Log in

No account? Create an account

Previous Entry | Next Entry

Lubos Motl today has ranked his confidence in a whole lot of different statements about physics. I could imagine someone in the world disagreeing or agreeing with all of these statements, so it's a pretty cool list, and I find it very useful to hear what his different levels of confidence in these statements are. I wish more physicists had the balls to make their predictions as public and quantifiable like this:

Lubos Motl: Probabilities of various theories of fundamental physics

Some selected entries from his list, if you're too lazy to click:

99.999% - String theory is a mathematically consistent theory including quantum gravity, even non-perturbatively, at least in some highly supersymmetric vacua

99.995% - Black holes exist

99.99% - The Standard Model with neutrino masses is the correct description for all interactions of currently known particles in particle physics below 100 GeV with errors that can be parameterized as small corrections

99.9% - The Big Bang, an expansion of the Universe from a very small size and huge density, is a correct description of its history, at least from the age of one minute or so

99.5% - The LHC will find a Higgs boson

99% - Evaporating neutral black holes in the real world preserve information when described by the exact correct quantum mechanical model

95% - The standard model of cosmology including the cold dark matter and dark energy is correct up to small corrections - no qualitatively new large terms or concepts are necessary to explain the data

85% - String theory is the correct unifying theory, i.e. when dust is settled, it will be clear that the same equations that describe our Universe more accurately than any other theory also imply the existence of the well-known 10-dimensional and 11-dimensional vacua

50% - Supersymmetry will be found at the LHC

20% - The anthropic selection will remain the only constraint on the value of the cosmological constant that will be accessible to theorists by 2100

0.2% - The LHC will produce evaporating black holes

0.002% - A discrete model without a continuum limit is more fundamental and accurate a description of the world than any continuous model

0.002% - A model that studies quantum gravity separately from other sources and assumes that nothing beyond pure gravity exists will lead to true and valuable insights about the workings of the real world by 2100

0.001% - Doubly special relativity is a refinement of special relativity that becomes much more accurate in a wide class of phenomena

0.0005% - Basic postulates of quantum mechanics such as superposition principle and the method to obtain probabilities as squared amplitudes will be modified and the extension will describe a class of phenomena much more accurately than orthodox quantum mechanics

0.0002% - Loop quantum gravity is able to describe physics whose low-energy limit is general relativity while avoiding an infinite number of fine-tunings

0.0001% - Loop quantum gravity, with the metric as the only and well-defined degree of freedom and with quantized area, is a correct description of gravity in the real world at the Planck scale

0.00001% - One of the ESP phenomena measured in the Princeton lab actually exists and can be measured again with a similar equipment

For the most part, I think his estimates are pretty dead on. There are a few things I would move up or down, but I'm surprised at how closely I would have picked most of these numbers... even knowing much less about the details of some of them. His estimate for detecting SUSY at LHC, for instance, is 50%, which matches my publically stated estimate exactly (as captured on video in my famous bet with mauitian at Burning Man last year). The likelihood of string theory being the correct theory of quantum gravity (or at least one limit of it), I also would have picked to be around 85%. (I'm somewhat surprised that he didn't put it higher, though, as sometimes he seems to be somewhat of a string theory zealot.)

A few things that I would have adjusted up or down from his values (these are based on my own personal knowledge, not on the sum total knowledge of everyone in physics, so some of it is based on my own ignorant guesses):

1.) His statement about the Standard Model as it stands, I would revise up above the highest on the list. He originally had a stronger statement, but weakened it when someone pointed out that axions being found would negate his original statement. I think he should have added several more 9's after the decimal after he weakened it.

2.) I would adjust the "Big Bang" statement upwards. I certainly think it should be above the statement about trusting "semiclassical gravity" :) Actually, I can't imagine how this statement could be wrong... at worst, we might find there is a better way to describe it, but it's still going to be a "correct" description in at least some sense. I think I'd put it at 99.997%, just above "black holes exist".

4.) I think I'd adjust "GUT theory with unified gauge interactions" up a couple percentage points. I also might adjust "SUSY at the GUT scale or lower" up a few percentage points.

5.) "anthropic selection will remain the only constraint on the cosmological constant until at least 2100". I'm not entirely clear on what he means by this. For instance, there are things you can add to Weinberg's anthropic calculation to tighten up the bound, so you could say you added some "non anthropic" thing, but the main constraint is still anthropic. I would put it at around 99.9% that the main constraint will always be anthropic. As for whether we will add minor corrections or new constraints to that, I don't think that's a terribly interesting question, and can't really estimate a likelihood for it reliably, but his 20% chance that we will add no more constraints at all sounds as good a number as any.

6.) I think I would adjust the chances that LHC produces "stringy modes" downard at least an order of magnitude from his 0.2%, but leave the probability for producing black holes where it is.

7.) I would adjust the chances of a "discrete model" being more fundamental than a continuous model up to around 5%.

8.) Due to my ignorance of loop quantum gravity, I would not be able to put it at as low a number as him. Anything I don't know much about has to be closer to the 50% mark indicating my ignorance. So for me, I would put it at up around 0.5% (1 chance in 200, something worth it for a few people to spend time looking into) rather than the 0.0001% he puts it at. I'm willing to believe that some people have studied those types of models enough to be that confident they don't work, but until I've actually studied them myself I will give them the benefit of the doubt.

9.) I would adjust the "Princeton ESP lab" (which is misleading since it was really just one guy funding his own experiments, not anyone supported by the university... and he was a psychologist, not a physicist) downward from 0.00001% to 0.00000000000001% (1 chance in 10^16). In other words, there is no reasonable excuse in the world for a respectable university to allow crackpot ESP research to go on using their webpage and their laboratory... and I think it's irresponsible of him to compare loop quantum gravity to it by putting it only an order of magnitude away, as ESP is clearly, much, much worse. They are not at all the same level of kookiness. (Thankfully, this lab has now been closed. A few years ago, when I first heard of this, I strongly considered sending in a complaint letter saying how poorly it reflects on the university, but enough others must have done it that I didn't have to.)

[Update: sorry, I was under a bit of confusion regarding #9. The guy I was referring to in number 9 (and in the threads here) was Roger Nelson, the head of what he calls the Global Consciousness Project; apparently, this was only a spin-off of the "ESP lab" that was shut down... that presumably was what LM was referring to. I never realized these were sort of two different groups. Both working on psychic/ESP type stuff at Princeton. Here's a link about the main ESP lab being shut down (http://www.theregister.com/2007/02/13/princeton_esp_lab/). I'm not sure now whether or not the GCP was also shut down too, but I would assume/hope so.]


( 32 comments — Leave a comment )
Apr. 18th, 2007 07:54 am (UTC)
All these 99s and 00s remind me of Eliezer Yudkowsky's paper on human cognitive biases where one of the phenomena emphasized was the consistent overconfidence of experts asked to give probabilities like that.

FWIW, my guesses would be closer to your figures than Mottl's.
Apr. 18th, 2007 08:40 am (UTC)
Well, shouldn't someone investigating ESP be a psychologist rather than a physicist? They'd have much better ideas how to do controls for behavioral experiments and the like. Though if they got any significant results, they'd probably want to call in physicists and all sorts of other scientists to figure out if there's anything like a plausible mechanism.
Apr. 18th, 2007 10:56 pm (UTC)

Well, shouldn't someone investigating ESP be a psychologist rather than a physicist?

Well, nobody should be investigating ESP, but some of the articles I read about it gave the impression that he was a physicist, which indirectly gives the impression that it's not insane from the perspective of someone who understands physics... which is a dangerous impression to give. :)

They'd have much better ideas how to do controls for behavioral experiments and the like.

This guy had no idea how to do controls or statistical analysis (see my comment to killtacular).
(no subject) - easwaran - Apr. 18th, 2007 11:15 pm (UTC) - Expand
(no subject) - onhava - Apr. 19th, 2007 04:00 am (UTC) - Expand
(no subject) - easwaran - Apr. 19th, 2007 04:42 am (UTC) - Expand
(no subject) - spoonless - Apr. 19th, 2007 04:44 am (UTC) - Expand
Apr. 18th, 2007 09:29 am (UTC)
Hi Jeff, are you a string theorist?

I strongly agree with your statement:

I wish more physicists had the balls to make their predictions as public and quantifiable like this:

Unfortunately, Lumo's confidence is also unshakable, and he has a very strong tendency to deny and censor any information that he doesn't like, and regardless of merrit.

I've known him for a number of years now, and the only time that I ever got him to admit that he was wrong about anything was when I corrected him in the moderated research forum and he had no control over it.

But I've lost all respect for him since.

Anyway, sorry to barge-in un-invited, but I found you via a key word search.

Maybe Lubos thinks that god is coming in the year, 2100?.. ;)

Apr. 18th, 2007 06:14 pm (UTC)

Hi Jeff, are you a string theorist?

I'm just a graduate student, but my advisor is a string theorist (actually, he was also Lubos's advisor). I've learned enough about strings to think string theory "looks very promising" but I have not published any work myself on string theory (my current work revolves around constructing a particular GUT that is compatible with certain TeV-scale assumptions). So like I say, a lot of my levels of confidence should be taken with a grain of salt since I'm less familiar with a lot of these things than actual people who have been working in the field for a while. It does look like the next project I work on will more directly relate to string theory, but the truth is, it's hard to say where the boundry between strings and particle physics is any more... everything is interconnected.
(no subject) - island01 - Apr. 18th, 2007 07:59 pm (UTC) - Expand
(no subject) - spoonless - Apr. 18th, 2007 11:10 pm (UTC) - Expand
(no subject) - island01 - Apr. 19th, 2007 01:25 am (UTC) - Expand
Apr. 18th, 2007 10:39 am (UTC)
Pffft. The part of this that is "quantifiable" is pretty bullshit quantities, IMHO. I don´t think it makes a lot of sense to assign subjective credence in numerical form. Maybe you can give ordinal credences, but I don´t know that assigning pure numbers makes sense.
Apr. 18th, 2007 06:00 pm (UTC)
I agree that beliefs are usually more complicated than just 1-dimensional numbers on a scale. (For instance, there are usually a lot of ways in which things can be "kind of wrong" about something and it's not clear whether to count it as wrong or right.) But I think the clearer the statements you make, the easier it is to assign actual numbers to them. Most of these are pretty clear statements, although there are a few such as the one about the anthropic prinicple and the one about the big bang that I feel are not clear statements... so I just sort of averaged the different interpretions I could think of for them to the best of my ability.

I think percentages do make sense if the statements are clear enough. The way I think of it is, out of the billions of beliefs I have had over the course of my life, does the strength of my conviction for this particular statement compare to the type of belief that has turned out wrong 1 in 10 times, 1 in 100 times, or 1 in 1000?, etc. You could also think of it as a question about how I would act if I were offered a bet on it. If I'm 90% sure of something, then that means I'm willing to match $10 for every $1 that someone puts up to me in a bet (minus a fee for the effort of thinking about it, risk, hidden costs, etc.) up to the point where it starts to get up to values I couldn't afford to risk.

So yes, there are problems with assinging numbers to subjective confidence levels, but a.) it's fun, and b.) it does have some validity, and c.) it's useful to know where other people stand on things compared to other things, rather than trying to sift through the various statements they've made in the past and guess.
Apr. 18th, 2007 06:22 pm (UTC)

I'm 90% sure of something, then that means I'm willing to match $10 for every $1 that someone puts up to me in a bet

oops, I mean $9 of course :)
Apr. 18th, 2007 11:11 pm (UTC)
The betting interpretation is the best way to make sense of it that I know. If the betting behavior is right, then the numbers have to satisfy the probability axioms, or else you're vulnerable to "dutch books", where you'll accept a series of bets individually that guarantee you lose money overall.

I think the strongest challenges point out that people don't buy and sell bets at the same price. The natural way to deal with these cases then suggests that people don't have a single probability function but rather a whole set of them. You can talk about the ranges of values assigned to particular propositions, but you lose some information about independence and correlations and the like if you don't consider the complete probability functions.
Apr. 18th, 2007 09:11 pm (UTC)
Very interesting, and also illuminating to non-physicists :).

"95% - The standard model of cosmology including the cold dark matter and dark energy is correct up to small corrections - no qualitatively new large terms or concepts are necessary to explain the data"

Why would you mostly agree with this? It seems that y'all have no good description of what "cold dark matter" and "dark energy" is, right? And yet, they appear to play pretty significant roles in the universe. If we can't describe them now using our current conceptual framework, might it be somewhat likely that new concepts will come into play to so describe them and why they are, at least right now, mysterious? So 95% might be too high? I dunno, I don't really understand much of it.

And, ESP like extra-sensory perception, or ESP meaning something else in physics?
Apr. 18th, 2007 10:49 pm (UTC)

It seems that y'all have no good description of what "cold dark matter" and "dark energy" is, right?

For dark matter, it's just an issue of identifying which high energy particles contribute to it, and what exactly their masses are. We have a whole zoo of different candidates, so one of them or several of them, or even all of them could be what comprises the dark matter. None of them require any fundamental changes to our current model. I think in some ways, the media may have misled people into thinking dark matter is more mysterious than it really is. It's actually necessary for things like galaxy formation, so if we hadn't seen dark matter there would have been huge problems with the current models. It shows up quite naturally in many models of low-energy supersymmetry which is something string theorists were expecting anyway. I would say if anything, it's confirmation that our current theories are on the right track.

For dark energy, pretty much everyone agrees that it is either vacuum energy, or vacuum energy plus a cosmological constant, which are both effectively the same thing. There are two main unanswered questions about dark energy. One of them is whether it has changed over the course of the universe or remained constant. And the other is why the current value we're measuring is so incredibly tiny (about 10^-120 in "natural units"). The second of these is often called "deep and mysterious" and there are many people who have suggested radical things such as large-distance modifications of gravity to account for it. However, I'm pretty confident that I already know the answer to why it's so small: the correct answer is the anthropic principle. I think Lubos Motl is more skeptical of this, but I would put that up there at 99.9%.

At any rate, both dark matter and dark energy fit in very well to our current models, so unless there are big unexpected surprises coming, there is no good reason to expect the current ideas will need to be modified. In contrast, something like MOND (an alternative to dark matter that has since been ruled out as a possible explanation for it) if it had been true would have completely overturned our current models of physics, and been shocking to nearly everyone... and would have required a whole lot of new bizarre concepts.

Anyway, I would agree with the 95%.

And, ESP like extra-sensory perception, or ESP meaning something else in physics?

ESP meaning extra-sensory perception. In particular the guy claimed to have set up a random number generator that suddenly got "less random" during September 11th due to all of the bad vibes coming from New York and all over the world. Unfortunately, the media made it sound like this guy was actually a physicist at Princeton, when in fact he was a crazy psychologist with a BS in physics who had conducted his own experiments with private funding and set up his webpage using Princeton's servers. Looking at his data, he had a severe lack of understanding for statistics... not to mention, the craziness of his ideas. Unsurprisingly, they finally shut him down and told him to leave, even though he never admitted that his data was complete bullshit.
(no subject) - killtacular - Apr. 19th, 2007 05:50 am (UTC) - Expand
(no subject) - spoonless - Apr. 19th, 2007 08:55 am (UTC) - Expand
(no subject) - killtacular - Apr. 20th, 2007 04:30 pm (UTC) - Expand
(no subject) - spoonless - Apr. 20th, 2007 06:26 pm (UTC) - Expand
(no subject) - spoonless - Apr. 20th, 2007 07:58 pm (UTC) - Expand
(no subject) - spoonless - Apr. 21st, 2007 06:07 pm (UTC) - Expand
(no subject) - spoonless - Apr. 21st, 2007 06:11 pm (UTC) - Expand
Apr. 19th, 2007 04:19 am (UTC)
a couple more comments on dark matter
It might also be worth mentioning that when he says "The Standard Model of Cosmology including Cold Dark Matter and Dark Energy", that's redundant. The phrase "Standard Model of Cosmology" is something that refers to the consensus that has formed within the past 5 years or so, that the energy in the universe is primarily dark energy, with some cold dark matter, and a tiny bit of regular matter. These percentages have been measured fairly accurately, and all of the data fits and looks very nice... it's perhaps the first time in history we've had such a nice model of cosmology, with no big missing holes or pieces. So as I mentioned in the previous comment, the main task left is finding which particles compose the dark matter... however, that answer will not really affect the cosmology. Or if it does, only very slightly... since the thing that matters in cosmology is the mass/energy density, not the details of how the particles making up the mass behave.
Re: a couple more comments on dark matter - killtacular - Apr. 19th, 2007 05:58 am (UTC) - Expand
two different Standard Models - spoonless - Apr. 19th, 2007 08:30 am (UTC) - Expand
Re: two different Standard Models - killtacular - Apr. 20th, 2007 04:32 pm (UTC) - Expand
Apr. 19th, 2007 01:29 am (UTC)
the correct answer is the anthropic principle, but I would put that up there at 99.9%.

Yeah, because it's just an energy conservation law.
Apr. 19th, 2007 09:26 am (UTC)
There is no standard model of cosmology yet.

There IS a most conservative mainstream approach, which looks nothing like what's been assumed here.

And there is a model that cutting-edge theorists most like, which is wrongly being assumed to be the non-existent "standard model".
Apr. 21st, 2007 02:41 am (UTC)
The term "Standard Model of Cosmology" is fairly new, and obviously doesn't have the long history of use that "Standard Model of Particle Physics" does (and is on nowhere near as firm a footing). But I do think it's starting to gain acceptence... at least some people are using that term. I suppose calling it Lambda-CDM would be better, in case it turns out the story is more complicated.

What do you mean by the statement that there are more "conservative" approaches? Lambda-CDM is the most conservative that fits all the data (for just about any sense of "conservative" I can imagine). Anything else is a bit of a long shot/speculation.
( 32 comments — Leave a comment )


domino plural

Latest Month

May 2017


Powered by LiveJournal.com
Designed by Lizzy Enger