enjoy the papers as much as I learned from your videos. Best,


KC

1) J.-y. Choi et al. Exploring the many-body localization transition in two dimensions, Science (2016). DOI: 10.1126/science.aaf8834
2) J. Smith et al. Many-body localization in a quantum simulator with programmable random disorder, Nature Physics (2016) doi:10.1038/nphys3783
3) C. Neill et al. Ergodic dynamics and thermalization in an isolated quantum system, Nature Physics (2016) doi:10.1038/nphys3830




Sabine Hossenfelder said...

Jo,

You say you got the point, but then you state that you think the anthropic principle is falsifiable. That's wrong. The anthropic principle is not a hypothesis. It's a requirement based on observations that have already been verified to extremely high precision. To falsify it you'd have to show that actually there's no life in our universe. I don't think you got the point. Best,

B.

akidbelle said...

Hi Sabine, I agree with your statement, so let me precise my thoughts and revisit my wording:

..only one set of parameters, symmetries, and particles listing can exist TOGETHER because it is constrained for some (self-?)consistence reason.

Here "consistence reasons" may refer to some underlying physical laws that we don't know of. For instance if you move alpha, some counter-reaction appears given by its impact on other parameters that puts it back to its "right" value (giving self-consistence).

J.

Sabine Hossenfelder said...

akidbelle,

Yes, that's possible, but not in the framework of the standard model. It was certainly a hope for string theory, but that didn't work out either. I'm not saying it's something we should give up on. I certainly hope (and most of the time believe) that it's possible to at least derive some of the presently unexplained parameters from a suitable more fundamental theory. On the other hand, maybe it just isn't possible. Best,

B.

Arun said...

Forget about fundamental parameters, does String Theory provide a probability measure for compactifications?

Sabine Hossenfelder said...

Arun,

No, it doesn't. It can't. You need a meta-theory to have a probability measure for the distribution of a theory's realizations. The problem is much like the idea of having varying fundamental constants or laws, for which you need laws etc. Best,

B.

akidbelle said...

Arun,

I can only not forget about the word "fundamental"; because naming something "fundamental" is just discarding the possibility that something more fundamental exists. Best,

J.

joel rice said...

The anthropic principle is a distraction. The important fact is that the universe has both patterns and pattern recognizers - and we wonder how it is that those Dirac Matrices actually have anything to do with electrons ! Wigner thought it 'unreasonable' that math is so effective. But it might just be possible that not only is it Reasonable, but Necessary !

Jim said...

"the anthropic princple hasn’t solved the problem it was meant to solve, because it does not single out one unique theory". Meant by whom? Most parameters being continuous, we'd expect some (perhaps "very small") volume of parameter space to allow life, not just a "unique theory" (which sounds like a point in parameter space).

"This study is hence another demonstration that a chemistry complex enough to support life can arise under circumstances that are not anything like the ones we experience today." How can you make such a claim? As Adams and Grohs point out themselves, "although carbon is (most likely) necessary for a universe to be habitable, it is not sufficient." The isotope ratios of other elements, distribution of stellar masses, etc. are also important. Demonstrating that very different parameters could lead to life is a horribly difficult problem! You will probably need to be concerned about the gory details of astrophysics, chemistry, geology, biology,... Surely no one could yet come close to stating that a very different set of parameters is capable of suporting life - it would be a monumental task. Similarly for the other anthropic counterexamples you mentioned. (Of course it may be that some such different parameters could support life - we simply can't say yet.)

Lambda is a very different case, since its effects on the largest (cosmological) scales are quite simple. It's when you get into messy small-scale physics that it gets very, very hard.

The point of the ^12C resonance is not that C and life are impossible outside a finely-tuned parameter range - again, such a statement is well beyond our ability to make at present. Instead, it's a hint that fine tuning may be present, which might be possible to address with anthropic arguments. Not so black and white.

Sabine Hossenfelder said...

Jim,

What I meant was that the theory be the sole maximum. That's the whole idea of our universe presumably being "finetuned for life". You can Google that phrase to see which people have popularized it. Among others, I believe Weinberg already alluded to this in his "Dreams of a final theory", that the theory we are after might not be the only consistent completion of the existing theories (the "old" dream of a toe) but the only consistent completion that is "rich" enough (I believe that's the word he used) to give rise to a universe we'd find ourselves in (the "new" dream of a toe).

Arun said...

Bee wrote: "You need a meta-theory to have a probability measure for the distribution of a theory's realizations."

So string theory fails to be a theory of everything because it can't provide the needed meta-theory?

JimV said...

The other Jim points out that evidence that other universes with significantly different physical constants can develop structures and chemistries analogous (but not identical) to those in our universe is not proof that these other universes could develop life. True, but it is also true that we cannot say definitively that any other set of random physical constants could *not* produce any form of life - defined as self-reproducing complex patterns with Turing computational characteristics. Life composed of magnetic fields, life composed of cellular automata, life composed of interacting fields, et cetera. Or at least, it would be a very hard problem to to prove that for anything but a universe of nothingness (and even then I would wonder whether nothingness was stable).

So the argument from mathematical proof (for and against fine-tuning) ends in a tie. Given then the references provided in Dr. Bee's post, that our universe is not unique in its ability to produce complex structures, I think the weight of the evidence is against the fine-tuning hypothesis, and it fails as a scientific hypothesis, unless and until new evidence for it is found.

Jim said...

Bee,

Anthropic arguments simply state that we're in a "typical" part of the region of parameter space that supports complex life. (Of course, "typical" requires a measure to explicate...) It would be a further claim that that allowed region is essentially unique (or that there is only one "maximum"). It is fine to criticize anyone (Weinberg and the "new dream"?) making this latter claim, but even if we had very good evidence of disconnected viable regions (via mechanisms akin to either of your references, Bee) the former anthropic arguments can still hold. In other words, we may still be in a "typical" part of the viable parameter space.

There is an interesting analogy, however, between:

old unique toe -vs- current very-non-unique string theory

and

unique viable region -vs- non-unique viable regions.

In both cases, there is no good reason a priori to expect uniqueness - it's just that some apparently wish(ed) it to be so.

JimV - I think the evidence is far from strong that other viable regions exist. So much still to work out, as I said. But also we know little about what steps were required to yield complex life here, so can't even evaluate yet whether other parameter regions could do the same. We'll be stuck for a long time yet.