by Hugo Mercier
sending £350 mil ion a week to Brussels that could be redirected
to the health ser vices instead (in fact, the number is nowhere near
that high, and most of the money goes back to the United King-
dom anyway).9 But fake news, which has always existed in one
form or another, was seen as particularly threatening this time
around because social media had vastly expanded its reach.10 In
the three months leading up to Donald Trump’s election, the
twenty most popu lar fake news stories related to the election gar-
nered more than eight mil ion shares, comments, and likes on
Facebook.11 Among the most popu lar fake news were stories
about Hil ary Clinton, Trump’s opponent, sel ing weapons to the
terrorists of ISIS, or the pope endorsing Trump. Through the
sharing of fake news, and of partisan news more generally, social
media have been accused of creating echo chambers that amplify
people’s prejudices and polarize the population, leading to ex-
treme po liti cal views.12
What do the humoral theory of disease, blood libels, and
Trump’s endorsement by Pope Francis have in common? Obvi-
ously, they are inaccurate pieces of information. They are also
linked with outcomes ranging from the clearly terrible (ethnic
attacks, the systematic mistreatment of patients) to the arguably
suboptimal (Trump’s election, Brexit). It would be natu ral to
think that these false beliefs led directly to the outcomes de-
scribed: physicians practice bloodletting because they accept
the humoral theory of disease; ethnic minorities are massacred
because of the atrocities they are accused of committing; people
vote the “wrong” way because they are misled by fake news.
202 ch ap t er 13
If this were the case, we would be dealing with very grave fail-
ures of our open vigilance mechanisms, in which people would
have been persuaded to accept the misleading ideas of influen-
tial physicians, rumormongers, and fake news purveyors. Unlike
some of the beliefs described in the previous chapters, these mis-
leading ideas would have dramatic consequences not only for
others but also for those who hold them: physicians who ask to
be bled, perpetrators of ethnic vio lence who get hurt, and people
who end up voting against their interests.
In this chapter, I argue that this account gets the direction of
causality wrong. By and large, it is not because the population
hold false beliefs that they make misguided or evil decisions, but
because the population seek to justify making misguided or evil
decisions that they hold false beliefs. If Voltaire is often para-
phrased as saying, “ Those who can make you believe absurdities
can make you commit atrocities,” this is in fact rarely true.13 As
a rule, it is wanting to commit atrocities that makes you believe
absurdities.
Every body Bleeds
Reading David Wootton’s Bad Medicine was an eye- opening
experience, revealing how until around a century ago doctors
were not only useless but positively harmful, and arousing my
interest in bloodletting.14 How could this practice have been
accepted for so long? My initial reaction was to trace it back
through the great physicians who had defended it, from Ben-
jamin Rush in nineteenth- century Amer ica to the Hippocratic
writers in ancient Greece. A specific link in this chain was
fascinating: from the eleventh century onward, hundreds
of thousands of people would be bled because a couple of
Galenic manuscripts on the humoral theory of disease sur-
f u t il e fa k e ne w s 203
vived down the centuries to reach the first Eu ro pean medical
schools.
But as I started looking into the anthropological lit er a ture, I
was quickly disabused of my Western- centric views. Far from
being a historical anomaly, bloodletting was practiced all around
the globe, by people who had never heard of Rush, Galen, or the
Hippocratic writers. The Guna (Panama and Colombia) used a
miniature bow to shoot a miniature arrow into the temple of
those suffering from headaches. When someone complained of
headaches, abscesses, or chest pain, the Bagisu (Uganda) sucked
a bit of blood from the ailing area with a hollowed horn. The Iban
(Malaysia) cut a small incision in the back when someone was
afflicted with back pain. The Dayak (Borneo) relied on a heated
bamboo to draw blood from any ailing part of the body. Blood-
letting was also practiced by major non- Western civilizations,
playing a role in ancient Indian and Chinese medicine.15
All in all, at least a quarter of the world’s cultures likely prac-
ticed some form of bloodletting at some point in their history.
In some of them— ancient Greece, ancient China— the practice
was accompanied by complex theoretical explanations. In most,
however, people were content with a cursory “ We’ve got to let
the bad stuff out.”16 If the humoral theory of disease can’t explain
why bloodletting spread in 99 percent of cultures, which have
never heard of humors, it does not explain either why it spread
in the cultures that embraced the humoral theory. Galen devel-
oped sophisticated theories to justify something people wanted
to do anyway: when they are sick or in pain, let a bit of blood flow
to evacuate hy po thet ical internal pollutants.
If bloodletting can be found throughout the world without its
Western trapping, the humoral theory, by contrast, rumors of
atrocities seem to be a standard component of the ethnic riot,
suggesting that these rumors play a significant causal role.17 In
204 ch ap t er 13
fact, the arrow of causality is unlikely to point in this direction,
as there is little fit between the rumors and the vio lence. We find
countless instances of rumors not followed by any vio lence, and
when vio lence does happen, its nature is typically unrelated in
form or degree to the content of the rumors.
When the Jewish population of Kishinev was accused of the
murder of a small boy, the lie took hold because people broadly
believed this ritual to be “part and parcel of Jewish practice.”18
Indeed, alarming rumors surfaced every year before Easter, with-
out any attendant pogrom.19 Shouldn’t this strike us as bizarre?
Who harbors in their midst people suspected of periodically kid-
napping children to bleed them to death? That the same beliefs
did not lead to vio lence most of the time suggests the beliefs
themselves do not explain why the vio lence erupted when it did.
If the local Christian population in Kishinev had genuinely
believed in the blood libel, we might have expected some terri-
ble reprisal, maybe the murdering of Jewish children, or of the
adults thought to be guilty. The reprisal is terrible indeed, but it
bears no relation to the accusations: How is pillaging liquor
stores going to avenge the dead child? In other times and places,
Jewish populations have been massacred, women molested,
wealth plund
ered under vastly flimsier pretexts, such as accusa-
tions of desecrating the host. Even in Kishinev, the allegations
piled up with no sense of proportionality, from kil ing children
to dishonest business practices: “ Those awful Jews. They bleed
our children to death. And they cheat us on the change!” By and
large, scholars of rumors and of ethnic riots concur that “partici-
pants in a crowd seek justifications for a course of action that is
already under way; rumors often provide the ‘facts’ that sanction
what they want to do anyway.”20
What about fake news, then? Can it sway momentous po liti-
cal decisions? Here I focus on the election of Donald Trump, the
f u t il e fa k e ne w s 205
event for which the most data are available. At the individual
level, there was a correlation between viewing fake news web-
sites, which overwhelmingly supported Trump, and being a
Trump supporter.21 At the state level, the more people visited
fake news websites, the more likely the state was to vote for
Trump.22 Does this mean that viewing fake news prompted
people to vote for Trump? Not necessarily. The majority of
people who visited fake news websites weren’t casual Republi-
cans but “intense partisans,” “the 10% of people with the most
conservative online information diets.”23 These people were very
unlikely to have turned from Hil ary voters to Trump support-
ers. Instead, they were scouting the web— not only fake news
websites but also the traditional press— for ways of justifying
their upcoming decision to vote for Trump, or of demonstrat-
ing their support.24
A study by Brendan Nyhan and his colleagues supports this
interpretation.25 Trump supporters were provided with accurate
information correcting some of Trump’s false statements (rather
than fake news, but the princi ple is the same). Most of them ac-
cepted the corrections. Yet the supporters didn’t waver in their
support for Trump. This suggests that the initial ac cep tance of
the false statements hadn’t caused their support for Trump.
Rather, they had accepted the statements because they supported
Trump.
Po liti cal scientists Jin Woo Kim and Eunji Kim observed a
similar pattern when they studied the rumors that Barack Obama
is a Muslim, which circulated prior to the 2008 presidential elec-
tion pitting Obama against John McCain.26 Kim and Kim com-
pared the answers to two waves of po liti cal surveys: one taken
before the rumors started to spread, and one after they had
peaked. The researchers found that the rumors did have an ef-
fect: they made people more likely to say Obama is a Muslim.
206 ch ap t er 13
However, this was only true for people who were already inclined
to dislike Obama. As a result, the rumors had no effect on
people’s general attitude toward Obama, or on the likelihood of
voting for him: accepting the rumor didn’t make people dislike
Obama; disliking Obama made people accept the rumor.
A Reason for Every thing
If people are going to do what ever they want anyway— from
practicing bloodletting to attacking their neighbors— why would
they bother with a variety of absurd and inert beliefs? Humans
are an uber- social species, constantly evaluating each other to
figure out who would make the best cooperation partners: who
is competent, who is nice, who is reliable. As a result, we’re keen
to look our best, at least to people whose opinions we value.
Unfortunately, we’re bound to do things that look stupid or
morally dubious. When this happens, we attempt to justify our
actions and explain why they weren’t, in fact, stupid or mor-
ally dubious. This lets us correct negative judgments, and it
helps observers better understand our motives, thus judging us
more accurately.
We not only spontaneously justify ourselves when our be hav-
ior is questioned but also learn to anticipate when justifications
might be needed, before we have to actually offer them.27 This
creates a market for justifications. But such a market arises only
when we anticipate that some decisions are likely to be perceived
as problematic.
As mentioned previously, the small- scale socie ties that prac-
tice bloodletting typically do not elaborate complex theories
to justify the practice; it is simply seen as the obvious option
when someone is suffering from a par tic u lar ailment. By con-
trast, in larger or more diverse communities, alternative treat-
f u t il e fa k e ne w s 207
ments are sure to be in competition, and physicians as well as
patients have an incentive to justify their decisions. This com-
petition and the attendant debates were certainly impor tant in
ancient Greece, where the humoral theory of disease was de-
veloped by the Hippocratic writers.28 The same competition
existed in Rome when Galen set up shop: it was only after his
treatments were questioned by the local doctors that Galen
developed a book- length defense of bloodletting, drawing
on his Hippocratic pre de ces sors.29 In a small- scale society,
you can practice bloodletting no questions asked, but in more
sophisticated cultures, to go around bleeding sick people, you
need a theory.
As for fake news, it also flourishes, when needed, as a form of
justification.30 In 2016, the year of the presidential election, six
of the top ten most shared fake news stories on Facebook were
po liti cal, from the pope’s endorsement of Trump to an ISIS lead-
er’s endorsement of Clinton.31 By contrast, in 2017, only two of
the top ten fake news stories were po liti cal (including a rather
funny “Female Legislators Unveil ‘Male Ejaculation Bil ’ Forbid-
ding the Disposal of Unused Semen”).32 Furthermore, more
than 80 percent of fake news related to the 2016 elections was
pro- Trump, and conservatives were more likely to share fake
news on social media.33 The abundance of pro- Trump fake news
is explained by the dearth of pro- Trump material to be found in
the traditional media: not a single major newspaper endorsed his
candidacy (although there was plenty of material critical of Clin-
ton as well). At this point, I should stress that the extent to
which fake news is shared is commonly exaggerated: during the
2016 election campaign, fewer than one in ten Facebook users
shared fake news, and 0.1 percent of Twitter users were respon-
sible for sharing 80 percent of the fake news found on that
platform.34
208 ch ap t er 13
Some po liti cal fake news— for instance, “WikiLeaks: Clinton
Bribed 6 Republicans to ‘Destroy Trump’ ”— might sound plau-
sible enough, at least to people with little knowledge of politics;
that is, most of the electorate. But many stories would presum-
ably sound quite absurd to almost every body (e.g., “[Evangeli-
cal leader Franklin] Graham Says Christians Must Support
Trump or Face Death Camps”). In this
re spect, po liti cal fake
news resembles other fake news. In 2017, the biggest hit was “Baby-
sitter Transported to Hospital after Inserting a Baby in Her Va-
gina”; in 2016, the runner-up was “ Woman Arrested for Defecat-
ing on Boss’ Desk after Winning the Lottery.”35 As suggested by
cultural evolution researcher Alberto Acerbi, the most implau-
sible fake news stories, whether or not they are po liti cal, spread
largely because they are entertaining rather than because they
offer justifications for anything.36 The most absurd po liti cal fake
news stories might also owe their appeal precisely to their over-
the- top nature, as they make for great burning-bridges material
(see chapter 12).
Whence Polarization?
When a piece of information is seen as a justification, we can af-
ford to evaluate it only superficially, as it will have little or no
influence on what we believe or do—by virtue of being post hoc.
This being the case, however, we should observe no changes at
all, not even a strengthening of views. After all, a strengthening
of our views is as much of a change as a weakening, and should
require equally strong evidence. Yet it has been regularly ob-
served that piling up justifications reinforces our views and in-
creases polarization. In an experiment, people had to say how
much they liked someone they had just listened to for two min-
utes.37 This confederate appeared, by design, either pleasant or
f u t il e fa k e ne w s 209
unpleasant. Participants who had to wait a couple of minutes
before rating the confederate provided more extreme evaluations
than people who answered immediately after hearing the con-
federate speak. During these extra minutes, participants had con-
jured up justifications for their immediate reaction, making it
more extreme.38
A similar tendency toward polarization has been observed
in discussion groups. In a study, American students were first
asked their stance on foreign policy.39 Doves— people who gen-
eral y oppose military intervention— were put together in small
groups and asked to discuss foreign policy. When their attitudes
were mea sured after the exchange, they had become more extreme
in their opposition to military intervention. Experiments that
look at the content of the discussions taking place in like- minded
groups show that it is chiefly the accumulation of arguments on
