“This means that fact-checking is the fake solution to the real problem of fake news. It helps debunk the simple hoaxes, but it only adds undeserved gravitas to fake political narratives and metanarratives.”
he term “fake news” has been used to describe everything from nonsense fabricated by Macedonian teenagers, to Breitbart, to The New York Times. That is, the term has been applied to people who make up stories with zero factual basis, to organizations with explicit political bias that report on real events, and to “the newspaper of record.” To some, it may seem dangerous to cast the “fake news” epithet so broadly. Sam Harris, for example, has lamented the equation of all these sources of media into the same category. And The New York Times itself has recently staked a claim on “truth“ with its first brand campaign in a decade, which presents the newspaper as the place to find truth in a world of conflicting and unreliable sources. There are certainly large differences between The New York Times and pure hoax websites. They cannot both be fake news in the same way, but they can both be fake news.
There is a simple way that factual news can be fake: selectivity. A purely factual publication can create a false narrative by means of including certain facts, while excluding certain other facts. Genuine facts can have genuine alternative facts, and the choice between facts is determined by deeper assumptions, ideologies, cognitive biases, and philosophies that do not lend themselves to be listed and stamped for approval by “fact-checking” organizations.
Generally, this fact-selectivity is simply understood as bias in media. Charges of media bias are nothing new. Jim A. Kuypers of Dartmouth College investigated the issue of media bias in his 2002 book, Press Bias and Politics. In this study of 116 mainstream U.S. papers, including The New York Times, The Washington Post, the Los Angeles Times, and the San Francisco Chronicle, Dr. Kuypers concludes that the mainstream press in the United States tends to favor liberal (American left) viewpoints. Dr. Kuypers identified liberal bias in the reporting of a variety of issues including race, welfare reform, environmental protection, and gun control.
Journalism may be biased, but it probably has been biased in some way or another for as long as it has existed. So how does mere bias justify the new moniker of ‘fake news’? Evidence suggests that at the same time that mainstream sources like The New York Times have literally framed themselves as objective authorities, their bias has increased to new heights. According to a nationally representative survey conducted by Indiana University scholars Lars Willnat, David Weaver, and G. Cleveland Wilhoit, the ratio of Democrat journalists to Republican journalists remained at approximately 2:1 from 1982 to 2002, but jumped to 4:1 by 2013. (The reader will notice that this large change cannot be attributed to a special repulsion to Donald Trump.)
Wishful believers in journalistic ideals may want to claim that reporters can remain impartial despite the rising ideological imbalance in their own ranks. But according to Gallup polls, the American public would disagree. In approximately the same timeframe that the already-high Democrat to Republican journalist ratio doubled, so did the ratio of Democrat to Republican approval of news accuracy. In 2017, proportionally four times more Democrats than Republicans thought that “news organizations get the facts straight.” When the question was last asked in 1998, 2000, and 2003, the Democrat-to-Republican news accuracy approval ratio was approximately 2:1. So the true metanarrative of news is that it has always been biased and that it is getting remarkably more biased. If that is the true metanarrative, then The New York Times’ “truth” brand is the fake metanarrative.
The point is not that news should necessarily be more conservative or right-leaning. That is just the dimension on which there exists data and documentation, which makes it a useful example. Reality can be distorted by bias in many dimensions—not just the Democrat-to-Republican dimension. There is a more fundamental bias that underlies the fake metanarrative, which is a foundation of the political bias in respectable news outlets. There is something about the nature of reportable political facts themselves, which tends to bias reporting. In the political context, the only facts that can be reported are the least important things to report. The following example reveals this unfortunate property of political facts.
What You Don’t Know Will Hurt You
When federal subsidies for college education were expanded in the 1960s with the goal of engineering social mobility and equity, some results were obvious: more students had the chance go to college. Identifiable students could be pointed to as instantiations of the fact that the subsidies were achieving their goals. These facts were naturally reported by The New York Times. But something was missing. In the 1960s, there was no tangible incarnation of the 1,000% increase in tuition prices that would occur over the next several decades. There was, at the time, no fact about how the inflating proportion of Americans with degrees would reduce the relative economic value of each degree, further reducing the ability of graduates to pay for those degrees. Even today, there is no concrete fact about what the contributions of student loans to the record high personal and national debt will do to the U.S. economy, the millions of people who depend on it, and the billions of people who depend on the heavily-linked global economy.
The problem is not just a simple failure to consider consequences. The problem is an asymmetry in considerability. There are certain economic theories that would predict unwanted consequences. But even if we assumed the sort of economic theory that would predict rising tuition prices and debt, such theories could not tell us how large the increases would be, how long they should take to materialize, or whether the increases would be observable at all, or be masked by countervailing factors. But we could immediately see, count, and touch the specific students who were to be helped by subsidies. So in the political debate concerning subsidies, we were left with solid facts on one side and nebulous theory on the other. The facts were well suited to be reported in The New York Times; the nebulous theory was not. Thus the totality of “all the news that was fit to print” composed a dangerously shallow and imbalanced political narrative.
In my exposition on alternative facts, I explained that for many political facts, there are equal and opposite alternative facts, which exist in alternative units of analysis. But here, things get a little more complicated; sometimes the alternative facts are not fully knowable. They exist in alternative units of temporal analysis (year vs decade vs century). As in the college subsidy example, the immediate effects are concretely factual, while the slower effects are not available for direct rational comparison. And worse yet, the importance of these alternative facts increases as their knowability decreases. The outcome of the decade is more important than the outcome of the year. It is also a lot less knowable.
A true (or at least truly coherent) political narrative would weight emphasis by importance. That is, the things that really matter should be most often cited and discussed. To do otherwise, while claiming to be a central source of truth, would be fake news. This is a big problem for newspapers who fancy themselves to be purveyors of the political truth. In the political context, the facts that can be reported in a newspaper are the least important things to report.
Not everything that matters is fully knowable, and what we are less likely to know is most likely to matter. And this epistemic asymmetry is very good at fooling humans. It is the tragically matching puzzle piece to a psychological asymmetry called the availability heuristic. Humans tend to overestimate the probability of events which are easily brought to mind and underestimate the probability of events which are more effortful to imagine. Concrete facts are easy to think about, while abstract theoretical possibilities require effort to conceptualize. So not only will the most important ideas be underreported, they will be underappreciated when they are reported. The steady diet of easily digestible surface-level facts makes narratives alternative to those facts seem to be comparatively remote and not worthy of consideration. That is, people give too much credence to trivial concrete facts—and not enough credence to gravely important abstract possibilities.
These viciously synergistic epistemic and psychological asymmetries can be found hiding under the surface of many political issues. If one side of a political conflict benefits from the asymmetric darkness (as educational subsidies did), that is the side that has underestimated dangers. One can see asymmetries operative in the issues of gun control, welfare, healthcare socialization, affirmative action, drug prohibition, and speech codes, to name a few. Policies on all of these issues may have some easily reportable factual outcomes and some grave abstract possible outcomes, which are far more important.
“Do something” is the structure of many left-wing political impulses, while conservatives, roughly speaking, prefer tradition and little change.
The epistemic and psychological asymmetries structurally support short-sightedness, interventionism, and authoritarianism. Policies thusly propped-up could possibly be politically right or left-wing. For example, while education subsidies were primarily a left-leaning initiative, the “War on Drugs” was mostly a right-leaning effort. Although conservatives are not immune from being fooled by asymmetries, left-leaning policies are more often bolstered by the epistemic darkness. “Do something about guns.” “Do something about poverty.” “Do something about racial disparity.” “Do something about gender disparity.” “Do something” is the structure of many left wing political impulses, while conservatives, roughly speaking, prefer tradition and little change.
The ‘Do something’ policies are easily supported by reporting facts about who is hurting and whom the policies will help. Anti-do-something policies typically cannot be supported by surface-level factual statements; they require either an appeal to tradition—or abstraction about possible higher order effects. People who are prone to do-somethingism should be expected to overvalue daily reporting of events (things to do something about). Knowing about the epistemic and psychological asymmetries of political facts should lead us to expect a high proportion of progressive do-somethingists to be involved in journalism and to promote their perspective through their work. Thus, the documented political bias in the news may be largely a manifestation of the inherent nature of news itself.
Facts are prone to triviality and irrelevance; theory is prone to naïveté and factual falsity.
If the news is a bad place to find a good political narrative, what else do we have? The only window we have to glimpse a portion of the unknown possibilities left out by the news is theorizing through reason. But theory is notoriously unreliable as well. Facts are prone to triviality and irrelevance; theory is prone to naïveté and factual falsity. Where facts fail us, not just any theory will do. Only theories that are logically coherent and falsifiable, yet not empirically falsified should be given any credence at all. But those theories that pass logical and factual scrutiny should outweigh their alternative facts in decisions about the future. And political decisions are all decisions about the future.
All of this means that the best possible political narrative would be less filled with facts than with reason. Given finite capacity, as fact content increases beyond a certain point, some important abstract reasoning is necessarily excluded, meaning the fakeness of the overall narrative increases.
This means that fact-checking is the fake solution to the real problem of fake news. It helps debunk the simple hoaxes, but it only adds undeserved gravitas to fake political narratives and metanarratives. We do not know to what degree factually false news hoaxes influence election outcomes. However, Alan Gerber, Dean Karlan, and Daniel Bergan of Yale University found in a controlled experiment that consuming reputable (but biased) news in an ecologically valid manner affected voting behavior. Voters who were randomly assigned a subscription to The Washington Post were 8% more likely to vote for the Democrat in the 2005 Virginia gubernatorial election than were voters randomly assigned to the control group. This finding suggests that the narratives crafted by factually correct yet biased news are pervasive and have real influence on the course of society. And given that the inherent bias in the deep structure of news promotes short-sightedness, interventionism, and authoritarianism, the influence wielded by news narratives is likely deleterious (even if the Democrat was a better choice for governor in Virginia).
This is not a comprehensive account of all of the problems in news or the media industry; it is merely meant to reveal one important problem. Mainstream news is necessary and useful—but only when it is understood to be shallow, biased, and fatally incomplete. Nothing so shallow as news reporting should take the role of political guide, nor proclaim itself to be “the truth.” All of this has a simple practical take-away for anyone interested in seeking truth in political narratives. Instead of asking what we know about any given issue or policy, ask what we do not know. It will pull you slightly away from the comfortable ground of concrete facts, but paradoxically, put you a little closer to reality.
Stuart Doyle is a Recon Marine with a BA in Neuroscience & Behavior from Columbia University and an MS in Criminology from the University of Pennsylvania. He has previously written at Quillette.