Wednesday, November 22, 2017

How do you prove that Earth is older than 10,000 years?

Mesosaurus fossil. Probably dating back
to the early Permian Period, roughly
300 million years ago. [Image source]
Planet Earth formed around 4.5 billion years ago. The first primitive forms of life appeared about 4 billion years ago. Natural selection did the rest, giving rise to species increasingly better adapted to their environment. Evidence, as they say, is overwhelming.

Or is it? Imagine planet Earth began its existence a mere 10,000 years ago, with all fossil records in place and carbon-14 well into decaying. From there on, however, evolution proceeded as scientists tell us. How’d you prove this story wrong?

You can’t.

I know it hurts. But hang on there, band aid follows below.

You can’t prove this story wrong because of the way our current theories work. These theories need two ingredients: 1) A configuration at any one moment in time, called the “initial condition,” and 2) A hypothesis for how this initial configuration changes with time, called the “evolution law.”

You can reverse the evolution law to figure out from the present configuration what happened back in time. But there’s no way you can tell whether an earlier configuration actually existed or whether they are just convenient stories. In theories of this type – and that includes all theories in physics – you can therefore never rule out that at some earlier time the universe evolved by an entirely different law – maybe because God or The Programmer assembled it – and was then suddenly switched on to reproduce our observations.

I often hear people argue such creation-stories are wrong because they can’t be falsified, but this makes about as much sense as organic salt. No, they aren’t not wrong, but they are useless.

Last week, I gave a talk at the department of History and Philosophy at the University of Toronto. My talk was followed by a “response” from a graduate student who evidently spent quite some time digging through this blog’s archives to classify my philosophy of science. I didn’t know I have one, but you never stop learning.

I learned that I am sometimes an anti-realist, meaning I don’t believe in the existence of an external reality. But I’d say I am neither a realist nor an anti-realist; I am agnostic about whether or not reality exists or what that even means. I don’t like to say science unveils “truths” about “reality” because this just brings on endless discussions about what is true and what is real. To me, science is about finding useful descriptions of the world, where by “useful” I mean they allow us to make predictions or explain already existing observations. The simpler an explanation, the more useful it is.

That scientific theories greatly simplify the stories we tell about the world is extremely important and embodies what we even mean by doing science. Forget all about Popperism and falsification, just ask what’s the most useful explanation. Saying that the world was created 10,000 years ago with all fossils in place is useless in terms of explaining the fossils. If you, on the other hand, extrapolate the evolution law back in time 4 billion years, you can start with a much simpler initial condition. That’s why it’s a better explanation. You get more out of less.

So there’s your band aid: Saying that the world was created 10,000 years ago with everything in place is unfalsifiable but also useless. It is quantifiably not simple: you need to put a lot of data into the initial condition. A much simpler, and thus scientifically better, explanation, is that planet Earth is ages old and Darwinian evolution did its task.

I’m not telling you this because I’ve suddenly developed an interest in Creationism. I am telling you this because I frequently encounter similar confusions surrounding the creation of the universe. This usually comes up in reaction to me pointing out that there is nothing whatsoever wrong with finetuned initial conditions if you do not have a probability distribution to quantify why the conditions are supposedly unlikely.

People often tell me that a finetuned initial condition doesn’t explain anything and thus isn’t scientific. Or, even weirder, that if you’d accept finetuned initial conditions this would turn science itself ad absurdum.

But this is just wrong. Finetuned initial conditions are equally good or bad explanations than not-finetuned ones. What is decisive isn’t whether the initial condition is finetuned, but whether it’s simple. According to current nomenclature, that is not the same thing. Absent a probability distribution, for example, an initial value of 1.0000000[00] for the curvature density parameter is scientifically equally good as an initial value of 0.0000001[00]… because both are equally simple. Current thinking among cosmologists, in contrast, has it that the latter is much worse than the former.

This confusion about what it means for a scientific theory to be useful sits deep and has caused a lot of cosmologists to cook up stories about the early universe based on highly questionable extrapolations into the past.

Take, for example, inflation, the idea that the early universe underwent a phase of rapid expansion. Inflation conjectures that before a certain moment in our universe’s history there was a different evolution law, assigned to a newly invented scalar field called the “inflaton.” But this conjecture is scientifically problematic because it construes up an evolution law in the past where we have no way of testing it. It’s not so different from saying that if you’d roll back time more than 10,000 years, you wouldn’t find planet Earth but god waving a magic wand or what have you.

A bold conjecture like inflation can only be justified if it leads to an actually simpler story, but on that the jury is out. Inflation was meant to solve several finetuning problems, but this doesn’t bring a simplification, it’s merely a beautification. The price to pay for this prettier theory, though, is that you now have at least one, if not several, new fields and their potentials, and some way to get rid of them again, which is arguably a complication of the story.

I wrote in a recent post that inflation seems justifiable after all because it provides a simple explanation for certain observed correlations in the cosmic microwave background (CMB). Well, that’s what I wrote some weeks ago, but now I am not so sure it is correct, thanks in no small part to a somewhat disturbing conversation I had with Niayesh Afshordi at Perimeter Institute.

The problem is that in cosmology there really aren’t a lot of data. There are but a few numbers. It’s a simple story already without inflation. And so, the current status is that I just don’t know whether or not inflation is a good theory. (But check back next month.)

Let me emphasize that the concordance model (aka ΛCDM) does not suffer from this problem. Indeed, it makes a good example for a scientifically successful theory. Here’s why.

For the concordance model, we seek the combination of dark matter, normal matter, and cosmological constant (as well as a handful other parameters) that best fit current observations. But what do we mean by best fit? We could use any combinations of parameters to get the dynamical law, and then use it to evolve the present day observations back in time. And regardless of what the law, there is always an initial state that will evolve into the present one.

In general, however, the initial state will be a mess, for example because the fluctuations of the cosmic microwave background (radiation) are not related in any obvious way to the structures we observe (matter). Whereas, if you pick the parameters correctly, these two types of structures belong together (higher density of matter corresponding to hotter spots in the cosmic microwave background). This match is a great simplification of the story – it explains something.

But the more you try to turn back time in the early universe, the harder it becomes to obey the scientific credo of storytelling, that you should seek only simpler explanations, not more complicated ones. The problem is the story we presently have is already very simple. This really is my biggest issue with eternal inflation and the multiverse or cyclic cosmologies, bounces, and so on and so forth. They are stories, all right, but they aren’t simplifying anything. They just add clutter, like the programmer that set up our universe so that it looks the way it looks.

On some days I hope something scientific will eventually come out of these stories. But today I am just afraid we have overstepped the limits of science.

98 comments:

Psmith said...

Tell us more about this disturbing conversation with Niyaesh !

Paul Boulanger said...

For me, falsification was always linked to usefulness. One way of being useful for a theory is to be able to provide predictions or offer new insights. By threading into the unknown (the unmeasured yet) it is putting itself at risk of being proven wrong when its predictions or insights are proven false by new measurements. This, for me, was the crux of Popper's suggestion: the desire for falsification leads to form theories that try to predict things or offer new insights, a.k.a. force us to concentrate on useful theories.

Sabine Hossenfelder said...

Psmith,

I have a feeling that before long you will hear more of it than you wished for ;)

Jayarava Attwood said...

I shocked a friend the other day by disagreeing that scientists were seeking the truth. As I understand science, as I was taught it and practiced it back in the day, scientists seek to accuracy. Whether or not a theory is true or not is hardly relevant as long as it accurately predicts new observations and plausibly explains existing observations.

We can make accurate inferences about reality by comparing notes about our observations and ruling out the purely personal. If 10,000 heads turn in unison at a tennis match, it is very much harder to explain that without reality than with it.

"Saying that the world was created 10,000 years ago with all fossils in place is useless in terms of explaining the fossils."

It depends how you define useful. For fundamentalists it is very useful to believe this, because God is the explanation for everything. So the fossils are just more proof that God has been at work and evolution makes no sense anyway. We cannot disprove this either. For *scientists* such theories are useless, because the use scientists make of knowledge is completely different. Usefulness depends on the use to which you wish to put something. A screwdriver makes a poor hammer. Usefulness is always defined with respect to human subjective judgements (John Searle makes this point in his book The Construction of Social Reality).

Defining reality in reductive terms appears useless to me. If reality is only the lowest level of structure in the universe, then biology makes no sense at all. One cannot understand organisms by dissecting them or reducing them to quantum fields. One has to observe them alive and interacting with their environment; i.e. as complex wholes, embedded in systems; doing stuff their parts cannot.

It seems to me that simplicity is just as bad as "beauty" as a criteria for what makes a good scientific theory. There is no a priori reason why anything should be simple.

naivetheorist said...

bee:

"But I’d say I am neither a realist nor an anti-realist; I am agnostic about whether or not reality exists or what that even means. I don’t like to say science unveils “truths” about “reality” because this just brings on endless discussions about what is true and what is real. To me, science is about finding useful descriptions of the world, where by “useful” I mean they allow us to make predictions or explain already existing observations. The simpler an explanation, the more useful it is."

i agree with you. this vew is actually fairly wide-spread amongst theoreticians even though it is not often stated in the literature. here are a few relevant quotes (you may have noticed that i'm a collector of quotes by scientists):

Stephen Hawking: "I don't demand that a theory correspond to reality because I don't know what it is."

Stephen Wolfram:"All any model is supposed to do whether it is a cellular automaton, a differential equation, or anything else - is to provide an abstract representation of effects that are important in determining the behavior of a system. And below the level of these effects there is no reason that the model should actually operate like the system itself."

Rudy Rucker: "I don’t think that the world is made of computations - any more than I think that it’s made of atoms or of curved space. These are all just useful modes of thought."

and my favorite quote on this topic is by the brilliant comedian:

Robin Williams: "Reality. What a concept."

richard "naive theorist" (this a self-description by Pierre-Giles de Gennes, winner of 1991 Nobel Prize in Physics)

Sabine Hossenfelder said...

naivetheorist,

Thanks for these quotes! I recently had a philosopher make fun of me for stating that science isn't about truths, and this makes me feel so much better. Next time I'll quote Hawking in my defense ;)

Uncle Al said...

1) 1857 meters of Grand Canyon vertical banding.
2) 3+ km deep South Africa gold mines with swamp remains (the gold ore).
3) Magnetic reversal banding of sea floor rifts.
4) 1.251×10^9 year half-life K-40 decay, 10.72% channel to (0.996035 atmospheric Ar-40 versus 0.000629 Ar-38, 0.9% of atmosphere overall) by electron capture, plus a 1.460 MeV gamma ray and a neutrino. Compress decay to 10,000 years and melt Earth’s crust
5) Core drill the South Pole 2.7 km-thick ice pile.
6) Mt. Everest is topped with sedimentary limestone sea floor.

These are jointly and severally consistent with an old Earth. Fast vast bulk is OK, but not its progressive fine structure integrated over all cases horizontal and vertical. People have been around longer than 10,000 years.

https://www.nytimes.com/2014/10/09/science/ancient-indonesian-find-may-rival-oldest-known-cave-art.html

But wait! If the past is contingent upon present free will, we can will the Earth to be 10,000 years old. How well did the Dark Ages succeed there as 1-6 were in place for both initiation and termination. I prefer an iPhone to Adam losing his (and thus our) baculum to Eve. Bring it back!

Aaron Boyden said...

You're an anti-realist. Anti-realism is so-called because it is a reaction to the various kinds of realist dogmatism and over-reach; it doesn't involve commitment to any rival understanding of things. The sort of pragmatism you describe is exactly what anti-realists in the philosophy of science are in favor of (well, most of them; admittedly, some people are anti-realists because they have a dogmatic agenda of their own which is a rival of realism, but those people are usually not in philosophy of science).

Don Lincoln said...

Well, Sabine, I now better understand your thesis. There is merit in it, but could one not state that your statement of "simple is better" is itself an aesthetic value? I'm not arguing that you're wrong, but probing the weaknesses of your argument in much the way you are arguing the weaknesses of, for example, inflation.

Personally, I think that science is about asymptotically approaching truth. But there will be a few more episodes of dabbling with metaphorical epicycles along that particular journey.

William R Somsky said...

Without any claims of truth or usefulness, consider the following:
- Perhaps the universe was created "just-so" 10,000 years ago so that things look "right" today.
- Perhaps it was created "just-so" *yesterday* so that things look "right" today, and all your past life never *really* happened, but is just an artificial memory you were created with.
- Perhaps it is ***yet*** to be created "just-so" ***tomorrow***, and everything happening today isn't really happening but is just an artificial memory that you will recall the day after tomorrow...

Henning Dekant said...

Back in high-school I told a Christian Evangelical friend who's church wanted him to believe creationist BS, that if a god had such a wicked sense of humour to create an earth that looks like it has been around for billion of years, then I really wouldn't consider such a god to be particularly benevolent.

What sums up my very reduced set of believes is that we can make sense of the world. I.e. as you put it, we have the ability of finding useful descriptions, and this faith is supported by the fact that science already created so many. Sometimes I toy with the realist view that this means we can find ultimate truth through the scientific method, but I usually remind myself that this is an article of faith in order to regain my agnostic composure.

Would have loved to attend your talk in Toronto, but family obligations always require me to travel to the US for American Thanksgiving this time of year.

Matthew Rapaport said...

Science as a process (methodological naturalism) is implicitly realist. What would be the point of doing science if you believed that you had no handle whatsoever on the mind-independent reality you observe and describe?

Arun said...

“Useful” suggests the question: “useful to whom?”.

E.g., a theory of race may survive because it is useful to the race in power, and not because it is true. The upper limit of sugar in your daily diet is something the sugar industry has a use in setting as high as possible. On reflection I think we would want “useful” to mean “corresponding to the world as it is beyond our human wishes, fears and goals” . But isn’t that another definition of “reality”?

Tanner said...

Who needs science when you have magic?

KMoustafa said...

An interesting book on this topic is "Darwin and the Ape," by Christopher Cosans, published by University of Indiana Press.

Koenraad Van Spaendonck said...

Hello Sabine,

" This really is my biggest issue with eternal inflation and the multiverse or cyclic cosmologies, bounces, and so on and so forth. They are stories, all right, but they aren’t simplifying anything. They just add clutter, "

Consequently a simpler model would rest e.g. on scenario's which do not need inflation, big bang, big bounce, even accelerated expansion.

But that would require different interpretations of what redshift entails (not cosm.redshift for the DE case, and not extra Doppler effect for the DM case.

If redshift were a property of space, we'd have a static picture, and there would be no talk of ' What happened in the early universe ?' as we have deduced it from extreme extrapolations.

This would require 'questioning the question'. The new question we would then ask is a different one : What is redshift if the universe is static ?

Can we model it as a property of space itself, within e.g. a generalisation of GR, thus avoiding the epicycle route and all of its issues and speculative nature ?

To create such a model with redshift as a property of space, one should start by digging into the foundations of the possible physical causes of gravity first.

Best, Koenraad


Peter said...

"Simplification" or "simplicity" seem to be good (historically validated) criteria for a new/larger scientific theory. But are they general? What if a situation IS actually complicated, and a theory seeking simplification, and therefore looking attractive, is an oversimplification? Multiple additional dimensions, or parallel universes, do not seem to simplify things, yet they might well be "true" or at least able to explain more than when excluding them. I suppose then a discussion and definition of "simplification" would be needed?

Angra Mainyu said...

Sabine,

While I would agree that some of the critiques of fine-tune arguments, Young Earth Creationism, etc., are not good, I don't agree with the assessment that Creationist hypotheses are not wrong, but useless. I hold that they are wrong, though not because they're not falsifiable. Perhaps, the following analogy will be of use:
Consider, for example, a murder trial. There is the following pieces of evidence:

a. Video footage from security cameras show the defendant kill the victim, take her wallet and some other objects, and then leave. There is no evidence indicating tampering.
b. DNA evidence and fingerprints are all over the victim's body, and in particular in the places the video shows the defendant touched.
c. The victim's wallet, purse and other objects were found in the defendant's place.
d. The victim managed to use turn her cell phone camera on during the attack. It's all recorded, and again, there is no evidence of tampering.
e. The perpetrator had already been convicted of robbery in the past.

But let's say that the defense attorney posits the following alternative hypotheses:

H1: The world is less than 1 day old, created with memories and all. The victim never even existed, and so there was no murder.
H2: We live in a simulation, and the defendant was framed by some of the simulation PTB. But he did not do it.
H3: Some powerful demons (or aliens, or whatever) planted the evidence, for some reason (e.g., for fun, or in an experiment, etc.).
H4: The laws of nature were suspended by God for mysterious reasons, and so the cameras can't be relied upon, the DNA evidence, fingerprints, etc., also can't be relied upon, etc.

So, the defense attorney posits that one of those hypotheses might be true, and it's not possible to prove otherwise - after all, each of them is consistent, and consistent with observations as well.
She also points out that there are infinitely many alternative hypotheses, pairwise incompatible but each of them consistent and logically compatible with observations.
So, she asks the jury to acquit, on the grounds that it's not known whether or not the defendant committed the crime, or even whether there was a crime in the first place, so the facts allegedly have not been established beyond a reasonable doubt.

Yet, the jury should not acquit. Those alternative hypotheses can be ruled out beyond a reasonable doubt. They're (individually and also as a disjunction) too improbable. But one does not need to be able to write down a probability distribution to do that. The members of the jury can just do a normal human probabilistic assessment, using their own brains/minds. That's not infallible, but it generally works properly, and without that, humans would not be able to assess (at least) almost anything.

So, the jury concludes on the basis of the available evidence that the defendant actually did it. It's not just that it's useful (for some purpose or another) to posit that the defendant did it. Rather, it's that he actually did it. If the proper conclusion were only that it's useful to make predictions to say that he did it, but that one doesn't actually know whether he did it, then it would be unjust to convict him. Now, Creationism, in its traditional form, did make predictions, based on which scientists in the past expected to find things they did not find. Of course, one can always come up with a Creationist hypothesis that hold that, say, God planted the fossils for some mysterious reason (and the light coming apparently from starts, etc.), or that the devil did it to confound us, or whatever, without incurring contradiction and without contradicting observations. But just as one can tell that this is not true in the trial case, I would say that similarly, one can make the same assessment in the case of (Young Earth) Creationism.

Angra Mainyu said...

Addition:


This isn't to say that one is always in a position to make one such assessment, and there are cases in which remaining undecided is epistemically proper. But Creationism is mistaken, and also, science generally does try to find what actually happened, is happening, etc., not just of making models to predict observations. If it turns out that sometimes one actually can't tell what happened and only has a model to predict observations, that's much better than nothing of course, but it's even better if we can find out what happened.

That said, I don't know much about cosmology, and for all I know, it may very well be that there is no strong evidence favoring one of the models over any of the others (or at least, over the disjunction of all of the others), and so scientists do not know what actually happened, even if they have good models to make prediction. And it may well be, for all I know, that some scientists are going beyond what the evidence warrants (actually, I reckon this is probable, for a number of reasons, though in that case, I'm basing my assessments on the behavior of the scientists involved, rather than on the theories themselves), or even that things like the early universe are so far away from our daily experiences that we do not have a means of properly assessing probability of some hypotheses - I don't know.

But in many cases, we do know what happened: It's true (not just useful) that humans evolved from monkeys, that humans, mosquitoes and trees have a common ancestor, etc. The evidence for that is overwhelming, just as the evidence against the defendant in the hypothetical scenario is also overwhelming, and regardless of the fact that in both cases, there are infinitely many alternative hypotheses that are pairwise incompatible, and each of them consistent with the observations. Similarly, it's false that there was a Garden of Eden where lions, T-Rex and sheep lived side by side peacefully, etc., so I would say that while you may be right for all I know in your critique of some present-day scientific stances, I don't think the rationale (i.e., in terms of usefulness instead of what's true) is correct.

John Anderson said...

The YEC crowd does not want and will not in fact be persuaded. You, Sabine, will only be sullied by the interaction. You are wasting your breath. I strongly recommend staying as far away from this topic as you possibly can. Laugh and move on.

The children of the YEC-ers will go their own way so don't worry about them. I grew up in the Gold Buckle of the Bible Belt and recognize that wrestling a greased pig is a very messy activity. You will not influence the pig's behavior.

Have you ever wondered why Michael Shermer publishes "Skeptic"? The people who believe garbage think he's a fool and a stooge, and those who agree with Shermer just self-congratulate themselves on their superior intellect and education. Absolutely nothing is accomplished.

BTW, I recommend visiting Gateway Canyon resort in western Colorado for dino tracks and beautiful landscapes. Moab, UT has a boat load of large dinosaur tracks. Woodland Park west of Colorado Springs has petrified sequoias with a 41 foot circumference. These people should look around. But you will never (>> 10^500 gigayears) force them to abandon their position. There is plenty of science that needs coverage for interested people now.

Sabine Hossenfelder said...

Don,

My reference to simplicity is really just Occam's razor. I am not saying simplicity is an absolute value, but using it in the form to mean 'as simple as possible to best fit data'. I elaborated on this previously elsewhere (and explain it in my book), but felt this post was getting too long. This is too mean, I am not saying simplicity is better in and by itself. I would agree that simplicity as an absolute value is another aesthetic criterion. (You find this eg in unification attempts.)

Having said that, I think without a requirement of simplicity you can't do science. Or rather, science becomes indistinguishable from religion. You would then be allowed to add any kind of magic explanation, angles, fairies, gods, silver strings tied to souls, and so on to your "theories".

Let me add that I also would agree that simplicity has a human component and is thus a little vague. While one could try to quantify it, in practice we chose theories that are simple for humans to use, which also depends on what we learned during training and what fits with our cognitive models of the world.

Best,

B.

Sabine Hossenfelder said...

John,

If you think I wrote this post for Creationists, you entirely misunderstood it.

Sabine Hossenfelder said...

Peter,

That's right, as Einstein put it so aptly, a theory should be as simple as possible but not any simpler.

Sabine Hossenfelder said...

Arun,

As I wrote, useful to describe observations. If that description depends on who does the describing or the observation you either have a theory or an experiment that is no good.

Myron A Penner said...

Such a great post, as per usual! I'm having hard time, though, figuring out how "useful for making descriptions" or the references to "accurate descriptions" means something different than "true descriptions of the world." Could you unpack how science's commitment to "accuracy" and seeking theories/models/laws that are "useful for making (accurate) descriptions" doesn't automatically entail the view that science is, minimally, seeking truth?

Koenraad Van Spaendonck said...

@naivetheorist

" Stephen Hawking: 'I don't demand that a theory correspond to reality because I don't know what it is' "


I agree.

Science at its best, always ends up with an improved model of merely the shadows on the wall of Plato"s cave.

Best, Koenraad

Sabine Hossenfelder said...

Angra,

I did not write this post to present an argument against Creationism. I know that Creationists don't merely pose the Earth was created not long ago but (at least most of them) also believe other things that are incompatible with scientific evidence for reasons other than what I have pointed out. I have chosen the example I laid out in my post carefully. I spelled out exactly what is the claim you are trying to disprove. I did not say this disproves Creationism or even addresses creationism or this is an argument I recommend anyone uses in conversation with Creationists. Frankly I am not interested even having a conversation with Creationists; I know better ways to waste time.

While there are cases in which you can appeal to probability, the history of our universe isn't one of them. We have only this one instance at our disposal. Hence you need a different argument. That's the reason for this blogpost.

Sabine Hossenfelder said...

Myron,

Suppose you have a set of data, you are looking for a theory that reproduces this data in a way that's both as simple as possible and as accurate as possible. There are various measures you can use to find that sweet spot. The word "true" is often colloquially used to mean "very accurate" or "very likely correct". I have no problem with that use of the word. It's just that I try to avoid it when it comes to scientific theories because I don't know how to define it. I know how to speak about accuracy and uncertainty when it comes to describing data. I know how to define truth for mathematical statements. But I don't know how to define truth for the description of observations.

Having said that, let me be clear that I am by no means implying that everyone uses these words the way I do. I add these explanations merely so that readers can understand how I use them. If I didn't explain it, I think my writing would be rather ambiguous.

Best,

B.

naivetheorist said...

bee:

"Next time I'll quote Hawking in my defense ;) ". Quoting Hawking will probably work (i.e. will shut your opponent down) but it is unfortunate when it is necessary to use 'argumentum ad verecundiam' (appeal to authority) in order to convince someone of the merits of one's case. especially since Hawking is no more of an authority than you are.

happy turkey day (this year, you europeans have more to be thankful for than we americans do - at least, you don't have Trump as your leader).

richard

Sabine Hossenfelder said...

naivetheorist,

In some cases an appeal to authority or popularity is justified, for example if the argument in question is about whether or not a particular line of thought is even used by anyone else. Ie, if someone claims "No one says X" and you demonstrate "The King of Sweden said X" that's not a logical fallacy.

Sabine Hossenfelder said...

(or maybe one just wouldn't call it an appeal to authority in that case, not sure about the terminology)

Phillip Helbig said...

"Absent a probability distribution, for example, an initial value of 1.0000000[00] for the curvature density parameter is scientifically equally good as an initial value of 0.0000001[00]… because both are equally simple. Current thinking among cosmologists, in contrast, has it that the latter is much worse than the former."

Several issues here. First, with regard to this particular example, there is no fine-tuning at all in classical cosmology. This aspect of the flatness problem is based on a misunderstanding and false argument from analogy. However, this is a particular (though famous) case, and does not imply that all fine-tuning claims are bogus.

Second, while 1.0000002[00] and 0.0000001[00] are equally simple, 1.0000000[00] and 0.0000001[00] are not. Why? Because the former is a "special value" which was known before any measurements were made. Yes, all lottery combinations are equally probable, and a sequence of 7 consecutive numbers, say, is just as probable as some "random" distribution. That is not the point. With the lottery, there is no special value, but with respect to the flatness parameter there is, which is why the lottery is a bad example.

Perhaps a better example: Suppose I write a random-number generator which should produce uniform deviates between 0 and 1. However, due to a code error (using integer instead of floating-point division, say), all numbers are rounded down to 0. I give you a list of 1,000,000 such "random" numbers, and all are 0. Would you say "no problem; that's just as probable as any other sequence"? If so, then you are missing the point.

Maybe someone will step up to the plate and defend me on this. :-)

Third, fine-tuning arguments and naturalness arguments are essentially opposed. Naturalness claims that ratios of parameters should be of order 1, while fine-tuning arguments claim that this is unlikely (yes, even if the underlying probability distribution is not known). If two things are equal, there is usually a reason for it.

johnduffieldblog said...

An interesting read, Sabine. Perhaps you might have finished up asking this question: How do you prove that the Universe is older than 13.8 billion years?

Phillip Helbig said...

Ie, if someone claims "No one says X" and you demonstrate "The King of Sweden said X" that's not a logical fallacy.

(or maybe one just wouldn't call it an appeal to authority in that case, not sure about the terminology)


Argumentum ad regem. :-)

Perhaps better than argumentum cum rege.

Best of course is praemium ex rege. :-D


Angra Mainyu said...

Sabine,

I didn't think you had written this argument as an argument against Young Earth Creationism (YEC). My disagreement was with the rationale that you gave against it (i.e., that's its useless, rather than false), not because it's an argument against YEC, but generally because of the reasons I tried to explain in my two posts.

On a similar note, I don't agree that the fact that there is only one instance of the history of the universe entails we can't use probability. Probabilistic assessments are not the same as frequencies. Take, for example, the court case I mentioned. The defense attorney may insist that, for all you know, you can't use probability, either, because for all you know, the universe is less than a day old, and if so, then you didn't actually observe repeated phenomena, etc.; and yet, the defense attorney failed to introduce a reasonable doubt, I would say that because the scenario is far too improbable. That seems to me like an ordinary intuitive epistemic probabilistic assessment to me.

In general, I would say that you can establish beyond a reasonable doubt that the universe is over 10000 years old (or 10 million, or a billion, etc.), even if you can't prove it in a mathematical sense. But the same applies to ordinary events, like - say - whether the Moon Landing happened, or whether a defendant committed a crime.

Sabine Hossenfelder said...

Phillip,

If you believe that 1.000000001 is a 'special value which was known before any measurement was made' you are, for all I can tell, making a sociological argument. As in 'someone has thought of this before, so it's special'. I don't see the fundamental relevance of this.

And yes, it is correct that not all finetuning arguments are wrong. They are fine if you do have a probability distribution.

Sabine Hossenfelder said...

Angra,

I don't know what sense it makes that you claim "an ordinary intuitive epistemic probabilistic assessment" is not a probability assessment.

Having said that, I think you didn't understand my previous comment. It is correct that in daily life you can make probabilistic assessments, which, for example, makes the thesis that the moon landing was a hoax ridiculously improbable. But you can't make such an argument for the history of the universe, which is what I was getting at. Best,

B.

Brian Dolan said...

Sabine,

I am not a philosopher, I don't know what an anti-realist is, but I don't think I want to be one. I do know that theoretical physicists build and study mathematical models of Nature. I like to tell my students that a mathematical model is just that -- a model. Never confuse your model with reality. I hope that doesn't make me an anti-realist.

Some models are more useful than others and some are more beautiful than others, which you consider to be more important is largely a matter of taste.

If two theories are both compatible with observations then it is usually best to choose the simplest one (which is also often the more beautiful). I think philosophers call this Occam's Razor?

Two more quotes that I like:

1) "It is more important to have beauty in one's equations than to have them fit experiment..."
— Paul A. M. Dirac, In 'The Evolution of the Physicist’s Picture of Nature', Scientific American, May 1963, 208, 47.

This always makes me think of Kepler's "Mysterium Cosmigraphicum", in which he found a relation between the orbits of five planets known at the time and the five Platonic solids. It was a very beautiful idea and it was completely useless (except for the fact that it made him famous!).

2) "If your theory is found to be against the second law of thermodynamics I can give you no hope; there is nothing for it but to collapse in deepest humiliation."
-- Arthur Eddington, The Nature of the Physical World (1915).

I would like to think that my firm belief in this one makes me a realist!

Edward H Russell said...

Sabine,

Due to the ambiguity of language there are so many synonyms -- useful, pragmatic, utilitarian, anti-realist, even true -- most of which carry an implicit value judgment. Perhaps an unambiguous term or symbol is needed to refer to that which is intrinsically or at least currently known only conditionally.

In case it hasn't popped up on your radar I would like to call attention to an attempt in that direction by Marcelo Gleiser. He comes at it from a different angle: http://www.hts.org.za/index.php/HTS/article/view/4713, concluding: "The logical insolvability of the problem of the First Cause [shows] the conceptual inappropriateness of current scientific models that claim to offer a solution."

Thank you for stating this somewhat controversial view so clearly.

Best, Ed

pete best said...

What about multiple lines of evidence from different disciplines and the balance of evidence argument?

In the case of the the earths age, fossils sit in earth and earth along with rocks is geology and geology offers a second line of evidence into the earths age along with dating evidence from meteorites etc. Dating ancient rocks using radioactive decay of carbon and uranium etc. Tectonic plates and shifts in the earths past etc. We also get evidence from DNA, mitochondrial etc. So on the balance of independent lines of evidence we can conclude that the earth is older than 10,000 years.

As for inflation and the origin of the Universe, arghhhh, nightmare. All we have is classic relativity which says the Universe is expanding and hence it must have been smaller in the past (albeit expanding) and hence a beginning taking the logic to its conclusion. However if we assume it all goes back to a singularity where we have no physics then we cannot run it forwards as we have no physics of singularities and hence no Universe physics ?

So relativity ironically brings about it own downfall via its predictions (very accurate). We see that the hydrogen to helium ratios and its isotopes are correct, we have a CMB from the beginning which is predicted etc so on the balance of evidence relativity is onto something but it defeats itself as singularities have no physics. So we turn to quantum physics and try to get merge the two etc but as Sabine goes on about we have no quantum relaitvity (quantum gravity) and hence we have no beginning.



Rob van Son (Not a physicist, just an amateur) said...

"... if a god had such a wicked sense of humour to create an earth that looks like it has been around for billion of years, then I really wouldn't consider such a god to be particularly benevolent. "


That was the original argument that silenced this theory in 19th century fundamentalist circles. If God created the world 10,000 years ago as if it had evolved for billions of years, then God is a liar trying to deceive us. That was an avenue of thought 19th century fundamentalist did not want to follow through. It seems creationists have become so desperate that they are trying to argue around this point: The earth was created mature.

Maybe God being the ultimate liar has started to resonate with their own practices? This is not unprecedented. As an answer to Kepler and Galileo, Catholics already claimed that whether the earth circles around the sun or vice versa was simply a matter of point view (relativism).

Phillip Helbig said...

"If you believe that 1.000000001 is a 'special value which was known before any measurement was made' you are, for all I can tell, making a sociological argument."

No, not 1.000000001, 1.000000000. The difference is the whole point.

Sabine Hossenfelder said...

Phillip,

The difference isn't physically relevant. You can never tell whether it's zeroes all the way down.

Phillip Helbig said...

True, one cannot ever tell if it is exactly zero. But that is also not the point.

The point is that an observed value is near a previously defined special value with only a small uncertainty in the measurement. Fine-tuning doesn't mean that the ratio has to be exactly one.

With regard to the curvature term, then exactly zero is special in a way in which a very small value is not, namely it does not change with time. (In dynamical-systems language, it is an unstable fixed point.)

Some have suggested that having exactly zero curvature would solve the flatness problem. This is wrong on at least two counts. First, the problem doesn't exist in the form in which many think it does, as I've mentioned before. Second, this would work only in a perfectly homogeneous and isotropic universe, not in the one in which we live (which appears to be a Robertson-Walker model on large scales but of course contains small-scale inhomogeneity and isotropy).

The point is that a value of exactly zero is qualitatively different from a very small value, because the former is stable and the latter is not.

With regard to fine-tuning of fundamental constants, qualitatively different behaviour is also the key, such as allowing the existence of stable nuclei, atoms, etc. The point is not that it is some number and not another (whether or not all are equally probable), but rather that some are special and others are not. (The flatness problem confuses things because the special number in this case, 0, is also one which is mathematically special.)


Sabine Hossenfelder said...

Phillip,

I am having a strong sense of deja vu here. As I believe I said previously (or if I didn't, I should have), arguments about stability (in the case of curvature) or sensitivity to initial conditions (in the case of naturalness) are one thing. Arguing that these have a relevance for what is the correct description of nature another thing entirely. The latter just doesn't follow from the former, which is what you are implicitly assuming.

Phillip Helbig said...

Yes, déjà vu all over again. :-)

We want to understand the universe. We want to know why things are the way they are. If something feels unlikely (I'm being deliberately vague here), then there are several options: we are mistaken in believing it to be unlikely (flatness problem), is is unlikely but it has to be something (pattern of the constellations, say), it is unlikely but there is a weak-anthropic explanation in the context of the multiverse, we are in a simulation, the world was designed by a creator, etc. If we understand why something is the way it is, we learn something about the universe. I don't see any hidden assumptions here. Of course, we can't be absolutely sure, just like we can't be absolutely sure that we are not dreaming. (I've actually wondered in a dream whether I was dreaming, then woke up. But I was still in a dream. Then I woke up again.)

Nady Shamy said...

IF WE CHOOSE THEORIES FOR SIMPLICITY THEN THE FACT OF CREATION IS THE MOST SIMPLE SINCE GOD CREATES BY NO MATERIAL MECHANISM BUT BY HIS COMMAND , IN ADDITION , ANY HUMAN CRITERIA FOR CHOOSING THEORIES ARE MERE SUBJECTIVE ACT WITH NO LOGICAL NECESSITY AT ALL , OUR PROOF IS THE THOUSANDS OF THEORIES IN COSMOLOGY WHICH PROVES THAT THEORIES HAVE NOTHING TO DO WITH ONTOLOGY BUT ARE MERE JUST SO STORIES AS YOU CALLED THEM .

JimV said...

This discussion reminds me of Socrates, wise because he knew that he knew nothing (in terms of models or explanations, with certainty). People do not like being told they know nothing, so it did not turn out well for Socrates.

A successful gambler (in some story I read long ago) was asked, how do you bet if a roulette wheel has landed on black for several successive turns: do you then bet on red, or black? I bet with the run, he replied. Perhaps that is the best we can do. (Then it becomes a matter of how we read the run.)

Benjamin Phelan said...

Thanks for the post, Sabine--For years I've been making a similar argument about magical fossils and YEC being logically unassailable, but my otherwise likeminded friends get a little heated. They seem to think the argument is that the YECists are, in an I'm-OK-You're-Okay sort of way, "right," but it's really about why they can't be convinced that they're wrong. A friend of mine teaches high school biology in the Bible Belt and when kids come up to him with the magic fossils argument he lets them explain it, in all its convolutions, and then asks them, "Do you believe God is good? [Yes!] Do you also think He is deceptive? [...umm ..."

Angra Mainyu said...

Sabine,

You say "I don't know what sense it makes that you claim "an ordinary intuitive epistemic probabilistic assessment" is not a probability assessment." I don't claim that. On the contrary, I'm saying we can make a probabilistic assessment. And it's extremely probable (beyond a reasonable doubt) that the Earth is over 10000 years old.

I think I did get your previous post, but you did not get my reply. Briefly, I'm saying that we can make an assessment like that at least when it comes to the age of the Earth.
I don't know enough about current cosmology hypotheses to tell whether one can when it comes to choosing between that sort of hypothesis. But an old Earth is a hypothesis we can evaluate with the usual manner. In order to establish that it's true, we look at the evidence, and it is indeed overwhelming. My central point is that having overwhelming evidence for a hypothesis does not imply that the observations one has are logically incompatible with all alternative hypotheses - in fact, there are always infinitely many hypotheses consistent with the empirical observations and not consistent with each other. Some (infinitely many) hypothesis also consistent with observations have (even together) a negligibly low prior, though.

Amos said...

If you extrapolate the evolution law back in time 4 billion years, you can start with a much simpler initial condition.

Is that really true? Does the information content of the universe - evolving under the (unitary?) laws of physics - increase with time? Or does "simpler" refer to something other than information content?

I suppose the entropy of the universe increases with time, and if we extrapolate far enough back in time we can find a state of very low entropy... but one could argue that the low entropy state is less probable (in the thermodynamic sense) than a high entropy state. From that point of view, the "simplest" state of the universe might be the end state, rather than the beginning.

Admittedly (and despite unitarity), it may be more difficult *in practice* to extrapolate backwards than forwards (with randomness or exponential sensitivity), but this may be due to the fact that in practice we are dealing with only part of the universe's wave function, with incomplete information.

If we were dealing with the wave function of the entire universe (admittedly a dubious notion itself) it might be just as easy to extrapolate forwards or backwards, and from any arbitrarily chosen foliation. Would any foliation then be considered simpler than any other?

Matias said...


Hello Sabine,

I have been reading your blog for a few years now and as a fellow physicist I have always found your posts to be fairly interesting, although my main area of research is in photonics. Your posts on naturalness are among my favorites! But what prompted my comment (first one in this blog btw), were your thoughts on reality and usefulness in this post.

In my opinion reality is fairly simple to define: reality is the state of things, whatever it may be, which is independent of any abstractions (thoughts or opinions, etc). Reality consists of things that exist, have existed or will exist, no matter if they are comprehensible or not. This does not, however, mean that reality cannot be emergent (i.e. a simulation or what have you). It is conceivable that there is something outside of reality, just like things exist within reality, but they do not "exist", because they are outside of reality. They do something else all together.

Then, by definition, humans are within reality (since we are things as well) and our senses make observations of our surroundings. Our surroundings are also things and thus exist and are a part of reality. Therefore, our senses are constantly probing reality, and since we can agree on certain observations (things fall towards the ground, wind blows, water flows, etc.), it is not a stretch to say that these observations are quite accurate in describing the domain of our existence (which happens to be reality). This is the idea behind reproducibility.

By extension we can say that experiments also probe reality, for we complement our senses with whatever experiments we have to grasp something about our domain that we otherwise cannot. Then, based on our observations, we produce some type of conjectures of reality that we call theories. Theories are tested with observations and are complemented or abandoned if necessary and so on.

That is, science is a pyramid structure, where we build upon earlier observations of our domain and search for new ways to probe it. We start from some very mundane observations (the base of the pyramid) and move towards some less obvious things (the peak) only once the whole pyramid has been build, we have a complete picture of the domain we inhabit (with some experimental uncertainty, of course). Thus, I believe that science ultimately increases the accuracy at which we are able to describe our domain, or, reality. This is also a definition of usefulness; if a theory we constructed does not describe our domain correctly, then it is neither useful nor true and there is no need to make a distinction between "true" or "useful" theories, because they asymptotically approach each other.

But, as I mentioned, this does not mean that there is nothing outside of reality that can affect it. Based on our observations, we just have no evidence of these types of things. And as long as there is no observational evidence to support some type of "meta-reality" it is completely okay to neglect it and only stare at our domain.


But these are just ramblings of a lesser mind. Keep up the good work, I really like your blog! It makes the brain juices flow very comfortably.

Matias said...


Phillip,

You said that:

"Second, while 1.0000002[00] and 0.0000001[00] are equally simple, 1.0000000[00] and 0.0000001[00] are not. Why? Because the former is a "special value" which was known before any measurements were made. Yes, all lottery combinations are equally probable, and a sequence of 7 consecutive numbers, say, is just as probable as some "random" distribution. That is not the point. With the lottery, there is no special value, but with respect to the flatness parameter there is, which is why the lottery is a bad example."

I think you don't know how probability works. For example, all combinations of lottery numbers are not equally probable. You can do the lottery with consecutive numbers (for example 1,2,3,4,5,6,7) everyday, until the end of the universe and you will not win even a single time.

Better than any mathematical proof on this, you can check it online, there are people who store all winning lottery numbers online for everyone to see. There are very rarely more than two or three consecutive numbers in a winning combination and there has never been six or seven.

Keith said...

Not sure about this argument, Time is an odd thing, and saying something like "Magical things happened 10000 years ago that created the universe as if it existed for 14 billion years doesn't really make much sense. Even if we can somehow say it just came into existence (lots of problems understanding any context outside of the universe let alone what language to use and whether concepts like Time are relevant), the age of the universe would still be ~14 billion years old (or whatever it happens to be), just like the size of universe is what it is. Time is woven into the makeup of the universe.

Arun said...

Yes, Bee, but observations are not theory-neutral.

Unknown said...

I am tired of responding to stupidity pretending to be intelligent. Of course you can tell when the fossils were laid down, etc., unless you do not believe in science. And your quotes are quite taken out of context.

Sabine Hossenfelder said...

Amos,

That's an interesting question, and one I haven't been able to wrap my head around. (At least so far.) No, I am not referring to entropy. If I had meant entropy, I'd have said entropy. I mean simple in the sense that you need few parameters to specify the state. Otoh, that you can do this clearly depends on the amount of coarse graining, so it's related to entropy. Well, the brief summary is I don't know. I don't particularly like talking about entropy when it comes to the whole universe because the probabilistic interpretation doesn't make sense. The universe is in one microstate, has always been in one microstate, and that's that. Best,

B.

Sabine Hossenfelder said...

Angra,

I did not say that this is the argument you should make or I recommend you make when you meet a creationist.

Phillip Helbig said...

"I think you don't know how probability works. For example, all combinations of lottery numbers are not equally probable. You can do the lottery with consecutive numbers (for example 1,2,3,4,5,6,7) everyday, until the end of the universe and you will not win even a single time."

Needless to say, it should be obvious to all here that you don't understand how probability works.

Please provide a proof of your claim.

"Better than any mathematical proof on this,"

No, in maths, if you can't prove it, it is usually not worth worrying about.

"you can check it online, there are people who store all winning lottery numbers online for everyone to see. There are very rarely more than two or three consecutive numbers in a winning combination and there has never been six or seven."

Sure, because there are 13 million combinations and only a fraction of those have ever been drawn.

Rob van Son (Not a physicist, just an amateur) said...

Amos
"Or does "simpler" refer to something other than information content?"

I think it does. Science assumes that (1) the universe is ordered and (2) humans can understand that order. In this light, a "simpler" condition means a condition we can better understand. In this sense, 1.000000.... is simpler than 1.000001... . In mathematics, the exact 1 or 0 tend to appear when there is a perfect balance. That is often seen as an indication we have made a step towards better understanding. For instance, a perfectly flat universe indicates a perfect balance somewhere deeper in the theory while any net curvature strongly indicates there must be an (unexplained) imbalance somewhere.

I can certainly understand that theorists think that 1.000001 is a reminder that they have missed something while 1.000000... feels like they can make the next step.

But there is no way we can understand an initial condition where the universe would be created with a 15 billion year fake history ready made out of nothing. By the definition of Creationism, we cannot understand God's actions and so the intervention of a God would defeat science at a fundamental level.

Sabine Hossenfelder said...

Matias,

I find nothing objectionable in your elaboration on reality, except that the reference to reality seems unnecessary to me. You can simply talk about describing observations.

As to your comment addressed at Phillip, it's just wrong.

Sabine Hossenfelder said...

Phillip,

What I am saying is that the number 0.00000001... doesn't require any more or less explanation than the number 1.00000000... at least not if you don't have a probability distribution that quantifies either one as less probable than the other.

As I said earlier, sure, we want explanations, and explanations are always good. I am merely saying it's wrong to think that some numbers are in greater need of explanation than other numbers. Plus, if an "explanation" is more difficult than just picking the parameters to fit observations, it's not scientific for the reasons I laid out above. If you'd allow scientific stories to be more complicated than necessary, science would become indistinguishable from religion. Best,

B.

naivetheorist said...

bee:

"I don't particularly like talking about entropy when it comes to the whole universe because the probabilistic interpretation doesn't make sense. The universe is in one microstate, has always been in one microstate, and that's that."

just so (i'm practicing my BBC english).

entropy is commonly 'explained' in the science literature (both in textbooks and articles) as being a measure of a system's disorder. This is incorrect. Entropy is a measure of the 'multiplicity' of a system.

and since the universe is, as you say, a single microstate. discussing the entropy of the universe is nonsensical (Penrose notwithstanding).

query: what do you think about the use of information in the definition of entropy? it seems to me to subjectivize physics and it makes no sense to me.

naive theorist

Sabine Hossenfelder said...

naivetheorist,

What the heck do you mean by "multiplicity"? I only know this from inelastic scattering.

I believe it was David Albert who said something to the extent that entropy was invented to understand steam engines, and applying it to the universe seems quite some stretch. Can't find the quote, but think it was something in this spirit. Or maybe it was someone else who said this. In any case, it's not my quip, but I think it is very apt.

I don't really know what you gain by talking about information. If you have a definition for it, fine. If not, why bother?

naivetheorist said...

bee:

here's one definition (the last sentence provides the defintiion:

We first introduce the very fundamental statistical ideas of microstates and macrostates. Given a system (e.g., a gas), we view it as built from some elementary constituents, (e.g., molecules). Each constituent has a set of possible states it can be in. The thermodynamic state of the system (which characterizes the values of macroscopic observables such as energy, pressure, volume, etc. ) corresponds to many possible states of the constituents (the molecules). The collection of states of all the constituents is the microstate. To keep things clear, we refer to the macroscopic, thermodynamic state as the macrostate. The vast disparity between the number of possible macrostates versus microstates is at the heart of thermodynamic behavior! The number of distinct microstates giving the same macrostate is called the multiplicity of the macrostate.

The Albert quote is great. almost as good his dead-on criticism of Lawrence Krauss's "A Universe From Nothing" (http://www.nytimes.com/2012/03/25/books/review/a-universe-from-nothing-by-lawrence-m-krauss.html?_r=5&)

naive theorist

JimV said...

As usual, this is going off-topic, but I agree that not all lottery numbers are equally probable. The lottery-draws I have seen pick ping-pong balls with numbers on them - without replacement. So if the first number is 1, there are fewer 1-balls than other balls to be picked for the next number. (This does not support Matias's conclusion, of course.)

I also disagree with Nady Shamy that a creation-god model is the simplest. It provides no explanation at all (no how or why), it just asserts that such details are beyond our comprehension. There is no basis for determining its simplicity, except that powers beyond our comprehension/duplication are not apt to be simple ones. I like to refer to it as the "god ate my homework" explanation.

ppnl said...


I think I have to go with Phillip Helbig on this.

Say I have a coin. I flip it a hundred times and it comes up heads every time. Does this "fine tuning" need an explanation? After all flipping a hundred heads in a row is just as likely as any other sequence. And yet I defy anyone to flip a hundred heads in a row and not look for a reason why.

Well why is this? I think the answer is that a hundred heads in a row is an algorithmically simple result. That is unexpected and improbable unless there is an underlying mechanism to explain the algorithmic simplicity or "fine tuning".

You say we don't know the probability distribution. But that isn't true. If our results look algorithmically simple and thus look like "fine tuning" then that implies that the probability distribution is actually narrower than we had any reason to think. Looking for a reason for that narrower probability distribution is to simply look for the reason for the fine tuning.

In the case of the coin we would first look to make sure it doesn't have two heads. Then maybe we would check it's weighting to see if it was out of balance. From there we would check ever more complex reasons for the "fine tuned" results. I am pretty certain that there would be an explanation and it would take a lot to make me accept that it was just a chance occurrence.



Phillip Helbig said...

"What I am saying is that the number 0.00000001... doesn't require any more or less explanation than the number 1.00000000... at least not if you don't have a probability distribution that quantifies either one as less probable than the other."

Maybe there is hope. Let's take the flatness problem as an example, though as I've pointed out there isn't really a problem here, but it is a concrete example and has often been used in this context.

The difference between exactly 0 and almost exactly 0 in this case is that the former is an unstable fixed point. So, it always has the same value, but other values will evolve with time. Thus, it is qualitatively different. (This example is misleading, though, because the evolution is highly non-linear; in this case, we DO have a probability distribution, but it is not what most people assume it is.)

With regard to what I consider legitimate fine-tuning arguments, the observed values are qualitatively different. Again, the book by Lewis and Barnes makes the point better than I can in a comment box. I have an unread extra copy; all you have to do is have lunch with me. I'll even buy! Send me an email.

Matias said...


Sabine (and Phillip),

What I meant in my rather lengthy comment to your text was that whether or not anyone believes that a reference to reality is necessary does not matter, since useful theories describe our domain of existence (which is reality if one chooses to define it like I do). In this case reality is implicit and there is no need to refer to it. Of course, there are other possible definitions for reality, but I will not go there.


As for my comment to Phillip, I am slightly disappointed of your response (and Phillip's too), so let me elaborate a bit. I am taking the lottery as an example again and I am assuming Finnish rules (since I don't know about other countries), where we draw 7 numbers from 40 possible ones, therefore there are 18 643 560 possible combinations.

One would think that the probability of winning with any combination is 1/18 643 560, but this is assuming a uniform distribution, which is usually justified because each number has an equal probability. This is not true, which is also a major feature in thermodynamics. In the context of thermodynamics, the probability of a single combination is a microstate.

To find out how the different combinations are distributed, one needs to look at the check sum of the sequences, that is, take a sum over all the chosen numbers. In the context of thermodynamics this would be a macrostate. The check sum has a minimum of 28, which is the state with the numbers 1+2+3+4+5+6+7, and a maximum of 259 = 34+35+36+37+38+39+40. There is exactly one possible combination for the minimum and maximum, so they have a multiplicity of 1. Whereas, for example, the sum 35 is satisfied by two combinations, 2+3+4+5+6+7+8 and 1+2+3+5+6+7+11, thus it has a multiplicity of 2, and so on.

It turns out that there are two states with the highest multiplicity, 143 and 144, (the mean of the distribution is 143.5), and the probability to find each state is simply the multiplicity of the state divided with all possible combinations. So the states 28 or 259 are vastly less likely than 143 or 144. In other words a sequence with a wide spread is more likely to win than one with a tight spread.

In statistical mechanics, the same ideas are applied through entropy, where every microstate is always equally probable (because if they werent, it would be difficult for a system to find equilibrium). In spite of this, the macrostates with the highest entropy (multiplicity) are the most likely, and we don't see egg white and yolk separating, although the microstates that describe this process are as likely as egg white and yolk not separating.

In this sense my earlier statement was slightly off: given any sequence with consecutive numbers, the probability is not always lower than for randomly chosen ones. Just the states with minimum multiplicity are the least likely, (1, 2, 3, 4, 5, 6, 7 and 34, 35, 36, 37, 38, 39, 40).

Michael John Sarnowski said...

I see, that people striking out on their own, as being more useful for finding out what is more precise or correct. However going with the herd, which may be an innate instinctual trait of humans, may be useful for ones own and the human species survival. It would therefor be very difficult to fight the herd mentality.

CapitalistImperialistPig said...

I find the distinction between "useful to describe observations" and "describing reality" to not be very useful as a guide to theory. One problem is that you cannot say anything useful about observations without some underlying theory that helps one interpret observations. Such theories are at least tentative theories of an underlying reality. Otherwise every observation is a unique sensory event with no relation to any other. We start with a built in theory of reality that assumes deep similarities among some kinds of events, and that observations made by others are similar to observations by ourselves and so on.

The problem with theories that assume the universe was created 10,000 years (or 10 microseconds) ago with pre-manufactured fossils, light from apparently ancient galaxies etc is not that they can be disproven but that they are ridiculous - or at least that they presume a creator with a malicious sense of humor.

When you argue against ideas such as Naturalness you are arguing about heuristics - ideas that might be useful in discovering theories - rather than a class of theories. Ditto for "mathematical beauty" and even falsifiability.

Kaleberg said...

I like your approach to physics. As far as you are concerned, physics is something humans do that lets them understand the universe in the sense of being able to predict and explain. I'm not 100% there with your use of the term "anti-realist". Your approach is that physics produces a set of logical postulates and a means for reasoning about them. Mathematics tells us that such a system is either inconsistent or there are at least two models that it describes. I get the impression that an anti-realist considers only the logical system and its consistency but doesn't worry about the models. Am I close here?

David Schroeder said...

When I first heard of the concept of inflation in the early Universe, years ago, I accepted it as gospel, and assumed the astrophysicists who conceived it had solid, irrefutable reasons to impute its former existence. I now see that the lay public needs to be cautious in accepting consensus views, which while strongly motivated to solve currently observable aspects of the Universe, are beyond the realm of testability.

Jack Stephens said...

There is nothing new about scientific theories pushing beyond the borders of extant knowledge which are often subsequently left in the the dustbin when new information appears. For example Lord Kelvin's theory that the sun is 20 million years old and the theory of ether to explain radiation transmission. It is just the nature of modern science to explore the unknown by creatively constructing theories in an attempt to make new information intelligible or old information more coherent. Occasionally one of these jabs into the unknown bears fruit such as General Relativity, but without this messy activity there would be no advancement in the sciences.

Jeffrey Baybick said...

The basic question is whether "on purpose" or "by accident."
The Creationist Young Earth is a theological doctrine that does not jive with the Biblical Account of Creation (as understood through the Received Tradition).

The Biblical account of creation is quite succinct; given to a scientifically non sophisticated with limited vocabulary. If one allows for this, there is striking similarity between the biblical account and modern day epistemology of cosmology and evolution.

The basic disagreement between modern day cosmology/evolution and biblical cosmology/evolution is the time frame. Although the biblical account refers to 6 days the question is what is a day from God's point of view. The biblical account according the Received Tradition is that the six days are the day from God's point of view and does not become Time as human understand it until the end of the sixth day with the emergence of humankind and the withdrawal of God from active creation.

The Received Tradition states that God is the habitation of the universe but the universe is not the habitation of God. The Psalmist says that a thousand years are but as a day that passes to God. This clearly indicates that Time should be seen from God's point of view and not man's. Logically, that being such as God, in which the universe sits, must see the world from a relativistic point of view for the first six days of Creation. And if one uses relativity the universe is about 6 days old.

Although humankind emerges at the end of the sixth day as the pinnacle of creation, the second chapter of Genesis tells us the emergence was through punctuated equilibrium (as the creature to become Adam is placed into a restricted area known as the Garden). The amount of time spent in the Garden of Eden is not stated. The Received Tradition states there were multiple lines of creatures that looked human but were not fully human. And that the descendants of Adam and Eve interbred with these creatures, as did Adam.

The age of the Earth is not approximately 6000 years but this is the age of civilization.

Uncle Al said...

Given Noether’s theorems, does physical theory simply arise from exact intrinsic symmetries? Is the Tully-Fisher relation bound by exact isotropic vacuum and exact conservation of angular momentum, hence dark matter as the simple answer?

Simple exact foundations of physical theory fail to exist - by observation.
Edward Witten, “Symmetry and Emergence,” arXiv:1710.01791

One physical property is emergent, unmeasurable, and shunned: strong arrow of time geometric chirality. A beautiful physics may be a simply explained young Earth justified by carefully not looking, followed by floods of paper.

Unknown said...

The previous posts about probabilities, 1.0000 vs 1.0001, etc , reminded me of this Feynman comment, from one of his popular books:

"You know, the most amazing thing happened to me tonight... I saw a car with the license plate ARW 357. Can you imagine? Of all the millions of license plates in the state, what was the chance that I would see that particular one tonight? Amazing!"

-- Tom H

tyy said...

I want to disagree. Do I?

I don't know.

Terry Taerum said...

On attempts to deliver 'proof' as compared to relying on 'strong(est) evidence', it's worth noting that what one person considers to be 'facts in evidence' (e.g. the painting is a forgery) may not be what we 'choose to believe'. What we choose to believe is based on facts founded on evidence born of assumptions expressed in language.

What I consider to be 'facts in evidence' that the earth is billions of years old is inseparable from my willingness to think in the language of my assumptions. Behind these assumptions are a complex scaffolding of languages interpreting genetics, geology, astronomy, physics... they are all interwoven, connected, and inseparable from one another. And yet, these languages remain distinct and separable - it would be foolish to treat a nucleus in chemistry as if it were the same as a nucleus in cell biology. In the same way, marrying science and the supernatural is not helpful.

Science has done brilliantly in the past century because it does not need to invoke the supernatural in order to explain and predict the natural. Using the supernatural to explain the natural is not required and, as such, is not useful in science.

Will Watchman said...

The Bible was written by man but are the infallible words of God. While the universe appears ancient, it is only 6,000 years old based on the genealogy of the Bible and by a constellation arrangement in Virgo on Sept 23rd called the Revelation 12 sign. The last time this sign occurred was 6,000 years ago.

Just as God created everything in six days and rested on the seventh so too shall mankind endure for 6,000 years and the last 1,000 years will be the millennial reign of Christ on earth after a seven year apocalypse known as the Tribulation Period for those who were not found worthy to partake of the rapture event.

While the day and hour of the rapture is unknown, it appears it is going to take place shortly after Israel becomes 70 years a nation this Nov 29th based on the parable of the fig tree in Matthew 24 as the final sign before the rapture. Most likely before 2018.

Only those whose sins have been atoned for by the blood of Christ and have been keeping their obligation to strive for righteousness shall be found worthy to escape all that is about to come to pass and to be able to stand before the Son of Man. We are almost there.

Tom Aaron said...

As a devout atheist I sometimes find myself defending Creationism. Not the theory but the derision it receives as being any more absurd than what many 'rational' scientists find acceptable. If one accepts that there is a god, intelligence, simulation maker, etc. that nudges one atom, then Creationism is no more silly. One can't be 'a little bit pregnant'. As soon as someone accepts some supernatural (as in outside of accepted physics) that can influence matter and energy then 'everything' is equally possible. The magic wand just needs to be waved.

In a muliverse universe of infinite possibilities is one of them the Universe of the Old Testament? Is it this one? I find this a spooky thought.

As a geologist, I understand time....billions of years of it! However, as a lay person struggling with quantum mechanics I'm open to the possibility that reality as we know it is an illusion or simulation. Perhaps our Universe wasnt created 6 thousand years ago but 6 minutes ago. Or...13.5 or 6 thousand are irrelevent as there is no 'time'. I'm no longer confident that our physics is driving along a valid track. I look forward to some General AI in the next few decades patting us on the head and setting us on course. I've gone from being a smug atheist to a more humble one.

bud rap said...

The fundamental problem with the standard model of cosmology is that it violates the underlying tenet of relativity that there is no universal reference frame. It does not matter that Friedmann employed the GR field equations, since he used them to model a Universe, a single, coherent simultaneous entity. That unitary assumption and its dependent redshift=recessional velocity assumption have left us with a cosmological model that bears no resemblance, in its particulars, to observed physical reality.

A proper GR description of the cosmos, even on a purely conceptual level, will probably require at minimum, two multidimensional reference frames. For instance, a three dimensional, local frame for matter (anything with rest mass) and a four dimensional, global frame for electromagnetic radiation might be a good place to start. Such a model, by design, would not contain a universal, unobservable, and inexplicable initial condition or a universal material expansion. As a consequence it might possibly bear at least a passing resemblance to the cosmos we actually observe.

UrDaddie said...

"...No, they aren’t not wrong, but they are useless..." ??

So they are wrong?

Ron Henzel said...

Sabine Hossenfelder wrote:

"Imagine planet Earth began its existence a mere 10,000 years ago, with all fossil records in place and carbon-14 well into decaying."

The problem with this is that I don't know of any Young Earth Creationists who believe that the fossil records were in place at the moment of creation. All the ones I know believe that the bulk of the fossil record is the result of Noah's flood.

Sabine Hossenfelder said...

Ron,

I did not claim that that's what Creationists believe. I don't know you what makes you think I did. I didn't even mention YEC.

nullnor said...

i wonder if believing in an external reality would cause a person to be less inclined to be an environmentalist? if a person would think they're not from this universe and someday they are going back where they came.. they might see the environment as a temporary thing.

recently i've been trying to look at the universe like it's an organism, or behaves like one sometimes. i guess that makes me an anti-realist? heh, like how someone said once.. how can a 3 pound mass of jelly contemplate infinity and think about it's place in the universe.

Dwight Thieme said...

Apologies, but I've never heard anyone say that the variant of Young-Earth Creationism you describe is 'wrong'. The most common evaluation on my radar (by far) seems to be 'not even wrong'. Even given the YEC assumptions, all you get is a variant of Roko's Basilisk argument ... first proposed some time earlier by some rather clever chap named Pascal. So YEC is not only not even wrong, it's not even original.

To your point on usefulness: IIRC, the deferent/epicycle model of planetary motions was still preferred as a computational model for some time after the Kepler's theory was known to be 'more correct'. A more recent historical example along the same line was the use of Newtonian mechanics -- not GR -- to calculate orbital trajectories for every one of the Apollo missions, forty-some years after GR was confirmed (yes, I know that this is a historical point of contention.) IOW, I'm not sure exactly what you're getting at here.

Sabine Hossenfelder said...

Dwight,

Apologies, but why does your experience count any more than mine?

Dwight Thieme said...

No, it doesn't, of course. But when you said:

I often hear people argue such creation-stories are wrong because they can’t be falsified, but this makes about as much sense as organic salt. No, they aren’t not wrong, but they are useless.

I am mystified as to your meaning. Is it that whenever you hear people argue such creation-stories this is the way they always argue and that you hear such arguments often? Is it that this is the argument you hear most of the time when this topic is discussed? Or is it that this is an argument that you don't hear relatively often when this topic is discussed, but you hear a lot of arguing over this and thus it is 'often'? For that matter, if only a single person makes that argument (or even none at all), you still could have made essentially the same post.

The point is thus not that my experience counts any more than yours, but that yours is no more privileged than mine. Thus I suspect it comes down to the type of people you hear these opinions from, why they make that argument, and why it matters enough to discuss it in the first place. The majority of the people I associate with are well-educated, relatively areligious, and only respond with this argument (to the extent they respond at all) when they are confronted by a YEC who asks this question. They are not especially prone to discuss it amongst themselves, if you take my meaning :-)

JimV said...

ppnl: in the case of coin flips, we have a probability distribution, the binomial distribution. I believe Dr. Hossenfelder was talking about cases for which we don't have a probability distribution.

Matias: the multiplicity of sums of lottery numbers would affect the chances of winning if there were prizes for matching the sum of the actual draw, but there aren't, so this has no effect. The chance of winning can be expressed as (probability of having the right sum)*(probability of having the right member of that sum-set). The latter term gets smaller the higher the multiplicity.

In general: Dr. SH was giving an example of an unfalsifiable theory, and happened to pick one with some similarities to what YEC's believe. I'll bet she wishes she had chosen some other example. Although, whatever one writes, probably some will misinterpret it.

Mozibur said...

I first read about the theory that the earth was created ten thousand years ago together with its fossils in the memoir by Edmund Gosse, Father & Son (it's a wonderful memoir by the way); it was a theory put forward by his father, who was deeply religious, in the hope that this would reconcile science and theology; instead, both sides simply laughed at him - they both found that narrative incredible.

I find inflation incredible, and even more incredible since it's postulates an unknown force that operated in the epoch of the very early universe; does this mean that, although we usually say that there are only four forces in nature - gravity, the strong and weak force and electromagnetism - we should add to that the inflationary force? Is it a force that will appear at only the very high energies, energies that were only found at the Big Bang itself?

Can I too add my voice in asking what disturbed you in the conversation with Niayesh; secrets are better by being divulged than kept!

Mozibur said...

There's a twist on Creationism called Occasionalism where the universe is not simply created at the Big Bang, but that it recreated at every moment; I quite like this position as 'democratises' creation - it's not simply an event that happened once, but happens all the time; and it seems to be consistent with QM, where measurements or rather interactions (a la relational QM) bring into existence elements of reality from an ephemeral indeterminism.

Christo said...

Hi Sabine Hossenfelder,

Sorry if someone has already mentioned/noted the following. As I understand it, and feel free to correct me, our current cosmological theories are drawn upon the theories with which cosmology can occur. Thus, why gravity is such an important force in cosmology, and why its' theory, the Theory of General Relativity, is prominently used - long range, inwardly acting, the the best theory of gravity we have, simple, etc.

So with that understanding I found this comment rather odd, "The problem is the story we presently have is already very simple. This really is my biggest issue with eternal inflation and the multiverse or cyclic cosmologies, bounces, and so on and so forth. They are stories, all right, but they aren’t simplifying anything. They just add clutter, like the programmer that set up our universe so that it looks the way it looks."

It seems to me that you can start of with a simple picture of a fundamental thing, and then build up the complexity with its' internal and external, multiple interactions. Human life would be an example of such a principle. You start off with quantum mechanics and chemistry to end up with complex proteins, interactions and cells resulting in us. Therefore, the bigger problem is with a better explanation of the fundamentals than with the emergent features consistent with the current accepted fundamentals.

So, if we had an accepted quantum theory of gravity, it would likely be simple which is what we desire. But, it could produce a complex theory of the universe. For example, Loop Quantum Gravity and the Bounce hypothesis. With LQG then, needs to be made testable predictions and have a greater explanatory power than GR, which as yet has not been achieved.

If this is standard information, than this is obviously not novel thoughts for you. Any deeper comments regarding this idea of complexity from simplicity in the sense of how I've brought it up?

In addition to this, I do agree with you about the Popperism/Falsification though. I've grown towards plausible explanations and Bayesian reasoning. I use to profess Falsification, but I became quite interested in measurements and probability theory which made falsification an obsolete philosophy. I've been reading 'Probability Theory: The Logic of Science' by E.T. Jaynes (But, I have yet to make it the quarter of the way through), which has honed my knowledge of probability theory and measurement, developing my philosophy.

I tend to explain proofs in this way also. Such as mathematical proofs require full information which is absent from the physicist, whom generally only have partial information from experiments/observations - the partial information we call evidence. Therefore, making the terms proofs and its' associations inconsistent with thinking outside of specific fields - philosophy and mathematics. Your title, "How do you prove that Earth is older than 10,000 years?", therefore shows inconsistency between what people want to know (Earth>10kya) and how they want to know it (by proof). A more useful question might be, "why should you accept the earth is >10kya?"

Regards

Sabine Hossenfelder said...

Christo,

You seem to have entirely misunderstood what I said. Yes, you can build up complexity starting from simple things, I didn't say anything to the contrary. Maybe re-read this.

nullnor said...

i wonder sometimes if the universe is building something. if you see a lot of examples of things starting simple and getting more complex. what is the most complex thing, DNA?

nullnor said...

it could be like the movie Time Bandits.. the most fabulous object in the world :)

nullnor said...

have you guys seen this video from jilli rose about the lord howe stick insect? i really like it. they just did the DNA and said it's not extinct. it's a crowd funded cartoon but true story... the universe is an organism. in my heart... in your heart.

can a universe exist without love? or order. or a plan. the answer is staring everyone in the face. where we come from. why we're here. were are we going. and together science and religion can help. we just haven't gotten farther down that road yet.

yeah i mean i get it. it's a toxic atmosphere. if you are a scientist, you don't ppl to think i am being subjugated by religion. everyone seeks truth. try not to forget that. and try not to lose sight of the really big questions. no-one is in a better position to answer them than all of you.

stay gold ponyboy.

https://vimeo.com/76647062

Matias said...


Sabine,

You mentioned that "the universe is in one microstate and has always been in one microstate". I believe you have said this somewhere before as well, but I couldn't find it in your blog. I am quite curious about this, can you elaborate on what do you mean? If you have some earlier blogpost about this topic, a link would be highly appreciated!


Also, I am very sorry about the off-topic commenting, I will stop after this.

JimV,

It doesn't matter whether you bet on the sum of the sequence or not, it will always exist and affect the outcome. I used it as an example for physics inclined people, because it shares features with temperature. Similarly in experimental work, it doesn't matter if we are measuring temperature or not, it will always affect our measurements.

So I will just leave a last example that should be very easy to understand. Let us say that we have a box from where we draw lottery numbers and all of the 18 643 560 possible combinations are there. What is the probability of drawing a single random sequence from all of these numbers? It is exactly 1, because whatever we draw from the box, it fulfills this criterion. Now we put the sequence back in and draw again. Now, the probability of drawing the same sequence again is exactly 1/18 643 560. Therefore, the probability of winning with a randomly chosen number sequence is 1*1/18 643 560, which we would expect.

What if we impose a condition on the first number sequence? Let us say that it has to be a sequence with all consecutive numbers. The probability of drawing such a sequence is 33/18 643 560, because there are exactly 33 different sequences with all consecutive numbers. Then, on the second draw you again have a probability of 1/18 643 560 to redraw the same sequence. So, the probability of winning the lottery with a sequence of consecutive numbers is 33/18 643 560*1/18 643 560 which is about 9.49e-14. In other words, you can expect to win at about every 1e13th draw, and if you play the lottery every single day then you can expect to win once after 2.9e10 years, which is about twice the age of the universe.