Related articles: LHC alarmists; their lawsuit; Nostradamus and the LHC
Jtankers, a reader of this blog, has asked me about an article written by the LHC alarmists called
It is enough to read the first few sentences to understand their thesis:
Do CERN employees have a culture of superiority? It is understandable that physicists might have a unique culture. But is that culture at CERN so far removed from the rest of society that they do not take the concerns of the rest of society seriously?
First, let me answer this question. Yes, the opinions of the CERN people about physics (not so much about other topics) are dramatically removed from the opinions of the rest of the society. Also, indeed, the CERN people don't take the concerns of the rest of the society seriously. Incidentally, the CERN people are right, too, and the rest of the society should trust the CERN people's opinions about particle physics, otherwise it would be a primitive non-scientific society.
The whole text by the LHC alarmists discusses the perceived arrogance of the particle physicists as well as various conspiracy theories about the destruction of the world by the new collider. That's why I want to write a mildly quantitative sociological essay about the probability that elites are wrong.
They ask various questions, usually invented by other authors such as Sabine Hossenfelder, for example:
Does the antichrist have a penis?
Well, why don't you ask Tipper? :-) This answer is not just a joke. She is indeed the right person to ask. She should know the right answer much more reliably than millions of other people combined. The emperor's assistant is likely to know the length of the emperor's nose more accurately than the average citizen of the monarchy. It brings me to the first point:
Elitism, good and bad
Should the ordinary people - and not quite ordinary people - trust the experts and elites? My key point is that the answer can't be universal. In different contexts, the correct answer is obviously different. It depends whom you mean by an expert.
Whenever the "expertise" is vacuous and the self-declared experts have no non-trivial special skills that have been verified - they only have some opinions that they are enthusiastic about - or whenever the "competition" they had to go through was unfair, fully stochastic, or vacuous, the answer is No.
When someone tells you what you should eat, whom you should love, how you should educate your children etc., because he or she considers himself an "expert", it is likely that he or she has no special credentials that would justify his or her "authority". People have made almost no progress in love or spanking for millenia and even if there were some progress, it couldn't really be attributed to experts.
You shouldn't pay too much attention to such wisdom. This category includes most of the activities of governments. You should obviously ignore most of them and try to minimize their impact on your nation.
However, there also exist cases in which the answer is clearly Yes. In many cases, ordinary people understand that they should trust experts and why. When you buy and drive a new car, you don't want the design to be made by the average citizens. You want the best engineers. You don't expect the average pedestrians to give you the best kernel for your new operating system. You won't buy stocks of an IT company dominated by incompetent people. Most people have common sense concerning these issues.
Why? Because they are a part of their life. They have made similar decisions many times and they have influenced their lives. They are learning things because these decisions are parts of their everyday life.
Expertise: related and unrelated to everyday life
This brings me to my second point. Regular people tend to trust experts if their expertise has some pretty much direct impact on their everyday life.
Clearly, this is not the case of high-energy physics. The existence of quarks and strings doesn't influence the life of ordinary people - maybe except for teachers - at all. A hypothetical lethal threat lurking at the LHC would influence the life of all of us. But such an event is so rare that people haven't developed any intuition for such "big" questions.
These are the reasons why most people don't realize how the expertise is critical to obtain the right (or qualified) answers in these "abstract" disciplines.
You might ask: don't they act rationally? Shouldn't they trust the expertise exactly in the cases when the expertise has been proven relevant for their everyday life? The answer is, of course, No.
Rationally speaking, the expertise cannot be divided to "trustworthy" and "untrustworthy" according to its impact on everyday life. The expertise should be divided to "trustworthy" and "untrustworthy" according to the amount of verification that has (or has not) validated it. It's a very different criterion than the "usefulness" in everyday life. The truth doesn't have anything to do with everyday life.
Expertise: verified and unverified
This brings me to the correct criterion to rationally decide whether some expertise or a membership in an "elite" is relevant or not. A creator of new operating systems has probably tried to write similar code in the past and he or she has either succeeded among the competitors or not. Everyone understands that this process makes top programmers much more - by orders of magnitude more - relevant than average people as far as programming goes.
But the same process occurs in the case of high-energy physics, too. Students have to pass many exams, PhDs compete for jobs, scholars compete for grants and citations as they propose new ways to evaluate the available evidence and to collect new evidence in new papers that compete for attention. None of these procedures is perfect but if some basic principles of merit are obeyed (i.e., unless the system is essentially broken), all of these steps significantly increase the probability that the "truly qualified" people are near the top.
It works in the same way even if the resulting "products" don't influence the everyday life of most people. But when the everyday life is unaffected, most people can't understand that the content of the expertise is nontrivial unless they are experts themselves.
Now, you can legitimately think about far-reaching sociological theories in which whole communities are corrupt, confused, or deceptive. In such a case, the whole mechanism based on competition must be broken. You wouldn't get more qualified people in such a framework.
But when you think about such a sociological hypothesis, you should neither accept it uncritically nor reject it uncritically. You should try to quantify the probability that your "conspiracy theory" is correct.
Probabilities of conspiracy theories
Now, conspiracy theories are usually unlikely. But why is it so? Why is this kind of a conspiracy theory unlikely? And when does it become likely?
If you have a pretty large number of players that are playing a game, one which tries to divide them into the qualified ones and the less qualified ones, and you end up with the situation in which the qualified ones agree about something while the unqualified ones may disagree, it is unlikely that such a mechanism led to the wrong answer by chance.
Every independent step in the competition process affecting every individual expert makes it more likely - essentially by a multiplicative factor - that the chosen elite is "real" and its opinions are more relevant or much more relevant than the opinions of the rest.
Even if you only increase the probability that the "elite" opinion is more correct than the "non-elite" opinion by a factor of 1.05 whenever you add a new person to the process, 100 people make the opinion of the "elite" 130 times more relevant than the opinion of the rest. Geometric sequences increase very rapidly, indeed. Independent pieces of evidence, unless they are almost vacuous, are able to falsify a hypothesis quickly.
This calculation is surely no proof that the elite is right. But it is an indication that the elite has evaluated the available, known evidence properly. Sometimes, the elite doesn't really know or have enough evidence to find a correct explanation. But:
The only plausible way how the scientific elite can systematically prefer a wrong answer even if the evidence supporting the correct answer is available is that the whole system is broken and individual scientists and individual papers actually fail to bring us closer to the truth.
In that case, the geometric series may become decreasing. What do I mean? Imagine that you have a binary question. For example, you ask: is the global mean temperature going to increase by more than 2 °C in the 21st century?
The evidence supporting one answer or the opposite answer is complicated and the papers are often uncertain. The question is: does a large number of papers help us to get closer to the right answer?
If you think about it, the answer to the previous question depends on another question, namely whether the average paper that influences the direction of research is helping to bring us closer to the truth. If it does, many independent papers are likely to make significant progress and the probability that we continue to keep a wrong answer - because of a "conspiracy" - decreases exponentially with the number of papers.
Unfortunately, there always exist situations in which individual papers don't help us to find the truth. For example, if the average author is more motivated to get the "Yes" answer than she is motivated to get the correct answer, it is reasonable to expect that a large number of papers will lead to the "Yes" answer regardless of the correct one. ;-)
This bad outcome is conceivable in climate science where you see a lot of "coherent" interest to obtain the "Yes" answer that may guarantee that the researchers converge to the "Yes" answer rather than the correct answer.
If a new honest paper with some characteristically fuzzy, incomplete climatological evidence would conclude that the probability of a 21st century warming exceeding 2 °C is 40%, while the probability of a smaller warming or cooling is 60%, then a reasonable large sequence of honest, independent papers would settle the answer: the warming will probably be smaller than 2 °C.
However, it is enough for the authors to be 1.5 times more motivated to obtain the "higher warming" answer and the signal is overwritten by the noise. These papers will help us to move closer to the "Yes" answer even though the right answer is likely to be "No". Well, I personally think that the typical climate scientists are much more motivated to find the "Yes" answer than the "No" answer and this bias makes it possible for the noise to overwrite the signal.
Fuzzy arguments vs exact sciences
The figures 40% and 60% mentioned above are meant to be serious estimates even though the precise value depends on the question and the methodology, too. The meteorology and climate science are so uncertain that it is very unlikely to "completely" rule out most hypotheses you can invent (or a semi-qualified person can invent). Even if you're swimming in the river right now, it doesn't mean that it won't be snowing tomorrow. Almost anything can occur by next month. There are so many terms, including "higher-order" corrections that are being routinely neglected, that they make the predictions of the global and continental climate for the year 2100 nearly equivalent to astrology.
In each individual paper, preconceptions and irrational motives are arguably pushing the "current state of knowledge" more strongly than the actual rational evidence.
However, other disciplines are much more exact than climate science. When you propose a wrong theory in high-energy physics - even if it is superficially sensible - the first relevant paper is typically able to rule it out at the 99% confidence level or higher. Instead of 40%, 60%, you have 1%, 99%, or even more extreme ratios. Why? Because theories in high-energy physics typically predict a lot of features including rather accurate continuous numbers (or discrete data with many options) and it is unlikely for a wrong theory to agree with all the experience. Even if your theory only predicts one number with an error margin, you get to the 99% confidence level (for the falsification of the theory) if the disagreement is more than 3 standard deviations - which happens very often.
This fact - the intrinsic accuracy in high-energy physics - makes the content of the scientific work much sharper.
Now, high-energy physicists may also be biased. But they would have to be much more biased to make their whole collective research dysfunctional. They would need to "irrationally" raise the probabilities of the wrong answers not by a factor of 1.5 but by a factor of 100 or more.
Destruction of the world: measuring the bias
Consider the following question:
Is the LHC going to produce a black hole or a strangelet that will destroy the civilization?
Now, as we have discussed, people can use various analyses based either on "observed phenomenology of catastrophic events in the Universe" or the fundamental physics (including estimates about some of its unknown features) to argue that the probability of the "Yes" answer is extremely low.
The most "catastrophic" scientists would tell you that phenomenological analyses imply that the probability of such a catastrophe is 10^{-10} per year or lower. This particular figure would double the probability that the life on Earth ends, relatively to the background threat caused by the finite lifetime of the Sun (10 billion years or so). That would be a significant risk that might even be enough to shut down the accelerators for security reasons.
Still, it would be no "imminent" threat. The Sun may also decide to explode in 2008 because of some timing mechanism that we don't fully understand.
However, more qualified estimates that take more complete bodies of evidence into account end up with much lower estimates of the probability of the doom, something like 10^{-100}. This is clearly negligible for all practical and semi-practical purposes you can think of. Get used to it. Nothing is 100% certain in this world. The probabilities of unusual things are not strictly zero but they are extremely tiny. Probabilities comparable to 10^{-100} are almost certainly not going to be "observed" by the whole mankind. And tiny numbers like that are commonplace in high-energy physics.
I have discussed the technical arguments why the LHC is not a real threat elsewhere - using several methods - but I also find it important for non-physicists to have some moderately sensible sociological arguments and computational methods to follow.
There exists no sensible reason to expect that all particle physicists want to hide a universal cataclysm. Many of them would be extremely happy if they could show that the risk is real. They would surely become very famous and rich. (Un)fortunately, no good arguments or scenarios that would predict such a catastrophe exist. Again, it doesn't prove that particle physicists are right. But it effectively disproves the hypothesis that they are choosing the wrong answer because of a conspiracy. Such a conspiracy would be unsustainable simply because new scientific evidence is much more decisive than the average bias.
The LHC is not going to destroy life on Earth. The probability of such an event is "zero" in ordinary people's (or experimental) understanding of the word "zero". The nonzero number we obtain is just an abstract academic issue.
String theory conspiracy
Let me talk about the probability of conspiracies concerning another important topic, namely the question
Is string theory the right framework to describe the fundamental forces and matter of the Universe?
There could clearly exist a "bias" in every new paper about these topics. This "bias" could be caused by the fact that many people have studied this subject for years and they have seen many things that have made them believe that string theory is right.
A quantitative discussion of "justified asymmetric expectations" could be subtle but we don't need it, anyway. Why? Because virtually none of the papers that are being written is actually trying to answer this "grand" question. This question is surely captivating, it subliminally motivates many people, and it controls the interests of crackpots and other laymen who talk about the subject.
But you won't find a recent paper by a serious researcher whose point would be to answer the "grand question" simply because we don't have any "grand arguments" that could achieve this goal. That's of course good news for string theory because wrong theories of high-energy physics may be falsified almost instantly. Having no tools to decide the question is pretty much equivalent to high chances that the theory is true. The situation is somewhat similar to the second Gödel's theorem showing that the consistency of a powerful axiomatic system cannot be proven within the system. If it could be proven, it would actually mean that the system is inconsistent because inconsistent systems allow you to prove anything.
In physics, we can usually determine the consistency of a system but it is still true that the "silence" - the absence of definitive statements about the validity of a theory - is the best news that a typical, complex enough theory may get. The only other conceivable option is that the theory becomes falsified.
Real papers are trying to answer more specific, "smaller" questions such as
Do black holes emit Hawking radiation?
This question is still "too general" and because the evidence overwhelmingly shows that the answer is Yes, physicists are answering much more concrete questions about the properties of this radiation etc. And even for these "very specific" questions, there is a huge body of evidence supporting the right answer. There are many questions of this kind.
But there is one more question related to - but not equivalent to - the validity of string theory, and it is this:
Is string theory a random man-made mathematical invention that can be forgotten as easily as a new Sudoku?
This question may be studied much more systematically than the question "Is string theory right?" Why? Because one can accumulate a huge body of evidence that the answer is No. How? Because if you assume that the answer is Yes, you can predict a huge number of things that can be shown to be wrong.
You would predict that the rules of string theory can be modified (much like the details of a Sudoku) and that the theory can mutate in many ways etc. However, everything we know confirms the opposite statement, namely that string theory is completely unique and one cannot modify it in any way. It doesn't have any siblings. It has many solutions but they're like mountains on Earth. There are many mountains but only one Earth that follows the same laws.
You would also predict that the theory is likely to fail in new consistency checks that we may construct. You would predict that we should violate some basic physical principles when we take the theory to the limit. There are many ways ho we can take it to the limit - e.g. limiting regions and special places of its moduli space. However, all consistency checks have worked so far and it starts to be sensible to extrapolate and expect that the future ones will work, too.
So even if someone were able to prove that string theory is wrong, there would still exist an amazing quasi-paradox in mathematics:
Why such a large, unique, and unexpectedly self-consistent structure that agrees with all qualitative features of our world exist?
Again, I would agree that this question would belong to mathematics - or quantitative philosophy, if you will - but I would still find this question extremely urgent - more urgent than most questions in "observable physics". And because of the strikingly physical character of string theory, the question above wouldn't be just another "millenium problem" of mathematics. It would still be much more grandiose a problem.
I don't believe that very unlikely objects routinely exist for no good reason. And because of the deeply physical character of string theory, the question above - despite its mathematical character - should still be essential for theoretical physicists even when someone proved that string theory doesn't describe our Universe (which I consider extremely unlikely).
More technical questions
But as I have said, the questions that the experts actually discuss are much more technical - or "modest", if you only care about big questions. And there exists a huge body of circumstantial evidence supporting the right answers to various questions of this kind.
Here I want to say that even though string theory hasn't really been proven to correspond to this Universe - by methods that would satisfy empiricists - many of its qualitative features have already been established. For example, very high-energy, trans-Planckian collisions lead to black hole production.
This is a statement that we can now derive from string theory, too. But even the people who don't work on string theory know the evidence that it is the case. It didn't have to be the case in a generic theory of quantum gravity and most "competitors" would have doubts about (or direct disagreements with) these trivial qualitative propositions.
On the other hand, string theory agrees with all the qualitative statements about physics beyond the Standard Model that have also been determined by other means. Even if all of these qualitative features - derived independently of string theory - are correctly reproduced by string theory, it still doesn't prove that string theory is right. However, if string theory were wrong, it would make the mystery "Why string theory exists?" even more pressing.
Who are string theorists?
But let me also say that from a sociological viewpoint, it is not really important whether the "partial results" consistent with string theory (such as the trans-Planckian black hole production or the existence of Hawking evaporation) imply string theory or not. Why?
Because, as I have already mentioned, string theorists are not writing papers claiming that "string theory is correct" on every page. Instead, they use their methods and tools - based on the highly non-trivial and unique system of stringy equations - to analyze more specific questions such as the character of the Hawking radiation, modes of particles moving in extra dimensions (i.e. the spectrum of new particles that can be observed in the future), or the black hole production in trans-Planckian scattering.
If someone wants to say something relevant for high-energy physics or quantum gravity, he must study these particular questions. Sorry, there is no other "easy" method. And the methods that are used to investigate these questions by the experts include the methods of string theory simply because these methods have proven to be very fruitful and "surprisingly" consistent with the observed properties of the elementary forces and particles. Consistency with the experience is what actually matters in physics. Because no other methods to answer questions unanswered by the framework of local quantum field theory are known, it is obvious that people must use the only methods they have to study these questions - and they are known as string theory - if they want to talk about these questions at all.
Summary
I have jumped into many related topics. But let me summarize some of the key points.
- It is recommended to listen to those elites that have some merit
- Experts who have been chosen by some fair tests or competition are more likely to have some merit than the self-declared ones
- The relevance of the expertise has nothing to do with the relevance of the discipline for the everyday life
- It is much easier for fuzzy and uncertain sciences to become misguided by bias and external pressures than it is for exact sciences
- Sufficiently exact disciplines are able to rule out wrong hypotheses much faster than the fuzzy ones
- The not-yet-falsified hypotheses still don't have to be correct
- However, in the disciplines with a high signal-to-noise ratio, the mistakes are more likely to be caused by a collective ignorance about concepts rather than collective deception or conspiracy
- Small but nonzero probabilities are predicted everywhere in physics
- Too small probabilities are "effectively" zero, especially if the calculated risk is significantly lower than the risks associated with the "natural background"
- While we can't prove that string theory is correct, we can practically rule out the hypothesis that string theory is a random invention that can be suddenly forgotten
- Physicists in serious disciplines are asking smaller, more modest, more technical questions most of the time
No comments: