by Hugo Mercier
the same side that leads people to polarize.40
It seems clear from the preceding that justifications for
beliefs we already hold aren’t always inert. Whether they are
self- generated or provided by people who agree with us, they
can push us toward more extreme versions of the same beliefs.
Why?
When we evaluate justifications for our own views, or views
we agree with, our standards are low— after all, we already agree
with the conclusion. However, that doesn’t mean the justifica-
tions are necessarily poor. In our search for justifications, or when
we’re exposed to the justifications of people who agree with us,
we can also stumble on good reasons, and when we do, we should
recognize them as such. Even if the search pro cess is biased—
we’re mostly looking for justifications that support what we
already believe— a good reason is a good reason, and it makes
sense to change our minds accordingly. For instance, a math-
ematician convinced that a conjecture is correct might spend
210 ch ap t er 13
years looking for a proof. If she finds one, her confidence in the
conjecture should be strengthened, even if the search pro cess
was biased (as she was looking to prove her conjecture, not
refute it).
In the mathematician’s case, there can be no polarization, only
an increase in confidence: the proof supports exactly the con-
jecture, not a stronger version of it. By contrast, everyday argu-
ments are much less precise: they point in a general direction
rather than to an exact conclusion. Most arguments, say, against
capital punishment, are not arguments for a specific position—
that the death penalty should be legal only in such and such cases,
with such and such exceptions. Instead, they tend to be argu-
ments against the death penalty in general—if it is state-
sanctioned murder, then it is a reason to get rid of the death pen-
alty across the board. Piling up such arguments can lead not only
to an increase in confidence but also to polarization— a stronger
opposition to the death penalty in this case.
Polarization does not stem from people being ready to accept
bad justifications for views they already hold but from being ex-
posed to too many good (enough) justifications for these views,
leading them to develop stronger or more confident views. Stil ,
if people have access to a biased sample of information, the out-
come can be dire.
Many commenters have linked the perceived increase in po-
liti cal polarization (in the United States at least) with the rise
of social media. In this prevalent narrative, social media feed us
a diet of news and opinions that agree with us— because we tend
to follow people with the same po liti cal leanings, or because the
algorithms that decide what news we see have adapted to our
preferences, creating so- called echo chambers.41 Legal scholar
Cass Sunstein even devoted a book to the issue: #Republic:
Divided Democracy in the Age of Social Media.42
f u t il e fa k e ne w s 211
Could it be, then, that an eagerness to justify our views, and
social media providing us with an endless source of reinforcing
justifications, are responsible for what is often described as one
of the worst prob lems in con temporary U.S. politics?43 As the
title of an article in Wired put it, “Your Filter Bubble Is Destroy-
ing Democracy.”44 In this light, could fake news, and partisan
news more generally, not be the mostly innocuous providers of
post hoc justifications I have portrayed, but a grave threat to our
po liti cal system?
How Polarized Are We? (Well, Americans, Really)
There might be a nugget of truth in the narrative of social media–
fueled polarization, but it is hidden under layers of inaccuracies
and approximations.
First, the degree of polarization is often exaggerated (through-
out this section, I mostly discuss the U.S. case, which is the best
studied). As po liti cal scientist Morris Fiorina and his colleagues
point out, the proportion of in de pen dents ( people who are nei-
ther Republicans nor Demo crats) hasn’t decreased in de cades; if
anything, it has increased in recent years, rising to 42 percent of
the population in 2017.45 Likewise, most Americans think of them-
selves as moderate, rather than conservative or liberal, a propor-
tion that has remained roughly constant over the past forty years.46
Moreover, a large majority of Americans think that Republican
and Demo crat politicians should compromise, and this is still true
for a plurality of those with the most consistently liberal or con-
servative views.47 Fi nally, on most issues only a minority of re-
spondents hold extreme views: for instance, little more than
10 percent of Americans polled said that there should be no re-
strictions on gun owner ship, or that only law enforcement officers
should have guns (an extreme opinion in the United States).48
212 ch ap t er 13
This lack of polarization can even be observed in the be hav-
ior of the vast majority of social media users.49 On Twitter, the
1 percent most active users behave according to the polarization
narrative, overwhelmingly sharing content that supports their
po liti cal stance. By contrast, the other 99 percent tend to depo-
larize the informational environment: the content they share is,
on average, more po liti cally moderate than the content they
receive.
The impression of increased polarization is not due to
people developing more extreme views but rather to people
being more likely to sort themselves consistently as Demo crat
or Republican on a range of issues.50 This increased sorting is
in part the outcome of people becoming better informed about
where Demo crats and Republicans stand on key issues.51 In
2000, barely half of Americans understood that presidential
candidate Al Gore was to the left of his opponent George W.
Bush on a range of central issues, such as what the level of gov-
ernment spending should be. In 2016, three- quarters could say
that Clinton was to the left of Trump on these same issues.52
The only reliable increase in polarization is in affective polariza-
tion: as a result of Americans more reliably sorting themselves
into Demo crats and Republicans, each side has come to dislike
the other more.53
But if social media are trapping people into echo chambers,
why do we not observe more ideological polarization? Because
the idea that we are locked into echo chambers is even more of
a myth than the idea of increased polarization.54 If anything, so-
cial media have boosted exposure to diff er ent points of view.
After all, Facebook users are regularly exposed to more “friends”
there than offline, allowing them to see the opinions of people
they would barely ever talk to. An early study by economists Mat-
thew Gentzkow and Jesse Shapiro found that the “ideological
f u t il e fa k e ne
w s 213
segregation of online news consumption is low in absolute terms”
and “significantly lower than the segregation of face- to- face in-
teractions.”55 Another study similarly found “no support for the
idea that online audiences are more fragmented than offline au-
diences.”56 Still another observed that “most people across the
po liti cal spectrum have centrist [online] media diets.”57
Research conducted in other countries reaches the same con-
clusions. In Germany and Spain, “most social media users are
embedded in ideologically diverse networks,” and in the United
Kingdom, “only about 8% of the online adults . . . are at risk of
being trapped in an echo chamber.”58 By and large, people on
social media mostly look at traditional, middle- of- the road news
outlets; when they are exposed to extreme views, these views
tend to come from both sides of the po liti cal spectrum.
Economist Hunt Allcott and his colleagues recently con-
ducted a large- scale experiment to test the effects of Facebook
on po liti cal polarization.59 They paid thousands of Facebook
users to deactivate their account for a month and compared these
users to a control group that kept on using Facebook. People who
kept using Facebook did not develop more polarized attitudes
and did not become more likely to support candidates from their
favorite party. However, on a number of ideological mea sures,
they were more likely to sort themselves consistently as Repub-
licans or Demo crats. On the other hand, the people who
stopped using Facebook were less well informed about the news,
and less likely to have seen news “that made them better under-
stand the point of view of the other po liti cal party.” Another
study found that increased Facebook use was related to de polar-
ization: as people were exposed to views diff er ent from their
own, they developed weaker attitudes.60 These outcomes are in
line with the observation that what ever polarization might be
taking place in the United States is more apparent among older
214 ch ap t er 13
than younger adults, with the former being less likely to use so-
cial media.61
Then, the puzzle should surely be: Why don’t we observe
more echo chambers and polarization? After all, it is undeniable
that the internet provides us with easy ways to find as many jus-
tifications for our views as we would like, regardless of how
crazy these views might be (see how many arguments in favor
of flat- earth theory you can find online). However, the desire to
justify our views is only one of our many motivations; usually,
it is far from being a paramount goal. Instead, we’re interested in
gathering information about the world, information that most
of the people we talk to would find in ter est ing and credible.
Even when looking for justifications, most of us would have
learned from experience that simplistic rationalizations won’t
fly with people who do not share our point of view.62
What to Do?
The main message of this chapter appears to be good news. Many
misguided or wicked beliefs— from the humoral theory of dis-
ease to fake news— are much less consequential than we think.
As a rule, these beliefs do not guide our be hav iors, being instead
justifications for actions we wanted to perform anyway. On the
one hand, this is good news indeed, as it means that people are
not so easily talked into doing stupid or horrible things. On the
other hand, this is bad news, as it means that people are not so
easily talked out of doing stupid or horrible things. If a belief plays
little causal role in the first place, correcting the belief is also un-
likely to have much of an effect.
The fact that people all over the world practiced bloodletting
without having ever heard of humors suggests that had the hu-
moral theory been soundly refuted earlier in the West, it wouldn’t
f u t il e fa k e ne w s 215
have stopped people from wanting to be bled. It was only when
evidence actually showed bloodletting didn’t work that doctors
stopped advocating for it. If a skeptic were to challenge gossip
about the local Jewish population, more rumors would likely pop
up as long as people were keen on going on a rampage, using ugly
tales to scapegoat their Jewish neighbors. A refutation from the
authorities might work not because it would be more convinc-
ing but because it would signal a lack of wil ingness to tolerate
the vio lence. Crowds are calculating enough: in Kishinev, they
paid attention to subtle signals from the police that they wouldn’t
interfere with the pogrom.63
Likewise, refuting fake news or other po liti cal falsehoods
might be less useful than we would hope. As a study mentioned
earlier in the chapter suggests, even people who recognized that
some of their views were mistaken (in this case, some of Donald
Trump’s untrue statements they had accepted) did not change
their under lying preferences (voting for Trump). As long as the
demand for justifications is pre sent, some will rise to fulfill it.
Before the internet made fake news vis ible for every one to gloat
at its absurdity, it could be found in the pages of specialized
newspapers— such as the canards of eighteenth- century
France— with exactly the same patterns as those observed now.
Most of the time, the news was pure sensationalism: one of these
canards announced the discovery in Chile of a creature with “the
head of a Fury, wings like a bat, a gigantic body covered in scales,
and a dragon- like tail.”64 But when people wanted to give voice
to their prejudices, the canards obliged, for instance, by insert-
ing Marie Antoinette’s head in lieu of that of the Fury to please
the revolutionary crowds. And if newspapers couldn’t do it, word
of mouth would. Each individual piece of fake news would un-
doubtedly reach a smaller audience that way, but more would be
created— witness, for instance, the rumors about nobles restricting
216 ch ap t er 13
the grain supply that emerged in de pen dently in countless vil-
lages as the French Revolution unfolded.65
Even if debunking beliefs that spread as post hoc justifications
appears a Sisyphean task, the efforts are not completely wasted.
People do care about having justifications for their views, even
if they aren’t very exigent about the quality of these justifications.
As a decision or opinion is made increasingly hard to justify,
some people will change their minds: if not the most hard- core
believers, at least those who didn’t have such a strong opinion to
start with— which is better than nothing.
14
SHALLOW GURUS
the christian god is an omniscient, omnipotent, om-
nipresent being, who loves all regardless of their faults. Other
Christian beliefs vary as a function of the specific church one
belongs to. Trinitarians believe that god is one, but also three in
one: the Father, the Son, and the Holy Spirit. Catholics believe
that the bread and wine of the Eucharist become the body and
blood of Christ (transubstantiation). Lutherans claim, by con-
trast, that the bread and wine of the Eucharist adopt a dual es-
sence, keeping their material identity while also becoming the
body and blood of Christ (consubstantiation). Other forms of
Chris tian ity, from Calvinism to Methodism, adopt yet further
variants of these views.
Scientists routinely defend notions that would other wise ap-
pear equally strange. It might seem as if you’re currently im-
mobile (or moving at a low speed on a train, say), but in fact
you’re moving at more than 600 miles per hour (the earth’s rota-
tion), plus 67,000 miles per hour (the earth’s revolution around
the Sun), plus 514,000 miles per hour (the solar system’s revolu-
tion around the galactic center), plus 1.3 mil ion miles per hour
(the Milky Way’s movement across space). You, along with the
rest of life on Earth, are descended from unicellular organisms.
Tectonic plates— the big rocks weighing up to 1021 kilograms on
217
218 ch ap t er 14
which we stand— are constantly shifting. When you take a plane,
time slows down because of your speed, but accelerates because
of your altitude. The list is endless, through quantum superpo-
sition to the Big Bang, but the point is easily made: before they
were established, many, maybe most, scientific theories would
have sounded nuts to every body but their creators.1
Some of the most influential intellectuals of the twentieth
century were notoriously obscure writers. Until 1998, the Bad
Writing Contest would select every year a scholar on the basis
of the abstruseness of their prose.2 The last person to win the first
prize was phi los o pher Judith Butler, yet she is just one of the
many Derridas, Kristevas, Baudrillards, and other (formerly)
fash ion able intellectuals known for their opaque prose. My per-
sonal favorite is Jacques Lacan, a French psychoanalyst who
makes the most abstruse postmodern scholar look like a model
of clarity. Here is an excerpt picked pretty much at random from
his latest published seminar:
To cut a long story short, I would say that nature’s specificity
is to not be one, hence the logical pro cess to broach it. By the
