Why does EPR need two observables?
Vonny N.
Bohr's 'measurement=meaning' interpretation of the quantum formalism,
one could show that a measurement in one part of space must
instantaneously influence the results of a measurement performed in
another part of space. As is well known, this argument makes use of a
pair of non-commuting observables for each of two entangled systems.
But it seems to me that only *one* observable is required to establish
Einstein's thesis. If, for example, one takes a pair of entangled
particles with total momentum equal to zero, then the results of the
measurement of one immediately imply the result of a subsequent
measurement performed on the other. Who needs a second observable
(position, say) for this argument to have force?
Obviously, two observables are necessary to get Bell's inequalities,
and also to discuss Heisenberg's Uncertainty Principle, but why are
they needed to establish Einstein's conclusion.
Vonny N.
a student
Vonny N. wrote:
> Einstein's main point in his EPR thought-experiment was that under
> Bohr's 'measurement=meaning' interpretation of the quantum formalism,
> one could show that a measurement in one part of space must
> instantaneously influence the results of a measurement performed in
> another part of space. As is well known, this argument makes use of a
> pair of non-commuting observables for each of two entangled systems.
> But it seems to me that only *one* observable is required to establish
> Einstein's thesis. If, for example, one takes a pair of entangled
> particles with total momentum equal to zero, then the results of the
> measurement of one immediately imply the result of a subsequent
> measurement performed on the other. Who needs a second observable
> (position, say) for this argument to have force?
The main point of the EPR paper was not to show that a measurement
instantaneously influences results elsewhere (indeed, the converse is
assumed in the paper). Further, even if a measurement on one particle
may imply the result of a measurement on a distant particle, there need
be no instantaneous influence involved - if two billiard balls collide,
I can work out the momentum of one from the momentum of the other
without anything non-local going on.
The EPR paper assumed that (i) NO instantaneous influence of
measurements, and (ii) a perfect pre-existing correlation between
distant particles implies that the correlated observables have
pre-existing values (otherwise how would one particle "know" how to
output the correct correlated value, given assumption (i)?). Here
"pre-existing" denotes "after preparation and before measurement".
But quantum mechanics allows perfect correlations between pairs of
non-commuting observables (position and momentum in the EPR paper, Bohm
later gave the spin example). The EPR paper therefore concludes that
(a) in such cases these observables must have pre-existing values, and
hence that (b) since quantum mechanics does not (and cannot) prescribe
such values, it must be incomplete.
Thus, noncommuting observables are needed, as the validity of (b)
depends on the non-existence of simultaneous eigenstates of such
observables.
Many people (particularly after the glorious sharpening by Bell, and
subsequent experiments by Aspect), regard the "locality" assumption (i)
above as suspect. However, given that no measurable non-local
influence is detectable (according to quantum mechanics), one can as
well suspect the "reality" assumption (ii). Some people also note
that the root of the trouble might be the implicit use of a third
"non-conspiracy" assumption, that experimenters have free choice in
what they are going to measure.
I.Vecchi
..
> Many people (particularly after the glorious sharpening by Bell, and
> subsequent experiments by Aspect), regard the "locality" assumption (i)
> above as suspect. However, given that no measurable non-local
> influence is detectable (according to quantum mechanics), one can as
> well suspect the "reality" assumption (ii). Some people also note
> that the root of the trouble might be the implicit use of a third
> "non-conspiracy" assumption, that experimenters have free choice in
> what they are going to measure.
I am not aware of any physically meaningful definition of free will (or
free choice), yet the concept keeps popping up in quantum mechanical
discussions. Here is my try.
In a quantum relational perspective free will and randomness are
undistinguishable and hence physically identical. They both boil down
to unpredictability. Bob can't tell when the Geiger will click and he
can't tell when his gf Alice will dump him (or more prosaically,
whether she will choose 0 or 1). Both the Geiger and Alice have free
will.
Cheers,
IV
Ilja Schmelzer
"a student" <of_1001...@hotmail.com> schrieb
> Many people (particularly after the glorious sharpening by Bell, and
> subsequent experiments by Aspect), regard the "locality" assumption (i)
> above as suspect. However, given that no measurable non-local
> influence is detectable (according to quantum mechanics), one can as
> well suspect the "reality" assumption (ii).
In the form used by Bell the "reality assumption" is very general, very
weak, does not even refer to spacetime. It may be argued that giving
it up is equivalent to give up parts of the scientific method (the search
for realistic explanations of observable effects.)
Einstein causality (i) is, instead, a quite special assumption about
spacetime. If we give up realism, the realistic version of Einstein
causality becomes meaningless anyway. The weak (probabilitistic)
version survives anyway. Thus, if we compare the two choices,
we only loose if we give up realism.
> Some people also note
> that the root of the trouble might be the implicit use of a third
> "non-conspiracy" assumption, that experimenters have free choice in
> what they are going to measure.
If we follow this line, Einstein causality becomes meaningless.
Even a working FTL phone would no longer falsify Einstein
causality, because it could be explained with conspiracy as well.
Ilja
Vonny N.
>
> Thus, noncommuting observables are needed, as the validity of (b)
> depends on the non-existence of simultaneous eigenstates of such
> observables.
I had a feeling that my question might result in a standard lecture on
EPR ;-)
Actually, I mustn't have made my point clear enough, so please let me
try again.
To convince someone today of the non-locality 'dilemma' QM confronts us
with, one typically takes an entangled quantum system, two
non-commuting observables, a proof of a Bell-type inequality, followed
by the results of some Aspect-type experiment, ... blah blah blah ...
This is the standard sequence of topics emerging from discussions about
EPR, and appears to be the one you have addressed. Unfortunately, this
is not what I am interested in here.
My present concern, which is somewhat more historical, lies in what
Einstein was trying to achieve through this thought experiment *at the
time*. This is a different task altogether.
Now Einstein, as we all know, believed that the world was most
certainly local. What is less frequently noted is that it is clear from
his arguments that he also assumed that *Bohr* would be against
non-locality. Thus he wanted to refute Bohr's interpretation by
demonstrating that it entailed non-locality. And I claim (or, better
perhaps, ask whether) this can be done with a single observable.
To do this, suppose you have an entangled state in which the total
momentum, say, is zero, but such that neither of the individual systems
is in a momentum eigenstate. Then a momentum measurement on one system
tells us immediately the result of a subsequent momentum measurement on
the other. As you say, there is nothing unusual about this, billiard
balls do it all the time. However, Bohr claimed that the billiard ball
model is *a priori* incorrect and that a value of momentum is only
meaningful upon measurement. But if Bohr wants to maintain his
'measurement = meaning' interpretation, he appears now to be forced to
concede some type of action-at-a-distance, with or without
consideration of other observables. And that, I think, is all Einstein
was trying to demonstrate.
Now, carrying on from this single-observable experiment we can
introduce a second, non-commuting, observable to show how the locality
assumption would lead to Heisenberg's Uncertainty Principle being
violated; but surely this is redundant since we have already shown that
Bohr's interpretation entails non-locality. We could also take
advantage of entanglement (which we haven't yet!) by establishing
Bell-type inequalities. But again, that is not my or Einstein's
intention at all.
In fact Einstein's aim of showing that Bohr's interpretation was
non-local doesn't even need EPR. Einstein had already, back in 1927,
argued that the appearance of a mark on a screen due the detection of
an electron already requires a kind of non-locality under Bohr's way of
thinking. I believe Einstein was using EPR as a more transparent and
convincing demonstration of the non-locality entailed in Bohr's view;
and for this, we appear to only need a single observable.
It seems to me that authors discussing EPR tend to confuse the initial
objective of the experiment, with the program it accidentally triggered
thanks to the later appearance of John Bell. 'EPR Logic' is confusing
enough as it is, without throwing these separate topics into the
blender.
Vonny N.
a student
> In the form used by Bell the "reality assumption" is very general, very
> weak, does not even refer to spacetime. It may be argued that giving
> it up is equivalent to give up parts of the scientific method (the search
> for realistic explanations of observable effects.)
>
> Einstein causality (i) is, instead, a quite special assumption about
> spacetime. If we give up realism, the realistic version of Einstein
> causality becomes meaningless anyway. The weak (probabilitistic)
> version survives anyway. Thus, if we compare the two choices,
> we only loose if we give up realism.
I don't know that one can be said to be definitively weaker than the
other - it possibly comes down to personal preferences. I find it hard
to give up either (i) or (ii) !
However, arguments for not giving up the locality assumption (i)
include
(a) "If we don't see it, why postulate it?": Locality is strictly
predicted/observed by quantum mechanics (operators with
spacelike-separated support commute). Hence, non-locality can only
consistently be brought in at some "deeper" level providing that this
level remains wholly unobservable for all known quantum phenomena.
This smacks of (unnecessary) conspiracy to me! Of course, it would be
a different matter if the deeper level did give rise to physical
predictions of verifiable nonlocality (while avoiding the shooting of
one's own grandparents).
(b) "Giving up locality is equivalent to giving up parts of the
scientific method": The main reason that physicists get away with
explaining so many things about the world around us is because we can
ignore so much of the world around us (we are also very good at picking
things that we can explain!). In particular, most interactions between
systems (and indeed most systems) can be ignored in modelling a
particular phenomenon - particularly if they are a long long way away
(eg, in the lab next door). If we suddenly say that there are in fact
nonlocal interactions going on between everything, we put this basic
principle at risk - how can I ever hope to understand the energy levels
of a single hydrogen atom if the electron is interacting with my Aunt
Jemima's hairdryer? Where is the noise?
> > Some people also note
> > that the root of the trouble might be the implicit use of a third
> > "non-conspiracy" assumption, that experimenters have free choice in
> > what they are going to measure.
>
> If we follow this line, Einstein causality becomes meaningless.
> Even a working FTL phone would no longer falsify Einstein
> causality, because it could be explained with conspiracy as well.
I agree that conspiracy is rather too strong to please anyone. It is
more of an unpalatable logical possibility (bit like religion
explanations really!). There was a very amusing paper by Shimony and
others in Dialectica a long while ago, with this scenario: Alain and
Philipppe order their "Bell experiment" apparatus and set it up. To
randomise the measurement settings for each correlated pair of spins,
they get their secretaries to say "heads" or "tails", and set the
detectors appropriately. They verify the Bell inequalities are
violated, and publish in the latest fashionable journal. However,
unknown to them, the manufacturer had provided two sheets, each with a
sequence of "heads" and "tails" written on them, and the secretaries
had merely read out what was on the sheets in order. It was all
predetermined!
Ilja Schmelzer
> Ilja Schmelzer wrote:
> > In the form used by Bell the "reality assumption" is very general, very
> > weak, does not even refer to spacetime. It may be argued that giving
> > it up is equivalent to give up parts of the scientific method (the
search
> > for realistic explanations of observable effects.)
> >
> > Einstein causality (i) is, instead, a quite special assumption about
> > spacetime. If we give up realism, the realistic version of Einstein
> > causality becomes meaningless anyway. The weak (probabilitistic)
> > version survives anyway. Thus, if we compare the two choices,
> > we only loose if we give up realism.
>
> I don't know that one can be said to be definitively weaker than the
> other - it possibly comes down to personal preferences. I find it hard
> to give up either (i) or (ii) !
>
> However, arguments for not giving up the locality assumption (i)
> include
> (a) "If we don't see it, why postulate it?":
That's easy for realists. We postulate lots of things we don't see,
like particles in the center of the moon or colors of quarks.
The reason for postulating things we don't see is that all simple
explanations (usually some more restricted subseto of explanations)
of what we see include the non-seen thing in one or another way.
> Locality is strictly
> predicted/observed by quantum mechanics (operators with
> spacelike-separated support commute). Hence, non-locality can only
> consistently be brought in at some "deeper" level providing that this
> level remains wholly unobservable for all known quantum phenomena.
> This smacks of (unnecessary) conspiracy to me!
There is no need for unnecessary conspiracy.
Assume you have an effect which allows only two explanations,
A->B or B->A. Above explanations violate Einstein causality.
Would you really believe into Einstein causality? But, of course,
in this situation it follows that we cannot use this for information
transfer A->B (that would violate the explanation B->A). Thus,
the weak QM notion of "Einstein causality" also holds.
> Of course, it would be
> a different matter if the deeper level did give rise to physical
> predictions of verifiable nonlocality (while avoiding the shooting of
> one's own grandparents).
You grandparents are out of danger if you simply accept a hidden
preferred frame.
> (b) "Giving up locality is equivalent to giving up parts of the
> scientific method":
No. That would be a normal application of the scientific method.
We have made an indirect observation A->B or B->A, each
explanation violates Einstein causality, thus, it is falsified by
indirect observation. We have a reasonable simple replacement -
classical causality in a preferred frame.
> The main reason that physicists get away with
> explaining so many things about the world around us is because we can
> ignore so much of the world around us (we are also very good at picking
> things that we can explain!). In particular, most interactions between
> systems (and indeed most systems) can be ignored in modelling a
> particular phenomenon - particularly if they are a long long way away
> (eg, in the lab next door). If we suddenly say that there are in fact
> nonlocal interactions going on between everything, we put this basic
> principle at risk - how can I ever hope to understand the energy levels
> of a single hydrogen atom if the electron is interacting with my Aunt
> Jemima's hairdryer?
Don't be afraid. Science has survived a sufficiently large time with
nonlocal theories (Newtonian gravity). But has there been science
without realism? Note, the notion of realism we talk about here
is very weak, even explanations including ghosts or gods would
be realistic in this very weak sense.
And it is completely reasonable to hope that locality with some
higher limiting speed will be reestablished later. We even an approximate
candidate for the preferred frame in nature and a criterion - the violation
of Bell's inequality should stop if the events are space-like separated from
point of view of the new limiting speed.
Last not least, your Aunts Jemima's hairdryer is allowed to interact
with your hydrogen atom even with Einstein causality. Thus, allowing
such interactions does not endanger the theory of stable hydrogen
atoms.
> I agree that conspiracy is rather too strong to please anyone. It is
> more of an unpalatable logical possibility (bit like religion
> explanations really!). There was a very amusing paper by Shimony and
> others in Dialectica a long while ago, with this scenario: Alain and
> Philipppe order their "Bell experiment" apparatus and set it up. To
> randomise the measurement settings for each correlated pair of spins,
> they get their secretaries to say "heads" or "tails", and set the
> detectors appropriately. They verify the Bell inequalities are
> violated, and publish in the latest fashionable journal. However,
> unknown to them, the manufacturer had provided two sheets, each with a
> sequence of "heads" and "tails" written on them, and the secretaries
> had merely read out what was on the sheets in order. It was all
> predetermined!
Yep. But the point is, if you have a phone so that you can really transfer
information in the usual sense FTL, the same type of conspiracy
explanation would also allow to explain it away. Thus, if you take this
conspiracy explanation serious, you remember the fool who denies the
existence of, say, faster than sound communication even if confronted
with a usual phone.
In other words, the conspiracy explanation does not allow you to
preserve Einstein causality. Instead, Einstein causality becomes
unfalsifiable and therefore physically meaningless.
Ilja
a student
> Now Einstein, as we all know, believed that the world was most
> certainly local. What is less frequently noted is that it is clear from
> his arguments that he also assumed that *Bohr* would be against
> non-locality. Thus he wanted to refute Bohr's interpretation by
> demonstrating that it entailed non-locality. And I claim (or, better
> perhaps, ask whether) this can be done with a single observable.
I think this is reading rather a lot into the EPR paper. Einstein
later commented that he didn't like Podolski's write-up of the paper.
But he did NOT say that he felt the point was to prove that Bohr's
interpretation was nonlocal. He gave his reading of the EPR scenario
as showing that quantum mechanics gave too many descriptions of the
unmeasured particle -depending on which (complementary) observable was
measured on the first particle (whereas Podolski emphasised
incompleteness). So it might be more reasonable (less shoehorning) to
credit the nonlocality reading as your own, rather than as Einstein's
:) - this doesn't detract from it being of interest.
> To do this, suppose you have an entangled state in which the total
> momentum, say, is zero, but such that neither of the individual systems
> is in a momentum eigenstate. Then a momentum measurement on one system
> tells us immediately the result of a subsequent momentum measurement on
> the other. As you say, there is nothing unusual about this, billiard
> balls do it all the time. However, Bohr claimed that the billiard ball
> model is *a priori* incorrect and that a value of momentum is only
> meaningful upon measurement. But if Bohr wants to maintain his
> 'measurement = meaning' interpretation, he appears now to be forced to
> concede some type of action-at-a-distance, with or without
> consideration of other observables. And that, I think, is all Einstein
> was trying to demonstrate.
I probably still don't fully understand your point of view. But it
seems to me that the hypothetical Bohr has no problem at this stage of
the argument - the real Bohr emphasised repeatedly that there is no
phenomenon before it is observed. Hence, for him, the momentum of the
second particle has not been measured/created/made meaningful at this
stage. It has simply been made predictable (should someone actually
measure the momentum of the second particle).
Bohr similarly held that, simply because a system is in an eigenstate
|a> of some observable A (and hence that A is predictable), it does not
mean that A actually "has" a value equal to a. The conditions for
measuring A must first be set up and the measurement carried out. Only
then is A measured, and a measured value created/defined/made
meaningful. Note that the conditions for measuring A will be
physically incompatible with the conditions for measuring some
complementary observable B, and hence only one can be
measured/defined/made meaningful. In particular, if B was measured
instead of of A, there is no sudden "disturbance" of a real value A=a -
there is simply a different phenomenon and a different measurement
result created.
Thus, I don't see Bohr actually conceding any nonlocality, at least on
the basis of your argument as I understand it.
Ilja Schmelzer
"Vonny N." <von...@hotmail.com> schrieb
> My present concern, which is somewhat more historical, lies in what
> Einstein was trying to achieve through this thought experiment *at the
> time*. This is a different task altogether.
His aim was to show that QM is incomplete.
> Now Einstein, as we all know, believed that the world was most
> certainly local. What is less frequently noted is that it is clear from
> his arguments that he also assumed that *Bohr* would be against
> non-locality. Thus he wanted to refute Bohr's interpretation by
> demonstrating that it entailed non-locality.
AFAIU, his aim was not to question the interpretation named
today minimal, that means, that |psi|^2 defines a probability
distribution. The aim was to show that this is not the end
of science, not the most fundamental theory.
> And I claim (or, better
> perhaps, ask whether) this can be done with a single observable.
The point of having two observables was that without measuring
observable 1 I can be sure that it gives the result defined by the
other measurement far away.
> But if Bohr wants to maintain his
> 'measurement = meaning' interpretation, he appears now to be forced to
> concede some type of action-at-a-distance, with or without
> consideration of other observables. And that, I think, is all Einstein
> was trying to demonstrate.
I think Einstein's interest was to find something beyond the uncertainty
relation.
> Now, carrying on from this single-observable experiment we can
> introduce a second, non-commuting, observable to show how the locality
> assumption would lead to Heisenberg's Uncertainty Principle being
> violated; but surely this is redundant since we have already shown that
> Bohr's interpretation entails non-locality. We could also take
> advantage of entanglement (which we haven't yet!) by establishing
> Bell-type inequalities. But again, that is not my or Einstein's
> intention at all.
I think Einstein would have liked Bell's inequalities very much.
(But not the result of Aspect's experiment.)
> It seems to me that authors discussing EPR tend to confuse the initial
> objective of the experiment, with the program it accidentally triggered
> thanks to the later appearance of John Bell. 'EPR Logic' is confusing
> enough as it is, without throwing these separate topics into the
> blender.
But with Bell's proof and the observable violation of his inequality we
have obtained a very deep insight into the nature of QM which is much
more interesting than the insight you can obtain considering only a
single observable.
Ilja
I.Vecchi
..
> (b) "Giving up locality is equivalent to giving up parts of the
> scientific method": The main reason that physicists get away with
> explaining so many things about the world around us is because we can
> ignore so much of the world around us (we are also very good at picking
> things that we can explain!). In particular, most interactions between
> systems (and indeed most systems) can be ignored in modelling a
> particular phenomenon - particularly if they are a long long way away
> (eg, in the lab next door). If we suddenly say that there are in fact
> nonlocal interactions going on between everything, we put this basic
> principle at risk - how can I ever hope to understand the energy levels
> of a single hydrogen atom if the electron is interacting with my Aunt
> Jemima's hairdryer? Where is the noise?
The argument is not new. Actually Newton had to deal with it. When the
"Principia" came out, mainstream localists-mechanicists yelled (not
without reason) that Newton's action-at-distance smelled of alchemy. He
famously replied "Hypotheses non fingo".
The core issue is still whether the "scientific method" is about
predicting experimental outcomes or fitting explanatory expectations.
Cheers,
IV
Andreas Most
> a student ha scritto:
> ..
>> Many people (particularly after the glorious sharpening by Bell, and
>> subsequent experiments by Aspect), regard the "locality" assumption (i)
>> above as suspect. However, given that no measurable non-local
>> influence is detectable (according to quantum mechanics), one can as
>> well suspect the "reality" assumption (ii). Some people also note
>> that the root of the trouble might be the implicit use of a third
>> "non-conspiracy" assumption, that experimenters have free choice in
>> what they are going to measure.
>
> I am not aware of any physically meaningful definition of free will (or
> free choice), yet the concept keeps popping up in quantum mechanical
> discussions. Here is my try.
Free will is certainly a philosophical term but also applies to physics.
As of my understanding physics assumes tacitly that an experimenter is
free to choose his measurement setup and what and if he is going to
measure. If you give up this assumption (we may even call it a
postulate) it could be argued that the measurement setup is
predetermined and the "free will" of the experimenter is only an
illusion. This could solve a lot of these philosophical puzzles
people have with quantum mechanics. But on the other hand it would
only shift the problem, because you might wonder now what (or who)
predetermines what we do...
Anyhow, being too arrogant I prefer to assume that we have a free will ;-)
Andreas.
I.Vecchi
..
> Now, carrying on from this single-observable experiment we can
> introduce a second, non-commuting, observable to show how the locality
> assumption would lead to Heisenberg's Uncertainty Principle being
> violated; but surely this is redundant since we have already shown that
> Bohr's interpretation entails non-locality. We could also take
> advantage of entanglement (which we haven't yet!) by establishing
> Bell-type inequalities. But again, that is not my or Einstein's
> intention at all.
>
> In fact Einstein's aim of showing that Bohr's interpretation was
> non-local doesn't even need EPR. Einstein had already, back in 1927,
> argued that the appearance of a mark on a screen due the detection of
> an electron already requires a kind of non-locality under Bohr's way of
> thinking. I believe Einstein was using EPR as a more transparent and
> convincing demonstration of the non-locality entailed in Bohr's view;
> and for this, we appear to only need a single observable.
..
Yours is a remarkable argument. For reasons that should be clear to
anyone who bothers to read my recent posts (e.g. [1]), I stand by the
two-observables interpretation (we are talking about two measurements
here) , but your point highlights the basic issue.
IV
[1]
http://groups.google.com/group/sci.physics.research/msg/ae1f348208b8206c
nightlight
> (But not the result of Aspect's experiment.)
He would have loved the result, too. Not just the Aspect's, but
the rest of them since. They all confirm what he believed.
He probably wouldn't have cared much about the usual wishful
spin on the _unbroken chain_ of experimental failures to violate
Bell Inequalities, which makes the gross failure appear as
'almost success', at least to those too impatient to read the
fine print.
E. Santos
"Bell's theorem and the experiments: Increasing empirical
support to local realism"
http://arxiv.org/abs/quant-ph/0410193
E. Santos
"Optical tests of Bell's inequalities not resting upon the
absurd fair sampling assumption"
http://arxiv.org/abs/quant-ph/0401003
sci.physics "Is There Bell's Theorem"
http://groups.google.com/group/sci.physics/msg/6a055035f346f309
sci.physics.research
On absence of QED prediction of B.I. violations
http://groups.google.com/group/sci.physics.research/msg/5d5dec33366d0bd9
http://groups.google.com/group/sci.physics.research/msg/c3e4a550fef154bd
http://groups.google.com/group/sci.physics.research/msg/fbd9858ee710e27a
http://groups.google.com/group/sci.physics.research/msg/7ee770990a31bd3f
PhysicsForum
"Photon 'Wave Collapse' Experiment (Yeah sure; AJP Sep 2004, Thorn...)"
http://www.physicsforums.com/showthread.php?t=71297
a student
> > I think Einstein would have liked Bell's inequalities very much.
> > (But not the result of Aspect's experiment.)
>
> He would have loved the result, too. Not just the Aspect's, but
> the rest of them since. They all confirm what he believed.
Even if not loophole-free, they certainly don't "confirm" local
realism!
> He probably wouldn't have cared much about the usual wishful
> spin on the _unbroken chain_ of experimental failures to violate
> Bell Inequalities, which makes the gross failure appear as
> 'almost success', at least to those too impatient to read the
> fine print.
Nearly all experiments have non-idealities. It is amusing, I suppose,
that the "no-enhancement" loophole (eg, that the detectability of a
photon does not increase on passing through a polariser), can be got
around only with atoms at present; while the "light-cone" loophole (eg,
that entangled atoms have not yet been measured in spacelike separated
regions), can be got around only with photons at present. However,
more decisive experiments are continually under construction, and there
is nothing to suggest, to date, that a decisive test is not possible.
It should be remembered, BTW, that such research does not attract the
same sort of interest and funding as, eg, quantum computing and
superstrings (pity!).
> E. Santos
> "Bell's theorem and the experiments: Increasing empirical
> support to local realism"
> http://arxiv.org/abs/quant-ph/0410193
This is an excellent paper for reviewing the loopholes of experiments
to date, and for making a cogent argument for the position that
loophole-free tests of the Bell inequalities may be physically
forbidden (but offers no actual model of such a model to replace
present quantum theory).
While it is important to keep an open mind, I suspect that some people
will in fact never be persuaded of Bell inequality violation by any
future experiment - they will always plead some new "conspiracy" or
"intelligent design" (or "flying spaghetti monster") loophole that
makes local realism OK. But that would not be science. Occam's razor
has to cut in at some point.