by Muhammad Aurangzeb Ahmad
As a data scientist, I imagined that an argument based on careful analysis of data coupled with sound statistical reasoning and proper used of machine learning should be enough to convince any person of one’s argument. However in many contexts this may actually have the opposite effect. For one, the previous statement may actually sound elitist and there is strong evidence that if people have strong convictions about a certain belief then offering contradictory evidence may actually strength their belief instead of weakening it. Thinking about why people act this way becomes easier if we rather drop the assumption that people are rational and start thinking that people’s rationality is mediated via emotions. Leibniz theorized that one day we would have machines that will be able to calculate answers to any question for us and so people instead of arguing will just say let us calculate. One might argue that the data driven society that we are currently building is taking us close to this ideal. However there is a hidden assumption in this assertion that that all people evaluate evidence in the same manner. The presence of conformation bias and other cognitive biases in humans tell a different story altogether. People are more likely to be skeptical and thorough in investigation if evidence presented to them goes against what they already believe. Even things like what people perceive as the scientific consensus varies from person to person. Thus Creationists pounce over any alleged evidence that “proves” that the theory of evolution is false while neglecting any data that goes in its favor. The point is not whether one can use data to make one’s point but rather evidence is powerless if one has already made up one’s mind, to quote Salman Hameed who studies the public perception of the theory of evolution.
Robert N. Proctor, a professor at Stanford, who also coined the term Agnothology is a pioneer in this area and did a classic study of how the tobacco industry willfully spread ignorance regarding the hazards of tobacco. The tobacco industry employed a massive advertising campaign to create doubt in the mind of the public about the veracity of claims regarding the harmful effects of tobacco. The industry created confusion in the minds of people by pointing out that the studies regarding the harmful effects of tobacco had been done on mice and not people. So-called experts were called to talk about the other side of the debate when in fact there was no other side and the evidence was unequivocal. Inadmissibility of evidence or data nihilism is however different from culturally mediated ignorance. The information age only makes things worse by enabling the creation of echochambers. Thus the Internet has created a free for all lunch buffet of information where cherry picking the facts can be used to prove anything. If that does not work then one does not even need to cherry pick anything, one can just invent alternative facts and feed them to a populace willing to eagerly believe and regurgitate the ‘facts.’ In a quote from BBC, Professor Proctor summarizes and warns, “Even though knowledge is ‘accessible’, it does not mean it is accessed.”
The current political climate of data nihilism did not arise in a vacuum. In the nineties we had the culture wars including the assault on Science in public schools. Post-modernism also played a role in relativizing truth; the Sokal Affair comes to mind here. All this may sound gloomy as journalists even wrote about the death of facts as early as five years ago but now people are even questioning whether one can no longer even trust the data that the government is collecting. This problem is however not unprecedented or unique to the United States. The history of the 20th century is testament to data manipulation by different communist regimes and ideologically driven research agendas like Lysenkoism in the Soviet Union that led to nothing. Denialism is not necessarily a Liberal vs. Conservative phenomenon. Vaccine denialism is a rampant problem in some sections on the left and no amount of evidence can convince a person if they have already come to the conclusion that the world is only six thousand years old.
Then there is the Dunning-Kruger Effect where people will low competency tend to greatly overestimate their ability in a given area of expertise. Couple that with culturally induced ignorance and we have a recipe for disaster. Confucius had a maxim that real knowledge is to know the extent of one's ignorance. How does one even begin to talk to people who think that they have the best of knowledge and all their opponents are ignorant? The dominant paradigm among the elites and the educated in the West, especially since the fall of the Berlin Wall, has been that people are rational creatures who are motivated by self-interest. The rise of populism and authoritarianism in the world as of late has been a grim reminder that people are far from rational. The phenomenon of data nihilism and culturally induced ignorance do have one thing in common, an appeal to values. Perhaps not all is lost; it may still be possible to change people’s minds and have a semi-rational discussion. Studies show that people are more likely to change their minds if there is an element of social desirability involved. If a person thinks that changing their position in a certain context is non-threatening or desirable then they are more likely to engage and even change their position. If you are a conservative then perhaps you won't change your mind on climate change if you hear it from a scientist but a sermon from your pastor will convince you. Similarly a vaccine denier does not need to hear about the advantages of vaccination from a scientist but rather from a celebrity. Thus one can appeal to people’s values and one can have a discussion; present them with something threatening then they are unwilling to listen. Chris Mooney described this best in his words; “you don't lead with the facts in order to convince. You lead with the values—so as to give the facts a fighting chance.” So data scientists and journalists have a harder task at hand, not only do we have to use data to tell stories but also we have to use idioms and framing that makes the other side listen. That solves half of the problem i.e., the presence of ignorance but if the problem is just data nihilism then all bets are off.