-1
$\begingroup$

There is a classic saying which goes more or less like this:

If your theory is not verified by the experiment then it is wrong!

Specifically for the field of particle physics, lately experiments done in particle accelerators present extraordinary results clashing with theoretical predictions up to 7σ discrepancies. The idea is these experiments to be more or less replicated by independent sources and research facilities and if the results are consistent then we can establish a confidence level about the results of the experiment.

But is this really so? What is the point of replicating an experiment that may be conceptually wrong in methodology and design or technically unfit to test a theory?

Who or what is checking out the experimenter?

Are today experiments under the high level of accuracy and complexity presented more prone to errors and false conclusions?

My concern about this is because a wrong conclusion from an experiment(s) can lead science historically to a wrong or death path, waste of time and resources.

$\endgroup$

3 Answers 3

4
$\begingroup$

When I started with particle physics experiments back in the '70's, there was a large number of independent experiments gathering data and slowly building up what is now the mainstream standard model. The number of authors in the papers at the time was order of 20, and many such experiments could be carried out by the hundred of groups studying the scattering of particles in bubble chambers and publishing the results. Now high energy physics papers are signed by thousands.

The particle data group laboriously added the statistical and systematic errors correctly of each experiment and came out with the results , changing continuously over the years. The only consensus is that the errors , statistical and systematic, should be smaller than 5 sigma for the valid experiments. As an example look at the rho decays and how the errors are calculated at page 4.

As larger and larger expensive instrumentation became necessary to gather more knowledge of the reactions, the groups became larger. The number ofthe same type of experiment at LEP was 4 exactly so as to have 4 independent confirmations for measurements, in order to reduce the possibility of systematic errors entering in the calculations. For LHC it is two, for the same reason . Then there are the independent experiments in the US.

Are today experiments under the high level of accuracy and complexity presented more prone to errors and false conclusions?

The results of the current experiments are scrutinized by many people ( remember the thousands of authors) , the errors calculated to the best of their ability, and if discrepancies are found in errors between independent experiments a lot of thought and effort go into trying to resolve mathematically the errors, before declaring new physics.

The OPERA result of super-luminal neutrinos, for example, is not validated by other oscillation measurements and the experiment itself found two possible systematic errors. Here is the history.

However, the ICARUS collaboration, which operates another detector in the same neutrino beam, analyzed its own data to show that there was no sign of the energy loss required by the Cohen and Glashow mechanism, a significant blow to the experimental finding.

Finally, in June the OPERA collaboration completed their own follow-up measurement with the fiber-optic connection fixed. The team found a speed consistent with both the light-speed limit and results from three other experiments using the same neutrino beam. From start to finish, the whole story took a bit under nine months, at the end of which physics was returned to the status quo.

$\endgroup$
1
  • $\begingroup$ In the LHC, there are two general purpose experiments, yes, ATLAS and CMS. Then there are two other main experiments, ALICE and LHCb, together covering all the four collision points. Then there are a number of smaller experiments, such as TOTEM. ATLAS and CMS in principle do they same thing and can thus cross-check each other, whereas the other experiments are focused on other things. $\endgroup$
    – a20
    Jun 14, 2022 at 13:01
3
$\begingroup$

Other experimentalists. Independent replication of results is important. Incremental advancements in the accuracy or refinement of the statistics by having more data. Refinement of theories as more data becomes available. Experiments even for studying the same subject may also use different equipment and the people running them may have different viewpoints and biases.

Some of that may seem like a waste of resources, but over time that is how a consensus of opinion is reached.

$\endgroup$
4
  • 2
    $\begingroup$ I object to your "a consensus of opinion is reached". Physics is not validated by votes. There should be errors given with statistical and sysematic ones clearly defined and a statistical confidence level, for particle physics over 5 sigma. $\endgroup$
    – anna v
    May 11, 2022 at 11:02
  • $\begingroup$ @annav I agree. I have seen reports about discrepancies found over 5σ but very few give also the σ value for the confidence level of the statistics. $\endgroup$
    – Markoul11
    May 11, 2022 at 11:21
  • $\begingroup$ @annav But consensus doesn’t mean everyone is in agreement. It also doesn’t mean that the majority rules, but generally there is some sort of agreement among experts after a while $\endgroup$
    – UVphoton
    May 11, 2022 at 11:38
  • $\begingroup$ I've always pined for a panel post-mortem inquiry by professionals and historians of science studying "results" withdrawn or expelled from the PDG, and the reasons and logic behind that. I've seen wildcat remarks of the kind, so I assume, at some internal level, the community is thinking about such things, and maybe the lessons "stay in Vegas"... $\endgroup$ May 11, 2022 at 13:32
2
$\begingroup$

Replicating experiments checks for human, equipment and random errors. These errors, I believe, are by far the most frequent ones, so it is always necessary to replicate experiments.

If experiment is conceptually wrong or at least unsound, there are other researchers reading your results that can raise objections to the methodology.

If no one raises objections, there is not much that can be done about it. However, the new results needs to be explained by new theory and this new theory then needs to be tested by new experiments. If this new theory proves to be false, either new theory needs to be invented or the original experiments reevaluated.

This iterative method should always lead science back to the right path and prevent the deathtrap that you are concerned about.

My concern about this is because a wrong conclusion from an experiment(s) can lead science historically to a wrong or death path, waste of time and resources.

You can see it as waste of time and resources, but it is waste only if better method is in existence. If there is none, I would call it necessary investment.

Also, any test that is supposed to prevent this waste can also be conceptually wrong and could also lead to the same problems.

$\endgroup$
7
  • $\begingroup$ "If this new theory proves to be false, either new theory needs to be invented or the original experiments reevaluated." This sounds very logical and effective method but IMO this is done today very unbalanced. Thus, a quite large effort is given to prove a theory wrong if the initial results are positive which is a good thing. However, the reverse is not so much practiced. If an experiment initially proves a new theory wrong very few to none experiments follow. $\endgroup$
    – Markoul11
    May 11, 2022 at 11:41
  • $\begingroup$ @Markoul11 If an experiment initially proves a new theory wrong, than the theory is wrong. Why would we need additional experiments? To prove it is even more wrong than we thought? $\endgroup$
    – Umaxo
    May 11, 2022 at 11:49
  • $\begingroup$ @Umaxo In particle physics that is not true. we plan on additional experiments for the same question. see my answer $\endgroup$
    – anna v
    May 11, 2022 at 11:51
  • $\begingroup$ @annav sorry, I don't understand what you are disagreeing with. I like your answer, you went into a detail of self-checks these experiments have, but I fail to see where it contradicts mine. $\endgroup$
    – Umaxo
    May 11, 2022 at 12:02
  • 1
    $\begingroup$ in answer to your" If an experiment initially proves a new theory wrong, than the theory is wrong. Why would we need additional experiments?" One needs the additional experiments. $\endgroup$
    – anna v
    May 11, 2022 at 12:08