Light's speed limit is safe for now
24 February 2012 by Robert
Garisto
News that the faster-than-light neutrinos that caused a
sensation in September may have a mundane explanation – a faulty
cable connection – comes as little surprise. There has never
been a convincing physical explanation for the result.
In the weeks following the original announcement theoretical
physicists wrote dozens of papers attempting to explain the
results, many of which found their way to me. As an editor at
the journal Physical
Review Letters, my job is to consult experts in the field
and ultimately decide which manuscripts to publish. I saw
nothing to convince me that they had cracked it.
First, a quick refresher. The result came from the OPERA
neutrino detector based at the Gran Sasso laboratory near
L'Aquila, Italy. OPERA detected neutrinos fired from CERN, about
730 kilometres away in Geneva, Switzerland, and found they
arrived 60 nanoseconds earlier than expected, implying they
zipped along at one part in 40,000 faster than the speed of
light.
This is as reassuring to a physicist as telling a mathematician,
"I added 2 and 2 and got slightly more than 4". Physicist Jim
Al-Khalili publicly offered to eat
his boxer shorts if the results are proved correct.
The experiment was not conceptually difficult. It calculated the
neutrinos' speed as you would do for any moving object: distance
divided by time. However, the researchers had to precisely
measure the neutrinos' path through the Earth and synchronise
the clocks at each end to an accuracy of a few parts per
million. It now seems they made an error, though we are awaiting
confirmation.
Another possibility still worth considering is that the result
is a statistical fluke. Data are sometimes described as
"statistically significant" if they reach the 95 per cent
confidence level. This is a disquietingly lax standard. It
doesn't mean that there is a 95 per cent chance there is an
effect. It means "supposing there is no effect, you will still
get data showing one 5 per cent of the time". Do hundreds of
trials and you will see many spurious effects, even at the 99
per cent confidence level. Quite a number of experiments have
disagreed with the standard model of particle physics at the 99
per cent confidence level but have fizzled out after more data
was taken.
However, the OPERA team claim a confidence level greater than
99.9999999 per cent. In other words you would have to do
billions of trials in order to expect to get this result by
chance.
So why were physicists underwear-eating-sure that there must be
a mundane explanation? Because, when it comes to data that
conflicts with established theory, we tend to be Bayesians – we
take our prior confidence in the established theory into
account. As Andrew Pontzen of the University of Cambridge and
Hiranya Peiris of University College London wrote
in New
Scientist in 2010:
"Bayesian statistics shows us that the anomalies in the data are
insufficient on their own to motivate drastic revision." We also
need a plausible theory.
Sure enough, the theories soon started rolling in. In the first
few weeks after the announcement, physicists produced more than
a paper a day on the subject. Many put forward new models
explaining how neutrinos could travel faster than light. The
rest either tried to find flaws in the experimental techniques
or pointed out constraints which rule out most of the new
models.
Many of these papers were submitted to Physical
Review Letters. Thus far we have received over 50 papers and
published just three, all in the last category (Physical
Review Letters, vol 107, p
181803, p
241802 and p
251801).
All reach the same conclusion: models which explain the result
by breaking relativity are ruled out. It doesn't matter how you
do it, if you break relativity, your idea is a goner.
How so? The paper by Andrew Cohen of Boston University and his
Nobel laureate colleague Sheldon Glashow (vol
107, p 181803) is the easiest to explain. Imagine an
aeroplane going faster than the speed of sound. It creates a
shockwave of air which you hear as a sonic boom, and which
carries energy away. An analogous thing happens to a particle
going faster than the speed of light: it loses energy via a
shockwave of particles. And yet the OPERA neutrinos arrived with
no energy lost. So there was no shockwave and thus no
faster-than-light travel.
Whenever a paper proposing a new model arrived on my desk, I
would ask the authors to explain how they circumvented this
constraint. None did so convincingly.
So even before the faulty-cable-connection news, boxer-betting
Bayesians were resting easy. Now, they are likely to move on to
something else.
Robert
Garisto is an
editor at Physical
Review Letters, published by the American Physical Society and
based in Ridge, New York
Source: New
Scientist
1 2 3 4 5 6 7 8 9 10 Newest
articles
|