Joy Christian's colourful exploding balls experiment

[Joy Christian]

... do I have to look it up for you?

No, you don't have to look it up. The settings are fine. I have used similar settings in one of my papers: http://arxiv.org/pdf/1106.0748v6.pdf (cf. the top of page 8).

[gill1109]

The bet is on based on what I have written and proposed in my experimental paper(s)

The settings are fine. I have used similar settings in one of my papers: http://arxiv.org/pdf/1106.0748v6.pdf (cf. the top of page 8).

So you are not concerned by Michel's worries.

It makes life a whole lot easier: we just do N runs, extract N pairs of directions u_k and minus u_k from the videos (any way you like) and compute as many correlations as we like, all based on the same N observed directions u_k and minus u_k. There is no coin-tossing by me, no random sub-samples.

You can look at all of the correlations and draw pretty curves to show the whole world that we got the cosine; fine by me. Your bet with me is decided by four particular points on all those curves: which four points, agreed in advance. Four particular points where the cosine curves are equal to +/- 0.7 but where local realism predicts +/- 0.5. Four times 0.5 is 2 (Bell-CHSH). Four times 0.7 is 2.8 (Tsirelson). (I take sqrt 2 = 7/5. Not actually true but a good enough and conservative approximation. (7/5)^2 = 49/25 < 50/25 = 2).

I insist that the four points are formed in the standard way E(a_i, b_j) from two directions for Alice, two for Bob. You can pick Alice's two directions and Bob's two directions yourself, but I insist you do that in advance. We both want no trickery by the other.

Of course you can do 100 runs first and take a look at the correlations there, maybe adjust directions a bit, then do the definitive 10 000. As they do in real experiments. First a phase of tuning, adjustments, synchronisation, alignment ... then a final definitive experiment after everything has been got just right.

[Joy Christian]

I insist that the four points are formed in the standard way E(a_i, b_j) from two directions for Alice, two for Bob. You can pick Alice's two directions and Bob's two directions yourself, but I insist you do that in advance. We both want no trickery by the other.

I thought we already agreed about the four settings. We choose the standard "Bell test angles": 0, 45°, 22.5° and 67.5°, respectively.

[Joy Christian]

I insist that the four points are formed in the standard way E(a_i, b_j) from two directions for Alice, two for Bob. You can pick Alice's two directions and Bob's two directions yourself, but I insist you do that in advance. We both want no trickery by the other.

I thought we already agreed about the four settings. We choose the standard "Bell test angles": 0, 45°, 22.5° and 67.5°, respectively.

Sorry. My mistake. These angles are for the photon states. The correct angles for the squashy ball experiment would be: 0, 90°, 45° and 135°, respectively.

[gill1109]

I insist that the four points are formed in the standard way E(a_i, b_j) from two directions for Alice, two for Bob. You can pick Alice's two directions and Bob's two directions yourself, but I insist you do that in advance. We both want no trickery by the other.

I thought we already agreed about the four settings. We choose the standard "Bell test angles": 0, 45°, 22.5° and 67.5°, respectively.

Sorry. My mistake. These angles are for the photon states. The correct angles for the squashy ball experiment would be: 0, 90°, 45° and 135°, respectively.

No problem! I keep making that same mistake, myself. Yes: 0, 90; 45 and 135 are good choices for you.

abs(0 - 45) = abs(90 - 135) = abs(90 - 45) = 45;
abs(0 - 135) = 135 = 180 - 45;
and then we have - cos(45) = - 0.7, - cos(135) = + 0.7.

Three correlations large and negative, one large and positive.
4 * 0.7 = 2.8 >> 2 = 4 * 0.5.
We put the dividing line for the purpose of the bet at 2.4. If by some miracle the experimental result was exactly 2.4 the adjudicators may toss a coin, as far as I am concerned.

So we are following your experimental paper and using the same set of N runs (same set of N reconstructed pairs of directions) for all correlations?

If on the other hand we want to follow the standard protocol of a CHSH experiment, and devote each run to just one setting pair, then you have to tell me which setting corresponds to heads, which to tails. You declared in advance your angles. Please also assign them in advance to "heads" and " tails". You may choose which is which, but you must declare your choice in advance.

How about

alpha_H = 0
alpha_T = 90
beta_H = 45
beta_T = 135

Then the relevant CHSH criterion would be

E(alpha_H, beta_T) - E(alpha_H, beta_H) - E(alpha_T, beta_H) - E(alpha_T, beta_T)

which according to the singlet correlations would be 4 * 0.7 = 2.8 while according to Bell's theorem could at most only be 4 * 0.5 = 2
(up to statistical error of the order of 1 / sqrt N).

N = 10 000 ought to make our bets safe for both of us, each according to our own theory.

[Joy Christian]

So we are following your experimental paper and using the same set of N runs (same set of N reconstructed pairs of directions) for all correlations?

Yes.

If on the other hand we want to follow the standard protocol of a CHSH experiment, and devote each run to just one setting pair, then you have to tell me which setting corresponds to heads, which to tails. You declared in advance your angles. Please also assign them in advance to "heads" and " tails". You may choose which is which, but you must declare your choice in advance.

How about

alpha_H = 0
alpha_T = 90
beta_H = 45
beta_T = 135

Then the relevant CHSH criterion would be

E(alpha_H, beta_T) - E(alpha_H, beta_H) - E(alpha_T, beta_H) - E(alpha_T, beta_T)

which according to the singlet correlations would be 4 * 0.7 = 2.8 while according to Bell's theorem could at most only be 4 * 0.5 = 2
(up to statistical error of the order of 1 / sqrt N).

N = 10 000 ought to make our bets safe for both of us, each according to our own theory.

Nothing prevents us from following both procedures. The analysis of data comes for free once the actual spin directions are experimentally recorded for N runs. Both procedures are advisable so that no one may feel cheated and the community also takes something home from our efforts. After all, what is the point of doing the experiment if the community does not benefit from it (either by being reassured that I was wrong as they believed, or by having a spectacular revolution in physics)?

[gill1109]

Nothing prevents us from following both procedures. The analysis of data comes for free once the actual spin directions are experimentally recorded for N runs. Both procedures are advisable so that no one may feel cheated and the community also takes something home from our efforts. After all, what is the point of doing the experiment if the community does not benefit from it (either by being reassured that I was wrong as they believed, or by having a spectacular revolution in physics)?

Indeed: the experiment is the experiment and we can do everything we like with it. The more the better. The data should be publicly archived on internet so everyone can look at it themselves.

But a bet is a bet and it needs to have one single and very clear criterion. We want to make the life of our adjudicators as simple as possible. We're asking a lot from them!

[Joy Christian]

Nothing prevents us from following both procedures. The analysis of data comes for free once the actual spin directions are experimentally recorded for N runs. Both procedures are advisable so that no one may feel cheated and the community also takes something home from our efforts. After all, what is the point of doing the experiment if the community does not benefit from it (either by being reassured that I was wrong as they believed, or by having a spectacular revolution in physics)?

Indeed: the experiment is the experiment and we can do everything we like with it. The more the better. The data should be publicly archived on internet so everyone can look at it themselves.

But a bet is a bet and it needs to have one single and very clear criterion. We want to make the life of our adjudicators as simple as possible. We're asking a lot from them!

Fair enough. We follow the first procedure. The second procedure---i.e., of following the standard protocol of a CHSH experiment---is problematic in any case because I cannot tell you in advance which setting corresponds to heads, and which to tails. That is the thing about rotations. You can't tell heads or tails until you see them.

[gill1109]

Fair enough. We follow the first procedure. The second procedure---i.e., of following the standard protocol of a CHSH experiment---is problematic in any case because I cannot tell you in advance which setting corresponds to heads, and which to tails. That is the thing about rotations. You can't tell heads or tails until you see them.

I don't see your problem with the second procedure. But if it worries you, instead of checking one particular linear CHSH inequality we could check all 8 CHSH inequalities.

ie we could look at

max{
+ E(HH) - E(HT) - E(TH) - E(TT),
+ E(HT) - E(HH) - E(TH) - E(TT),
+ E(TH) - E(HT) - E(HH) - E(TT),
+ E(TT) - E(HT) - E(TH) - E(HH),
- E(HH) + E(HT) + E(TH) + E(TT),
- E(HT) + E(HH) + E(TH) + E(TT),
- E(TH) + E(HT) + E(HH) + E(TT),
- E(TT) + E(HT) + E(TH) + E(HH)
}

The best of all 8 linear CHSH inequalities. Then you don't have to decide anything in advance. You could be happy with *my* assignment

Alice H -> 0
Alice T -> 90
Bob H -> 45
Bob T -> 135

But if you are happy with the first procedure, so am I! It leaves *nothing* to chance. I would be happy with N = 100, say. Makes the experiment a whole lot cheaper, again.

[Joy Christian]

Fair enough. We follow the first procedure. The second procedure---i.e., of following the standard protocol of a CHSH experiment---is problematic in any case because I cannot tell you in advance which setting corresponds to heads, and which to tails. That is the thing about rotations. You can't tell heads or tails until you see them.

I don't see your problem with the second procedure. But if it worries you, instead of checking one particular linear CHSH inequality we could check all 8 CHSH inequalities.

ie we could look at

max{
+ E(HH) - E(HT) - E(TH) - E(TT),
+ E(HT) - E(HH) - E(TH) - E(TT),
+ E(TH) - E(HT) - E(HH) - E(TT),
+ E(TT) - E(HT) - E(TH) - E(HH),
- E(HH) + E(HT) + E(TH) + E(TT),
- E(HT) + E(HH) + E(TH) + E(TT),
- E(TH) + E(HT) + E(HH) + E(TT),
- E(TT) + E(HT) + E(TH) + E(HH)
}

The best of all 8 linear CHSH inequalities. Then you don't have to decide anything in advance. You could be happy with *my* assignment

Alice H -> 0
Alice T -> 90
Bob H -> 45
Bob T -> 135

But if you are happy with the first procedure, so am I! It leaves *nothing* to chance. I would be happy with N = 100, say. Makes the experiment a whole lot cheaper, again.

We leave nothing to chance. We follow the first procedure. All the other variants with the data will be tried later by many people if the strong correlations are seen.

[gill1109]

We leave nothing to chance. We follow the first procedure. All the other variants with the data will be tried later by many people if the strong correlations are seen.

OK. Michel is not going to like this.

[Joy Christian]

We leave nothing to chance. We follow the first procedure. All the other variants with the data will be tried later by many people if the strong correlations are seen.

OK. Michel is not going to like this.

I am hoping that Michel will register his worry here. To my part, I am not worried for two reasons. The first is the simple reason that Fred has already mentioned. The choice between the two procedures does not really matter if the correlations are strong---i.e., E(a, b) = -a.b. The CHSH inequality would be violated in that case.

The second reason is rather involved and it has to do with the topology of S^3. I have explained in detail why and how CHSH is violated by Nature and in my model.
The violation has nothing to do with entanglement, non-locality, or non-realism. It has to do with the topology of the physical space we happen to live in. Nothing will prevent us from seeing that if we follow the procedure described in my paper and calculate each correlation separately---either as a full curve or as just four points.

[gill1109]

We leave nothing to chance. We follow the first procedure. All the other variants with the data will be tried later by many people if the strong correlations are seen.

OK. Michel is not going to like this.

I am hoping that Michel will register his worry here. To my part, I am not worried for two reasons. The first is the simple reason that Fred has already mentioned. The choice between the two procedures does not really matter if the correlations are strong---i.e., E(a, b) = -a.b. The CHSH inequality would be violated in that case.

The second reason is rather involved and it has to do with the topology of S^3. I have explained in detail why and how CHSH is violated by Nature and in my model.
The violation has nothing to do with entanglement, non-locality, or non-realism. It has to do with the topology of the physical space we happen to live in. Nothing will prevent us from seeing that if we follow the procedure described in my paper and calculate each correlation separately---either as a full curve or as just four points.

Well, your two reasons are not independent of one another. The second reason, is the reason why you are confident that E(a, b) = -a.b.

But I can say that I agree with your reasoning, entirely. If your theory is correct, then it would be possible that E(a, b) = -a.b. And if E(a, b) = -a.b then you have absolutely nothing to fear, whichever procedure we use, as long as N is decently large so that experimental correlations are close to the theoretical correlations. A random sample of size N/4 taken from a random sample, is also a random sample. And if N is large then N/4 is also large.

Given your belief in your work, you have absolutely nothing to fear.

Given my belief in my work, I have absolutely nothing to fear.

Michel already registered his worry about the "all correlations based on the same, full, sample" here. He seems not to have any worries about the other option. But maybe he will change his mind soon. I hope so, for him, because I have very high regard for his simulation programs and I think he could benefit from some statistical insights into what they can achieve and what they cannot achieve. The whole loophole discussion.

In that case, however, I would be a bit unhappy if he would then communicate his worries to you, because then you would pull out of the bet and withdraw your experimental papers, as fast as possible.

That was an example of a chain of counterfactual reasoning. It is probably an example where counterfactual reasoning is pretty pointless.

[Joy Christian]

Given your belief in your work, you have absolutely nothing to fear.

Given my belief in my work, I have absolutely nothing to fear.

So the bet is on. We may believe what we like to, but let Nature be the final judge. Whether my claim is vindicated or refuted, she will surely be the winner.

I can't help recalling, however, that Clauser was absolutely convinced before the very first experiment by him and Freedman that CHSH would hold and QM would be falsified. He was in for a major surprise (Shimony, on the other hand, was convinced that QM would hold and CHSH would be violated).

With that in mind, would you like to draw up a document? I can then review it and see if I agree with everything. We can then pass it on to our adjudicators to see what they think. We may then also consider to post the document here, either on this thread or on a new thread (these are all just suggestions).

[gill1109]

Given your belief in your work, you have absolutely nothing to fear.

Given my belief in my work, I have absolutely nothing to fear.

So the bet is on. We may believe what we like to, but let Nature be the final judge. Whether my claim is vindicated or refuted, she will surely be the winner.

I can't help recalling, however, that Clauser was absolutely convinced before the very first experiment by him and Freedman that CHSH would hold and QM would be falsified. He was in for a major surprise (Shimony, on the other hand, was convinced that QM would hold and CHSH would be violated).

With that in mind, would you like to draw up a document? I can then review it and see if I agree with everything. We can then pass it on to our adjudicators to see what they think. We may then also consider to post the document here, either on this thread or on a new thread (these are all just suggestions).

I keep an open mind and am willing to be proved wrong.

Yes I will make a first draft of a protocol for you.

[FrediFizzx]

I would suggest in the protocol, that you make the 5K Euro bet as a donation to the winner's favorite charity in their name. You will probably get more interest in the crowdfunding phase.

[Joy Christian]

I would suggest in the protocol, that you make the 5K Euro bet as a donation to the winner's favorite charity in their name. You will probably get more interest in the crowdfunding phase.

Sounds good to me.

[FrediFizzx]

And also that if there is any money left over from the crowdfunding after the experiment is done, that it will go to the winner's charity.

[minkwe]

We leave nothing to chance. We follow the first procedure. All the other variants with the data will be tried later by many people if the strong correlations are seen.

OK. Michel is not going to like this.

Richard, I will prefer if you do not speak for me. I can do that myself. I have already expressed my worries very clearly and I am doing that in the related thread about the CHSH. Let me do it again:

My only worry is that we are comparing apples to apples and oranges to oranges.
Apple: All correlations are calculated from the same set. (ie, the terms in the CHSH inequality)
Orange: Each correlation is calculated from a disjoint set from the others (ie, terms from QM and terms from Aspect-type experiments)

Richard believes Apples are really Oranges due to some kind of statistical tricks. If he has confidence in his work, he shouldn't mind if you guys bet to do an Orange experiment. But he appears to mind because he keeps insisting on calculating everything on the same set. So my advice to both of you has been the following: Pick one and stick with it. Keeping in mind the following: If you choose to do an Apple experiment, you will both be deluding yourselves if you compare it to QM or Aspect-type experiments which are Oranges. And if you do an Orange experiment, you will both be deluding yourselves (like most physicists/mathematicians these days anyway) if you compare it with the CHSH inequality, for reasons I have clearly explained elsewhere.

This is my only concern. Again let me reiterate: You can either bet to do an Apple experiment about the CHSH inequality or you can bet to do an Orange experiment about QM. QM says absolutely nothing about Apple type experiments. The QM predictions are for Orange type experiments. This is the reason Bell's theorem is flawed as I've explained many times. Bell took Orange type predictions and compared them with an Apple type inequality, himself falling into the grave he had dug for von Neumann's no-go theorem.

[minkwe]

Here is what Mermin had to say about it:

John Bell, 1966 rediscovered the fact that von Neumann's no-hidden variables proof was based on an assumption that can only be described as silly -- so silly, in fact, that one is led to wonder whether the proof was ever studied by either the students or those who appealed to it to rescure them from speculative adventures

In case you think Mermin was being discourteous, here is what Bell himself said about it:

Yet the von Neumann proof, if you actually come to grips with it, falls apart in your hands! There is nothing to it. It is not just flawed, it's silly! ... When you translate [his assumptions] into terms of physical disposition, they are nonsense. You may quote me on that: The proof of von Neumann is not merely false but foolish!"

http://lin25.thphys.uni-heidelberg.de/~ ... mermin.pdf

Now if those quotes got you interested in what von Neumann's silly assumption was, it turns out, it is exactly equivalent to the assumption that Oranges are Apples. The suggestion that you can use QM and Aspect-type experiments as sources of correlation terms for purposes of comparing with the CHSH inequality.