The Big Bell Test needs your help

November 29, 2016 9:51 AM

Some scientists are hoping that you can help them prove Einstein wrong by choosing 1 or 0s!
posted by of strange foe (53 comments total)

So I just served as a RNG for some scientist types? Fine, I suppose. I'm guessing their IRB approval took far longer than it should have, all things considered.
posted by RolandOfEld at 9:58 AM on November 29, 2016 [3 favorites]

Sadly, I can't find it in myself to scroll through the twitchy mobile interface to find out if this is interesting or not. Why do people saddle their sites with these gaudy displays that never render right in the wild?
posted by howfar at 10:01 AM on November 29, 2016 [14 favorites]

I'm a little allergic to low density media, so you can read through it at this link:http://thebigbelltest.org/#/science?l=EN.
posted by the Real Dan at 10:03 AM on November 29, 2016 [8 favorites]

What value does having people generate random numbers do? People are terrible at randomness. Computers are only sort of good at it, but much better than people.
posted by jacquilynne at 10:07 AM on November 29, 2016 [1 favorite]

Hmm. Aren't people bad at generating randomness? Quickly pick a number between 1 and 10 before you click that link...
posted by clawsoon at 10:07 AM on November 29, 2016 [4 favorites]

Reading the FAQ, they do have a physical RNG in addition to human input. This RNG was used in a prior experiment in 2015. The FAQ remains pretty vague about why the physical RNG isn't sufficient. Here's the prepublication version of the 2015 paper published in Nature which may address this question (I'm still reading).
posted by muddgirl at 10:23 AM on November 29, 2016 [1 favorite]

> I'm a little allergic to low density media
> so you can read through it at this link

Bless you.
posted by hank at 10:26 AM on November 29, 2016 [5 favorites]

I understand why it is the way it is, and it's a reasonable way to present it to an all-ages audience that doesn't know anything about QM, but reading popular science explanations is like reading an article that was google translated from English into Hungarian and back into English.

It all sounds very poetic but you'd save a lot of time if you just said this is generating randomness to measure entangled pairs.
posted by danny the boy at 10:26 AM on November 29, 2016 [5 favorites]

Unlike electrons or protons or the Higgs boson, which are perfectly interchangeable particles that behave similarly under the same conditions, every human-being acts genuinely on his/her own, and this is very valuable for the Bell test requirements.

So free will is presumed?
posted by Obscure Reference at 10:33 AM on November 29, 2016 [6 favorites]

This is a test of Bell's Inequality Theorem. Basically it states that whatever weirdness is going on with Quantum Mechanics, behind the scenes there must be something non-local. I.e., faster than light. Two people, spread across many light years, performing a coordinated experiment on an entangled system will appear to instantly influence the other's result. Conveniently (or inconveniently if you're trying to build a time machine), this occurs in a way that prevents FTL. This is because the correlation only shows up in a probabilistic way-- sort of like if I flip a coin and get heads, you'll always get tails (the actual experiment is a little more complicated).

Why use people if people are bad at generating randomness? Probably because one of the "outs" for Bell's Inequality, is super-determinism:

According to Bell:
There is a way to escape the inference of superluminal speeds and spooky action at a distance. But it involves absolute determinism in the universe, the complete absence of free will. Suppose the world is super-deterministic, with not just inanimate nature running on behind-the-scenes clockwork, but with our behavior, including our belief that we are free to choose to do one experiment rather than another, absolutely predetermined, including the ‘decision’ by the experimenter to carry out one set of measurements rather than another, the difficulty disappears. There is no need for a faster-than-light signal to tell particle A what measurement has been carried out on particle B, because the universe, including particle A, already ‘knows’ what that measurement, and its outcome, will be.

I believe the Many-Worlds Interpretation does away with non-locality in Bell's Theorem a similar fashion.
posted by justkevin at 10:35 AM on November 29, 2016 [8 favorites]

People asking me to pick a number between 1 and 10 will always get a non-integer from me. Pi is usually a good choice.
posted by Confess, Fletch at 10:36 AM on November 29, 2016 [2 favorites]

People asking me to pick a number between 1 and 10 will always get a non-integer from me. Pi is usually a good choice.

I picked Rho.
posted by radwolf76 at 10:37 AM on November 29, 2016

I'm a little allergic to low density media

Now I have a word for that thing I hate!

See also: audio and video without transcripts. I read faster than you talk, stop wasting my time.

Especially you, Slate. Any 'shortcut' that can be expressed in a single sentence does not need a video.
posted by leotrotsky at 10:43 AM on November 29, 2016 [29 favorites]

Pi and Rho are not numbers! They don't have digits 0, 1 ,2 etc..
posted by blakewest at 10:43 AM on November 29, 2016

People are terrible at randomness

True! For example, I predicted this thread would be full of grumbling about the linked site and the methodology therein.
posted by beerperson at 10:44 AM on November 29, 2016 [16 favorites]

I don't think they are using the technical word "random" correctly. Anything a human chooses is by definition systematic. I can predict my own responses. It's just that you can't. My choices may be desultory, but definitely not random.
posted by Mental Wimp at 10:47 AM on November 29, 2016

Pi and Rho are not numbers! They don't have digits 0, 1 ,2 etc..

They aren't rational. I believe they are called irrational numbers.
posted by Mental Wimp at 10:48 AM on November 29, 2016 [7 favorites]

Confess, Fletch People asking me to pick a number between 1 and 10 will always get a non-integer from me. Pi is usually a good choice.

Feigenbaum's (first) constant, or maybe a hyperreal like say [2]+[harmonic sequence], or "x in (1,10)", depending on how annoying I want to be about the ambiguity of the word "number."
posted by yeolcoatl at 10:50 AM on November 29, 2016 [2 favorites]

Why is it surprising that measurements on entangled particles are correlated? If the particles' characteristics are set correspondingly at the time they are entangled, then no FTL communication would be needed when they are measured. I am probably misunderstanding something here.
posted by DarkForest at 11:07 AM on November 29, 2016 [1 favorite]

I read faster than you talk, stop wasting my time.

leotrotsky, you're my hero.
posted by I_Love_Bananas at 11:09 AM on November 29, 2016 [9 favorites]

I bet I can type 0's and 1's more randomly than you can!
posted by straight at 11:14 AM on November 29, 2016

I got to the end and I guess it turns out this is just a demo? Typing 100 1s and 0s in 20 seconds without just leaning on one of the keys turned out to be trickier than I would have thought.
posted by Copronymus at 11:18 AM on November 29, 2016

Pi and Rho are not numbers! They don't have digits 0, 1 ,2 etc..

Pi and Rho are not integers. They are numbers or, more specifically, names we give to specific numbers.
posted by jacquilynne at 11:27 AM on November 29, 2016 [1 favorite]

Wow, justkevin, thanks for the layman's explanation, which I could not extract from the BigBellTest website.

So that makes this a pretty scary experiment, if it proves Bell right.
posted by beagle at 11:32 AM on November 29, 2016

My hands are SO TIRED now. Stupid Oracle.
posted by sldownard at 11:33 AM on November 29, 2016

I've always rather liked gamma (~ 0.57721) the limit of the difference between the harmonic series and the natural logerithm, as it isn't even known if it is irrational.
posted by Death and Gravity at 11:48 AM on November 29, 2016

I've been playing this for an hour and can't figure out how to upgrade my accelerator. Has anyone beaten the Boson King yet?
posted by beerperson at 11:48 AM on November 29, 2016 [7 favorites]

I really don't understand the choice to use humans here. If you ask 16 people to pick 4 random binary digits, probably 0 of them will produce 0000 or 1111 as their output, when you'd prefer one of each. Humans are really bad RNGs.
posted by 0xFCAF at 11:54 AM on November 29, 2016

Why is it surprising that measurements on entangled particles are correlated? If the particles' characteristics are set correspondingly at the time they are entangled, then no FTL communication would be needed when they are measured.

It's not just that they are correlated, it's that there are two different predictions on how they will be correlated (one based on local realism, and the other on quantum mechanics), and both predictions can't be true.

I really don't understand the choice to use humans here. If you ask 16 people to pick 4 random binary digits, probably 0 of them will produce 0000 or 1111 as their output, when you'd prefer one of each. Humans are really bad RNGs.

They've already done the experiment with a pretty good physical RNG, but what if this "hidden variable" that classical theorists are looking for affects both the RNG and the particles? Then what we think are independent events, really aren't. I think the theory is, if we had 300 million (admittedly pretty bad) number generators, instead of just one, the likelihood that our choices are independent of the particle spin should be higher.
posted by muddgirl at 12:25 PM on November 29, 2016 [4 favorites]

Also, it does look like they are doing a fair bit of scrubbing for predictability. On my first mission I got an unpredictable score of 43 out of 50.
posted by muddgirl at 12:28 PM on November 29, 2016 [1 favorite]

If you want random binary data with human influence, just feed live posts from Twitter into a hashing function.
posted by 0xFCAF at 12:34 PM on November 29, 2016 [2 favorites]

Why not run a Bell-type experiment with a CSPRNG (cryptographically secure pseudo-random number generator)? The "hidden variable" of local realism can't affect the output of the CSPRNG (it's deterministic), so when the Bell experiement turns out just like all the rest but you still don't want to give up local realism, you are forced to conclude the "hidden variable" has a structure just like the specific cryptographic algorithm of your CSPRNG, which is a crazy conclusion. (and even crazier once you perform the experiment using 2 or 3 unrelated CSPRNGs)
posted by the antecedent of that pronoun at 12:51 PM on November 29, 2016 [2 favorites]

probably 0 of them will produce 0000 or 1111 as their output

Jokes on you! I inputed nothing but 0's into their little test. So there. :-P
posted by oddman at 12:53 PM on November 29, 2016 [4 favorites]

Tthe idea that these theories about possible hidden variables are crazy or outlandish is actually discussed frequently in the scientific debate, as in this paper on the "memory loophole":

One might wonder what the point of considering all these loop holes is. Each seems to involve more conspiracy on Nature’s part than the last, and none of them appears to lead to plausible physical models. Given the importance of the Bell-type experiments, however, and their consequence s for our world view, we feel that it is important to analyze the experiments as rigorously as possible and in particular to distinguish between logical impossibility and physical implausibility of the models.

There is another more practical motivation [7,8]. It is well known that quantum key distribution schemes which use entanglement have significant security advantages over other schemes; they can also be extended by the use of quantum repeaters to allow secure key distribution over arbitrary distances. The security of these schemes relies crucially on the fact that the states created and measured are genuinely entangled. The most obvious and seemingly reliable way to verify this is to use Bell-type tests as security checks within the protocols. However, any such tests need to be interpreted with care. If a quantum cryptosystem is acquired from a not necessarily reliable source, or possibly exposed to sabotage, then a cautious user must consider the possibility that devices have been installed which use classical communication to simulate, as far as possible, the behaviour of quantum states, while allowing third parties to extract illicit information about the key. Such devices effectively define a local hidden variable model, and the usual criterion of physical plausibility no longer applies.

posted by muddgirl at 1:37 PM on November 29, 2016 [1 favorite]

I'm curious how local hidden variables, super determinism, strings and whatever else compare to just accepting that there are truly random phenomena in the universe? I take it super determinism is not being used in the sense of deterministic vs. stochastic, but more some kind of 'super non-causality'? For a moment I thought maybe true randomness would turn out to be indistinguishable from hidden variables, but now I've talked myself out of that idea..

I've skimmed over the notion of entanglement many times, always a bit bewildered at the assumptions being made. Like when a high school student asks what happens if you take all of the electrons in all the atoms in this litre of air at STP, and put them over here, and then take all of the neutrons and protons and put them over there. I may or may not have been that high school student :P The correct answer is, of course, you can't actually do that. Go ahead, try.

CBC's Ideas had a great show just the other day: Is That All There Is?
posted by Chuckles at 2:02 PM on November 29, 2016

probably 0 of them will produce 0000 or 1111 as their output

Jokes on you! I inputed nothing but 0's into their little test. So there. :-P

Same here. You way underestimated my laziness.

I didn't get very far though due to the interface and narrative I just couldn't follow.
posted by bongo_x at 2:10 PM on November 29, 2016

e FTW!
posted by mikelieman at 2:12 PM on November 29, 2016

It's not just that they are correlated, it's that there are two different predictions on how they will be correlated (one based on local realism, and the other on quantum mechanics), and both predictions can't be true.

What are the two different predictions? I've never been able to understand well enough how the two predictions differ in order to understand what was being tested. All I've been able to gather is that they are slightly different probabilities, but I can't understand why.
posted by Hactar at 2:24 PM on November 29, 2016

Quantum physics seems to say something different: that the act of observing the world can change it. Niels Bohr, a towering figure in quantum mechanics, even claimed that observables such as "the position of the atom" have no meaning until someone measures them. If this is true, the act of observation at least alters and maybe creates the world, quite the opposite of an independent reality. Physicists and philosophers have debated Bohr's interpretation of quantum mechanics since it was published in 1927.

I am not a scientist but this reeks of some kind of new age quackery. The claim that observation alters the world makes sense because you have to interact with something to measure it. The claim that the world coalesces when WE observe it sounds like BS.

How do you define observe? When you look at something it's photons bouncing off the object and entering your eye to interact with chemical in your eye that creates a nerve impulse that (through a series of other electrochemical reactions) finally arrives in your brain where you realize you are looking at a particular object.

Is observation when the photon enters your eye? When it interacts with your photo receptors? When your visual cortex assembles the image? When you are aware of something in your visual field? When you remember the name of said object?

What is the difference between a person looking at an object and a frog looking at an object - a camera - a space probe from a long dead civilization - a single molecule of hydrogen being heated by the impact of a reflected photon?

I am skeptical of any theory that tries to prescribe some great meaning to our existence; that humans are the true arbiters of reality is a bit too religious for me.
posted by kzin602 at 2:34 PM on November 29, 2016

Why not run a Bell-type experiment with a CSPRNG (cryptographically secure pseudo-random number generator)? The "hidden variable" of local realism can't affect the output of the CSPRNG (it's deterministic)

The results of the "loophole-free" experiments of 2015 already necessitate that hanging on to local realism requires one to postulate a hidden variable with extraordinary power and reach. In particular, the experiment by Shalm et al. tested two different sources of random noise as well as a "Cultural" pseudorandom source:

To generate the file for Alice, we XORed the binary data string of “Back to the Future 1”, “Back to the Future 3” (which was used in reverse order), a concatenation of episodes of “Saved by the Bell” (see below), and a concatenation of 1 × 10⁹ digits of π after applying a modulo 2 operation with “Monty Python and the Holy Grail”.
...
To generate the file for Bob, we XORed episodes of “Doctor Who” (which were all concatenated together), “Back to the Future 2”, and the concatenation of Leonard Nimoy: Star Trek Memories with episodes of “Star Trek” (this complete file was used in reverse order).

A CSPRNG requires a seed value, if both sides of the experiment are using the same seed you can't rule out the possibility that excess correlations between the results of the measurements wasn't the result of a correlated choice of measurements. If the experimenter chooses a different seed for each side, you're basically performing a small version of the "human provides randomness" experiment, as the experimenter's choice of seeds is ultimately the choice of measurements to make. If you seed the CSPRNG with a value from a "true" RNG, you're repeating the earlier 2015 experiments which used physical RNGs as the source of randomness to determine the choice of measurement to make.
posted by RichardP at 2:34 PM on November 29, 2016

That final Psychic level took me forever! I guess this proves psychics exist.
posted by Sparx at 2:48 PM on November 29, 2016 [1 favorite]

The claim that the world coalesces when WE observe it sounds like BS.

I've always taken it as the world we experience coalesces when we observe it. Whatever the actual world is may not change at all. When I look at a bunch of dots and see a picture of a frog then it's a picture of a frog. The dots didn't change.
posted by bongo_x at 3:39 PM on November 29, 2016 [2 favorites]

Now I have a word for that thing I hate!
See also: audio and video without transcripts. I read faster than you talk, stop wasting my time.

I'm fourthing this. GOD I hate five minute videos (sometimes with ADS), complete with lengthy intro animation, personal introduction, pseudo-amusing factoid, etc., until they finally get to THE POINT in the last 30 seconds. You who make these! I banish thee to the Plane of Gottfried!
posted by JHarris at 3:53 PM on November 29, 2016 [7 favorites]

So, we have a 'MeFi' group on this, yes?
posted by motty at 4:26 PM on November 29, 2016

If particles can be in a superposition, then so can people.

If a person looks at a particle in superposition, that person then becomes part of the superposition.

This is the many-worlds theory. It is really not that complicated once you come to terms with the idea of superposition itself.
posted by jnnnnn at 5:36 PM on November 29, 2016 [1 favorite]

I have a Geiger counter, some radioactive sources and a Raspberry Pi I can configure as a USB input device, so I could put in proper random 1s and 0s as if I were typing them. (Or an avalanche diode, or the thermal noise from one of my radios, or...)

Wonder what that would do to the experiment...
posted by Devonian at 5:51 PM on November 29, 2016

RichardP: "In particular, the experiment by Shalm et al. tested two different sources of random noise as well as a "Cultural" pseudorandom source:"

Amazing! The selection of sources is pretty interesting to me. Back to the Future trilogy? Sure, of course. Doctor Who episodes? I'm more of a Tom Baker guy but these Matt Smith episodes are fine and totally understandable. Two documentaries about Star Trek? Ok, that's a little weird but maybe they didn't have any actual episodes lying around. Thirteen non-consecutive episodes of Saved by the Bell? Alright, who picked those?

Also, they describe using "the binary string of Back to the Future 1, [etc...]" (emph. added). Except, there isn't one binary string representing it. Which version of the movie did they use? Was it from laserdisc, DVD, Blu-Ray, or digital download? Or did they digitize an old VHS recording off of a network TV broadcast? If it was DVD, was it the 2002 release, 2009 release, 2010 25th Anniversary Edition, or 2015 30th Anniversary Trilogy Edition? Or was it a shady torrent rip? Was it encoded in H.264? MPEG4? RealVideo? Reproduction-minded scientists want to know.
posted by mhum at 7:27 PM on November 29, 2016

Jokes on you! I inputed nothing but 0's into their little test. So there. :-P

Loser! Team 1 all the way!11!!oneoneone!!1!
posted by sebastienbailard at 7:52 PM on November 29, 2016 [4 favorites]

How about some balance? #team (01)*
posted by Pyry at 8:24 PM on November 29, 2016

Is there a more reliable way to generate a 1 or a 0 randomly than a coin flip?

For speed take a pack of cards and just use the #'d aces through tens and shuffle and flip 'em over one at a time and an even number gives a zero and an odd number gives a one. Since every shuffle will generate exactly 20 zeroes and 20 ones this is highly non-random by some measure but I have not suffered through sufficient statistics classes yet to tell you what this measure is named.
posted by bukvich at 9:58 PM on November 29, 2016

On the annoyance of explainer videos with long intros and people talking introspectively into the camera or whatever, I installed a chrome extension called "Video Speed Controller"
It let's you speed up any video. As a result I watch most Youtube videos at at least 1.6 often much higher.
(Also podcasts and sometimes TV shows, once you get used to it normal speed seems frustratingly ponderous)
posted by Just this guy, y'know at 4:00 AM on November 30, 2016 [1 favorite]

A pal has an iron rule - any Youtube video that starts 'Hello, Youtube' can be stopped right there. I wouldn't go quite that far, but the list of channels with putative interesting content rendered unwatchable by presenters who are either unwatchably annoying or anesthetically dull is quite large. (For reasons obvious to those versed in the art, ham radio videos very often qualify - hams have been selectively bred for a century to achieve very long monologues with minimal content, as they've been broadcasting to random listeners without restraint or feedback for longer than anyone.)

YouTube has a speed control, no plug-ins required. The 2x is usually too much, but 1.5 is a saviour for the hungry ear and eye grasping at thin gruel.
posted by Devonian at 8:00 AM on November 30, 2016 [3 favorites]

Things with slow, clunky animations but flawless timers annoy me to no end. Why are you including the time I have to sit and wait for some animation to complete?
posted by Avelwood at 11:54 AM on November 30, 2016 [1 favorite]