The crisis of science and academia
Universities are considered as the most trusted institutions in Western societies – a possible reason why some readers might view this post as exaggerating, if not conspiratorial. But the sad reality is that trust in science and academic institutions is increasingly compromised. In fact, there are three main reasons why science and diplomas are not a religion and why we should embrace scepticism towards science and academia: (1) technical and scientific errors, (2) corruption and (3) a complicated relationship between science and journalism.
Technical and scientific errors
The analysis of the Corona crisis in academic journals and the media was based on technical and scientific errors. But technical and scientific errors were not unique to the C-crisis. To the contrary, technical errors are a widespread phenomenon in science and academia.
In 2005, Stanford top scientist and epidemiologist Prof. John Ioannidis, one of the most prominent critics of lockdown measures, published a widely known paper with the title Why Most Published Research Findings Are False. According to Prof. Ioannidis the reason why most research findings are false is that the estimated effects and relationships between variables are based on data biases. The statistical computations might be correct, but the data is not based on random samples, which is why the distribution of the items in the data does not reflect the distribution of the items in the real world.
Or let’s take the problem of multicollinearity. When researchers develop a regression model in order to search for explanatory variables, they have to make sure that the input variables do not correlate with each other because the purpose of the regression method is to estimate the independent effect of a variable. But if the independent variables correlate with each other, the effect on the dependent variable would not be independent.
Such scientific errors are widespread in academia and in some cases these errors can have severe consequences on politics, the economy or society. Opinion pollsters, for instance, which can influence voting behavior and the outcome of an election, often use biased data. Another statistical error, which contributed to huge damages in the financial sector, is the value of risk formula, which is based on the assumption that price changes in financial markets are normally distributed, when, in fact, the empirical data follows a Pareto shaped distribution.
During the course of the history of science, technology improved experiments and error detection. As measuring instruments became more and more accurate and revealed larger variations of natural constants, dominating theories turned out to be wrong, of which the consequence was that old paradigms and theories were replaced by new ones that fitted the data more accurately. For instance, the concept of Newton’s mechanics could not explain what physicists had observed at the particle level, which is why Newton’s mechanics was replaced by quantum mechanics.
In addition to technology, a fundamental aspect of the progress of knowledge was the process of trial and error that was pushing the limits of our knowledge. By the first half of the 20th century positivism was the dominant paradigm in science and academia, but as it turned out, there are not only white swans. The principle of falsification, which was famously formulated by Sir Karl Popper, replaced positivism in the academic arena and became the core of critical rationalism and modern epistemology. In math the Austrian mathematician Kurt Gödel once and for all rejected Bertrand Russel’s idea that mathematics is a logically closed system independent from human determinism. In the era of quantum mechanics physics has moved away from deterministic thinking due to the counter-intuitive observations at the quantum level. Logical positivism was replaced by probability.
One of the main problems today is the absence of epistemological debates, in so far that empirical research was separated from theory. In economics, for instance, empiricism is still the dominating methodology, while paradigms such as praxeology, which takes a purely theoretical approach, are completely ignored and neglected.
Empiricism has also remained a prominent paradigm in science and at tech firms in the Silicon Valley. In 2008, Chris Anderson published a controversial article in Wired titled The End of Theory, arguing that with the emergence of the big data paradigm theories have become obsolete. According to Anderson, Google algorithms do not care about explanations and causal relationships, they simply use correlations and it works. But is the hype around Google algorithms legit?
In a Nature article published in 2009, Google researchers argued that their flu trend algorithm could predict the dynamics of flu seasons by finding correlation patterns in Google search requests. As it turned out, however, their flu predictions did not work, which is why Google silently abandoned the project. And this is not the only Google project in this field that failed.
In quantitative sciences but also in medical research correlations are often used to draw theoretical conclusions. As an important and widely quoted rule tells us, however, correlation does not equal causation – a principle that is often ignored in today’s academic practice. A correlation between variables can have intermediating variables that are unknown or the correlating variables can be two independent outcomes of the same cause. Spurious correlations are a widespread phenomenon. Just because a higher number of storks might correlate with a higher human birth rate in rural areas, a theoretical linkage between the two makes absolutely no sense.
But what can mitigate the appearance of technical and scientific errors? Scientific progress depends on pluralism in order to detect flaws. This includes competing theories arguing with each other, getting accepted or rejected through a process of natural selection, but also the necessity of multiple perspectives and methodologies in finding the truth and solving complex problems. The field of artificial intelligence, for instance, borrowed concepts not only from logic, math and engineering, but also from economics, philosophy, psychology or linguistics.
Even if scientists knew all the concepts in their own field, they might not know about theories in other disciplines. Modern academia with its highly specialized expert knowledge relies on inter- or multidisciplinary research to connect the dots between different areas. If complex and multidimensional problems are not solved by multidisciplinary teams with different views, methods and experience, the probability that these problems remain unsolved increases to a considerable extent.
An institutionalized mechanism, which is supposed to avoid scientific errors, is quality management and review processes. But review processes take time and can fail due to human errors of the reviewers or a flawed reviewing process. Another problem of academic studies is the lack of transparency of the data - a problem Nature magazine coined as the reproducibility crisis. In contrast to publications in disciplines such as IT, where data is often publicly available on GitHub or other platforms, many other disciplines suffer from disclosed data, which makes it difficult to reproduce an analysis and to assess potential biases.
Corruption in science and academia
Six months after political scientist Michael J. LaCour had submitted his study When contact changes minds: An experiment on transmission of support for gay equality in Science magazine, the editorial board had to retract the paper because LaCour had made false claims about funding and was accused of committing scientific fraud. In this case fraud was detected and the editors retracted the paper, but it can be assumed that in many other cases corruption remains undetected.
In 2011, three psychology professors published an article titled False-Positive Psychology, which led to a controversial debate within the psychology community about the credibility and reliability of psychological studies. Similar to the so-called Bellford distribution, which is used in accounting fraud detection, the authors developed a simple algorithm called p-curve, whose purpose is to identify p-hacking strategies, by which researchers manipulate the significance level of their correlations.
Fraud and manipulation have gained a foothold in academia because scientists are working under pressure. Studies without results are not getting published and if someone’s research led to nowhere, academics loose funding. Many researchers have focused more on the quantity of scientific output than on its quality. At the same time, it has become more difficult to publish in high quality journals, which is why a large number of journals with lower quality standards emerged.
What is even more problematic than funding especially in disciplines that are less mathematized is that politics can interfere. Government funding is often directed to issues and academics that favor the interests of the establishment such as climate change, virus research, gender, or mass migration – issues that allow politicians and governments to extend their control over their populations.
Government funding serves as gate keeper, by which those, who do not share the opinions of the political elite, are kept outside the academic institutions. One prominent example is the brilliant economist and philosopher Ludwig von Mises, whose skepticism towards the political establishment prevented him from attaining a professorship at a respected university.
Today, governments or companies exploit their connections to academic institutions to more or less buy the research that supports their view. The same can happen when scientists are asked to serve as expert witness at courts. Generally speaking, there is nothing wrong about this kind of practice since judicial or political decisions can rely on expert knowledge, however, if scientists contributed to policy-making, they are naturally leaving the scientific domain. To call policy formulation “science” is grossly misleading. Science is about finding the truth and not catering political interests.
Probably one of the biggest challenges in the academic world is that corruption also hijacked the academic discourse and, to be more precise, freedom of speech – a principle that is a powerful weapon against human errors and corruption. If individuals cannot express freely anymore, governments or certain ideological movements have the power to unilaterally determine what is right and wrong.
Today, freedom of speech gets increasingly undermined by radical ideologies that dominate the campus. During my study at the University of Edinburgh, for instance, I personally experienced how save space policies were exploited to discriminate against unpopular positions. The student union organized a session in order to support BDS demands – policies that were supposed to boycott Israeli products and academic cooperation with Israeli scientists. After students, who opposed the BDS demands, signalized their opposition, BDS supporters brought forward a motion based on save space policies in order to remove the student from the hall. The Telegraph even reported about this incident.
Within the context of the Corona crisis, the attacks against freedom of speech became more severe. The media narrowed the space of what can be said and what can’t be said and even discredited critical perspectives by labeling them as conspiracy theorists or fascists – a strategy that has been traditionally applied by authoritarian regimes.
Science and journalism: it’s complicated
In the past few years science journalists were faced with a difficult responsibility. They saw their main purpose in the reduction of an ever growing complexity to an extent as of which is appropriate to the average reader. As a result, science journalism tended to hide the complexity of problems.
In addition to this, journalists often lack time, resources or the intellectual capabilities to detect scientific errors, to understand the technical details or to put small pieces of information together into a broader theoretical context. It would not be surprising if science journalists relied on a publisher’s or scientist’s authority rather than on their own scientific judgement.
But human errors are not the only reason why the relationship between science and journalism has become complicated. In fact, another problem is sensationalism, which can get out of control and lead to hype and panic. Journalists have a natural incentive either to hype promising scientific results or to portray them in a dramatic way. Claims are frequently confused with arguments backed by actual empirical evidence and single studies and their impact on knowledge formation are often overstated.
In 2015, three Dutch biomedical researchers published a widely quoted study investigating the frequency of words such as ”robust”, ”novel”, “unprecedented” and “innovative” in medical journal databases between 1974 and 2014. As the study found, scientists have increasingly used more optimistic expressions about their results and tend to hype their own research more than in previous decades. Scientific hype combined with media sensationalism can quickly lead to far-fetched conclusions. The development of scientific knowledge, however, is a slow and decentralized process incompatible with the modern news ticker bubble.
Science journalists have grossly failed to serve as a necessary corrective against errors and corruption. They failed to inform the public about the crisis of science and academia and instead even contributed to hype and panic and to the silencing of critical views.