Slashdot Mirror

http://users.aber.ac.uk/ajs99/Altairhtml/Altair.sh tml

and photos:

http://users.aber.ac.uk/ajs99/Altairhtml/presspics .shtml

I'm not connected with them, I just work down the road.

here's a lan case / pc I made :) The PortaPC

Oops, bad copy-paste! That's the link I intended; sorry to all others for the misunderstanding. While we're at it, there's also this one; while Thompson certainly doesn't seem to have "what it takes" (as Trevor Marshall does), her site is also quite interesting.

The ion based tests open their own set of loopholes, while closing the common photon tests loopholes.

Unlike photons, ions can be detected with high efficiency (theoretically 100% since they're massive particles). But since the spin coupling energy (with Stern-Gerlach magnets, the analogue of polarizers in optical experiments) is much lower than their kinteic energy, the spin measurements (selection) are much less reliable than those of optical photon polarizers. This results in large background counts (due to depolarization) and this subtraction is a well known loophole for LHV models (also occuring with photon tests when the sensitivity of the detectors is increased).

Additional problem is in reliable production of the entangled pair state, again due to the low energy of spin-spin coupling (between the spins of the ion pair) relative to other energies involved in the process of pair production and collimation. The result is again a large number of "accidental coincidence" detections, i.e. another contribution to the background to be subtracted.

Hence the background subtractions make ionic tests very similar to the case of photon tests when the photo-detection is increased to near 100%. This can be achieved by using very high energy photons (e.g. gamma photons from the electron-positron annihilation), in which case one can have near perfect detection, but polarization measurement doesn't work too well (via Compton scattering), producing very large background which needs to be subtracted (exactly as with ions, and for essentially the same reason).

An alternative way to increase the photo-detection efficiency is to use very sensitive (low threshold) detector, but that produces large dark current, the background noise, which again has to be subtracted. The resulting unsubtracted data in such case are almost exactly what a classically entangled EM wave packet would produce (i.e. what Maxwell equations would predict). You can, for example, see the actual raw data from Asepect's PhD thesis, where he did his famous cascade experiments, on Caroline Thompson's web site. For more modern PDC based experiments, see the similar Stochastic Electrodynamics models (which are again the Maxwell equations based models, but with stochastic initial & boundary conditions) which reproduce the raw data for these experiments, on the Trevor Marshall's site.

It is true, as far as I know, that no single experiment has gotten sufficient sensitivity using spacelike separated measurements... yet. It seems somehow perverse though to hang a defense of local realism on this fact.

This is the oldest handwaving argument for dismissing the loopholes in Bell tests, i.e. why would some future increased sensitivity (for detection or polarizer efficiency) suddenly switch from the good agreement with QM (modulo loopholes) and give preference to the local realism.

First one should note that the non-adjusted experimental data is already consistent with the local realism (and there are numerous local models for variety of the setups reproducingt he experimental non-adjusted data). So nothing here has to change for more sensitive experiments. Only the wishfully adjusted data (when the loopholes are dismissed via ad hoc unverifiable assumptions) exclude local realism.

Second, it is perfectly natural for any local realistic model to make the efficiency of the detector or polarizer dependent on the values of the local (hidden) variables. Such dependency is outright excluded by the fair sampling (be it sampling by the detectors or the polarizers; in the ion case the critical fair sampling problem is at the Stern Gerlach magnets, the "polarizer," not the detector, which is here near-perfectly efficient; while in photon case it is at the photo-detectors, and not at the near-perfect polarizers). The ion tests only shift the fair sampling (in the hidden variable space) problem closer to the source, at the Stern-Gerlach and the pair source production depolarizations (which results in the background counts, which is presumed to be a fair sample in the LHV space, thus it is flatly subtracted subtracted from the total counts).

To illustrate this point in a form more accessible to non-physicists here, consider a national poll using email. Such poll will fail to detect people who do not have computers with internet connection and email account. Suppose now the poll analysis expert wasn't informed about the method of communication used to obtain the poll data. This is analogous to the Bell test experimenter looking at the counts of particles detected, but not being able to measure or know anything about the so-called local hidden variables (analog to the email, which is here invisible variable/value to the poll analyst).

The analyst, not knowing anything about the underlying means of communication, proclaims now that he will assume that the chance of being detected by the pollster doesn't depend on the (invisible to him) method of communication, hence the sampling is proclaimed to be fair, by declaring it to be so. This is exactly how the experimenters in the Bell tests establish that their sampling is fair (at detectors or polarizers), independent of the hidden variables -- by proclaiming it to be so.

Now suppose in our poll, a question asked "what was your income?" Clearly, the sample is biased here, and the income discovered this way will be higher than the national average. In other words, the poll is more sensitive in "detecting" higher income than lower income people. In the extreme, if the question is "are you homeless," the sample will completely miss any homeless person, i.e. this "detector" is made completely insensitive to the "homelessness" by virtue of the particluar value of the hidden variable (the means of communication = email). OTOH, if the means of communication were walking through the parks at 10PM, and asking anyone encountered the same questions, the results would be biased the other way. Here the hidden variable "means of communication" = "asking people in the parks at 10PM" -- the variable has different value, and the detection profile is quite different.

In other words, the hidden variables will perfectly naturally bias the sample, almost by necessity, since they do affect (or are correlated with) what is detected and what is missed. Ruling out the sample bias by fiat (as is done in every Bell test), is effectively excluding the hidden variables upfront, by declaring it so.

Extremely challenging, like in "it can't work and it won't ever work, but I hope the government and the industry sponsors won't find that out, at least until I retire, preferably after I am dead."

The whole field of Quantum Computing is a mathematical abstraction (fine, as any pure math is, as long as you don't try to claim that's how the real world works). Its vital connection with the real world is based on a highly dubious (even outright absurd, according to some physicists, including Einstein) conjecture about entangled quantum states (roughly, a special kind of "mystical" non-local correlation among events) which was actually never confirmed experimentally. And without that quantum entanglement the whole field is an excercise in pure abstract math with no bearing on reality.

While there were number of claims of an "almost" confirmation of this kind of quantum correlations (the so-called Bell inequality tests), there is always a disclaimer (explicit or, in recent years, between the lines as the swindle got harder to sell), such as "provided the combined setup and detection efficiency in this situation can be made above 82%" (even though it is typically well below 1% overall in the actual experiment; the most famous of its kind, Aspect experiment from early 1980s had only 0.2% combined efficiency, while 82% is needed for actual, "loophole free" proof) or provided we assume that the undetected events follow such and such statistics, etc. The alternative explanations of those experiments (requiring no belief in mystical instant action-at-a-distance), which naturally violate those wishfull assumptions, are ignored, or ridiculed as unimportant loopholes when forced to debate the opposition, by the "mystical" faction. After all, without believing their conjecture all the magic of quantum computing, quantum cryptography, quantum teleportation, along with funding, would vanish.

For those interested in the other side of these kinds of claims, why it doesn't work and why it will never work, check the site by a reputable British physicist Trevor Marshall, who has been fighting, along with a small group of allies, the "quantum magic" school for years:

Quantum Mechanics is not a Science"

Unfortunately, the vast bulk of the research funding in this area goes to the mystical faction. As long as there are fools with money, there will always be swindlers who will part the two.

For a more popular account, accessible to non-physicists, of the opposing view, you can check a site by a practical statistician (and general sceptic) Caroline Thompson:

Caroline Thompson's Physics

vlax worte: That 5% was a concession to the handful of linguists (mostly anthropologists) who still take some portion of Sapir-Whorf seriously.

Give it a rest. Add the ethnographers, media theorists, cultural theorists, historians of science, etc. etc. etc. and together they far outnumber the hard linguists. As Daniel Chandler notes, there is a "broad acceptance" within academia of a weak version of the Sapir-Whorf hypothesis.

See: http://www.aber.ac.uk/~dgc/whorf.html

A lot of geeks resist the idea that language constrains reality because they can't bear the thought of their world being C++. :-7

Danny