by Hugo Mercier
“on this pop psy chol ogy B.S.”56
140 ch ap t er 9
The main explanation that had been previously offered for the
inefficiency of po liti cal campaigns is that each side invests in re-
action to the other’s investment, so that their effects cancel each
other out. But this cannot be what happens in the studies re-
viewed by Kalla and Broockman. Who received the treatment—
the flyers, the calls, and so forth— was random, making it impos-
sible for the other side to target these people in par tic u lar.
Po liti cal campaigns just don’t seem to persuade a significant
number of voters, at least in impor tant elections. What about the
wide swings in the polls observed throughout the campaigns
then? A recent analy sis suggests that they are largely artifactual:
when a candidate is perceived to be doing wel , people inclined
to vote for them are more likely to answer the pol s, creating the
illusion of swings, when in fact few people are changing their
minds.57
By contrast with most po liti cal campaigns, the news media
“have an impor tant effect on the outcome of presidential elec-
tions,” as statistician Andrew Gelman and po liti cal scientist Gary
King put it more than twenty- five years ago. However, Gelman
and King specify that this effect is achieved “not through mis-
leading advertisements, sound bites, or spin doctors, but rather
by conveying candidates’ positions on impor tant issues.”58 As a
rule, the main role played by the media is to provide the infor-
mation without which citizens couldn’t make even minimally
informed po liti cal decisions, information such as what party each
candidate belongs to, or what the candidates’ platforms are.
Recent research vindicates Gelman and King’s pronounce-
ment. More news media coverage makes for a more informed
electorate;59 citizens who trust the media more are also the best
informed;60 more informed electorates are less susceptible to
persuasion, not more. As a result, the more news sources are
available to the public, the more the public is aware of what poli-
p r o p a g a nd is t s, c a mp a i g ne r s, a d v e r t ise r s 141
ticians are doing, and the more efforts politicians make to fulfil
their constituents’ wishes.61 At least in wel -publicized elections,
the media and po liti cal campaigns play a largely positive role of
informing citizens— even if, when it comes to American po liti-
cal campaigns, it is easy enough to imagine that the same result
could be achieved at a fraction of the cost.
Advertisers
The sums spent on po liti cal campaigns pale by comparison with
the amounts lavished on advertising. In 2018, more than half a
trillion dollars was spent on advertising worldwide.62 This
money can (in theory) heavi ly influence customers’ preferences,
making them choose more expensive products, or even favor in-
ferior alternatives— for example, buying Coca- Cola when they
are supposed to prefer Pepsi in blind tests.
As for po liti cal campaigns, mea sur ing the effects of advertising
is difficult. Researchers at Google and Microsoft have argued that
to know whether an online ad produces positive returns at all, it
has to be tested on more than ten mil ion people— and that is in
the ideal scenario of a perfectly controlled experiment.63 If adver-
tising effectiveness is so difficult to mea sure, it is not for technical
reasons but because ads have small effects at best, making it dif-
ficult to tell whether they have any effect at all.
Early work on advertising efficiency suggested that most ads
had no discernible effects whatsoever. An early article from 1982
was already asking, “Are you overadvertising?,” the answer being
a clear yes.64 A review of studies conducted in the ten years after
1995 claims to have observed some small but significant effects
for TV ads.65 As is the case for po liti cal ads, the main variable
moderating the effectiveness of consumer advertising is whether
or not the audience has preconceived opinions. Ad campaigns
142 ch ap t er 9
have no effect on consumers who have already experienced a
product.66 This result is impor tant, as it means that advertising
doesn’t function by making some products gain a better image or
appear more prestigious—if this were true, people who know the
product already should be just as likely to be influenced by the
ads. Instead, advertising functions mostly by “giving [customers]
information on inherent product characteristics,” information
that is superseded by personal experience when it is available.67
A sad example of advertising working is that of TV ads for
cigarettes. Cigarettes aren’t exactly a hard sel : long before adver-
tising existed, people had been smoking everywhere tobacco
was available. Merely pointing out the existence of cigarettes
should be sufficient; the effects of nicotine on the brain, target-
ing reward centers and soon making itself indispensable, would
then do the bulk of the work. As expected, cigarette ads were
most efficient when they could tap into a market of people who
had not been aware that smoking was an option, such as young
Americans in the 1950s.68
When advertising affects consumers, it doesn’t affect them in
a way that reflects sheer credulity. For example, the effectiveness
of ads relying on celebrities depends on whether the celebrity
is perceived as a trustworthy expert in the relevant domain.69 By
contrast with relevant expertise, gratuitous sex and vio lence in
ads are more likely to decrease their impact.70
These results might be difficult to believe: we can all think of
celebrities associated with products they have no known exper-
tise in. Boyd and Richerson, defending a bias to do what ever
prestigious people do, mentioned Michael Jordan advertising
underwear, but a better- known example might be that of George
Clooney and Nespresso. As far as I can tel , Clooney has no rec-
ognized expertise in coffee, yet he has come to be associated with
the brand. However, the direct effects of his endorsement are not
p r o p a g a nd is t s, c a mp a i g ne r s, a d v e r t ise r s 143
clear. Nespresso was already growing more than 30 percent a year
before Clooney became its ambassador in Eu rope, in 2006. In the
following years, the brand kept growing at similar rates.71 Nes-
presso had also achieved formidable growth in the United States
long before Clooney started advertising for the brand there, in
2015.72 Ironically, if it is unclear how many customers Clooney
brought in, we know Clooney was brought in by customers: to
reward early buyers for the brand’s success, Nespresso asked
them to select an ambassador. They picked George Clooney.73
Advertising cannot even be blamed for making people choose
sodas (supposedly, Coca- Cola) they wouldn’t favor in a blind
test. Most people are simply incapable of distinguishing between
Coke and Pepsi.74
Even if advertising in this domain had huge
effects (which it doesn’t), not much persuasion would be in-
volved in making people choose Coke over Pepsi, products
essentially indistinguishable by taste or price.
Marketing researcher Gerard Tel is drew from his review of
advertising effectiveness these words of caution: “The truth, as
many advertisers will quickly admit, is that persuasion is very
tough. It is even more difficult to persuade consumers to adopt
a new opinion, attitude, or be hav ior.”75
Patterns of Mass Persuasion
As attempts at mass persuasion pile up, from demagogues ha-
ranguing crowds on the Agora to advertisers vying for our
attention on smartphones, a clear pattern emerges. Mass persua-
sion is tremendously difficult to achieve. Even the most dreadful
propaganda attempts, from Nazi Germany to Stalinist USSR,
have been surprisingly in effec tive at changing people’s minds.
Any message that clashes with our prior beliefs, any in-
junction to do something we aren’t happy to do anyway, is
144 ch ap t er 9
overwhelmingly likely to fall on deaf ears. The Catholic Church
at the height of its power could not get peasants to fast, con-
fess, make penance, willingly pay the tithe, or abandon their
pagan practices. Nazi propaganda failed to make the Germans
abhor the handicapped or like the Nazis. Once voters have their
minds set on a candidate, all the campaign money in the world
isn’t going to sway them. Ads are wasted on consumers who have
firsthand experience of the product advertised.
Mass persuasion fails when it encounters re sis tance. An audi-
ence needs to have positive reasons to believe a message if the
message is to have any effect. The most effective messages echo
the prejudices or serve the goals of their audiences— anti- Semites
defending their hatred with Nazi propaganda, revolting crowds
recycling millenarian themes— but then can we really talk about
persuasion? At best, mass persuasion changes people’s minds on
issues of little import, as when voters select a party whose plat-
form fits with their opinions on significant issues and then fol-
low the party’s lead on less (personally) significant topics.
Clearly, the patterns of mass persuasion aren’t compatible with
widespread credulity. Instead, they reflect a cautious evaluation
of the information communicated, as people decide whether
messages fit with their prior opinions, and whether they come
from reliable sources.
If We Are So Vigilant, Why Are Some
Misconceptions So Popu lar?
In these past chapters, I have detailed the functioning of our open
vigilance mechanisms: how we decide what is plausible and wel
argued, where expertise lies, who is trustworthy, and how to react
to emotional signals. A wealth of psychological experiments
shows these mechanisms function broadly rationally, allowing
p r o p a g a nd is t s, c a mp a i g ne r s, a d v e r t ise r s 145
us to avoid harmful messages and change our minds when con-
fronted with good enough evidence. Open vigilance mecha-
nisms are certainly efficient enough to stop nearly all mass
persuasion attempts from changing our minds.
This optimistic conclusion might seem out of touch with the
list of patently wrong beliefs— power ful witches are afoot, Barack
Obama is a Muslim, vaccines aren’t safe— enjoying widespread
cultural success. Yet professing a mistaken belief doesn’t neces-
sarily make one gullible. In the next six chapters, I explore a
laundry list of misconceptions, from rumors to fake news, show-
ing that the ways in which these misconceptions spread, and
the effects they have on our thoughts and actions, are best ex-
plained by postulating efficient open vigilance mechanisms
rather than outright credulity.
10
TITILLATING RUMORS
in 2015, 20 percent of Americans believed that Barack
Obama, then the sitting U.S. president, had been born abroad.
Forty- three percent of Republicans— the opposition party—
also thought he was Muslim (Obama was born in Hawaii, a U.S.
state, and is a Christian).1
In April 2017, David Dao was forcibly removed from an over-
booked United Airlines plane. The situation was handled so
poorly that he lost a tooth, broke his nose, and got a concussion
(according to his lawyer). After Dao and the airline com pany
settled, rumors were flying around the popu lar Chinese social
media platform Weibo that the settlement had reached the sum
of $140 mil ion.2 Although the real amount was never divulged,
the likelihood is that it was a hundred times lower.3
In early 1969, there appeared in the French provincial town
of Orléans a rumor that young women were being abducted from
the changing rooms of Jewish retailers, to be sent abroad as pros-
titutes.4 In spite of official rebuttals from the police, politicians,
and other authority figures, the rumor grew for several months
before slowly dying over the summer.
Besides these examples of wildly inaccurate rumors, the low
level of accuracy of some rumors is borne out by more system-
atic studies.
146
t i t il l at in g r um o r s 147
In June 1950, the Indian town of Darjeeling was hit by devas-
tating landslides. Psychologist Durganand Sinha studied the
rumors that proliferated in the aftermath— rumors about what
had caused the landslides, the number of casualties, the amount
of rainfall, and so forth.5 The rumors were uniformly false, wild
exaggerations and dramatizations of the actual events. The same
outcome was observed after an earthquake hit the Indian state
of Bihar in 1934.6
The University of Michigan saw a large strike in 1975. Recog-
nizing that “rumors tend to proliferate in times of crisis,” two
psychologists, Sandord Weinberg and Ritch Eich, attempted to
counteract their spread. They set up a rumor crisis center that
employees could call to check the veracity of the rumors they
heard through the grapevine. Only about 15 percent of the re-
ported rumors were accurate.7
Rumors of Crisis
Why are false rumors so common? After psychologists Gordon
Allport and Leo Postman published their influential book The
Psy chol ogy of Rumor, shortly after World War II, most theories
of rumor diffusion focused on the state of mind of those who
believe and spread rumors.8 As a review put it, “Rumor genera-
tion and transmission results from an optimal combination of
personal anxiety, general uncertainty, credulity, and outcome-
relevant involvement.”9 According to Allport and Postman’s
theory, changing environments— a black president in the
White House, uncertainty about the outcome of a strike—
generate anxiety. Anxiety makes people credulous toward infor-
mation related to the anxiogenic events. Rumors help people
make sense of current happenings, reduce uncertainty about the
&
nbsp; future, and assuage their anxiety. On top of the anxiety- inducing
148 ch ap t er 10
situation, which would make people more gullible, some indi-
viduals are supposed to naturally suffer from a lack of “critical
sense,” making them particularly good transmitters of rumors,
however ludicrous these rumors might be.10
These explanations may sound plausible, but they do not fit
with the theory put forward here. Uncertainty should make us
yearn for certainty, anxiety should make us clamor for
reassurance— but only if the certainty and the reassurance are
real. Being lulled into a false sense of certainty or security might
feel nice, but it is a recipe for disaster. Open vigilance mecha-
nisms should reject messages that we don’t have good enough
grounds to accept, irrespective of how they make us feel.
The main issue with the theory that people credulously look
for anxiety- assuaging rumors is that most rumors are more likely
to fuel than to extinguish anxiety.11 Do we feel safer thinking the
local shop keep ers are kidnapping young girls? Do exaggerated
claims of disaster- related damages assuage our concerns?
But even if the standard theories can’t explain the full pattern
of rumor transmission, the transmission of so many false rumors
might still challenge my argument that people are not gullible, but
are good at evaluating communication. Whether false rumors are
anxiety reducing or anxiety increasing, many people accept them,
often on the basis of flimsy evidence. This seems like a glaring
failure for our mechanisms of open vigilance. But to properly
gauge this failure, and to better understand its causes, we must
take a look at more efficient cases of rumor transmission.
All That Spreads Is Not False
For many years, the Wall Street Journal has published a daily col-
umn, “Heard on the Street,” which rec ords gossip and rumors
flying around the world of finance. In an analy sis of this column,
t i t il l at in g r um o r s 149
economists John Pound and Richard Zeckhauser focused on ru-
mors of takeover attempts— when a com pany makes a bid to
acquire another com pany.12 They found that nearly half of these
rumors were accurate, making them a valuable source of infor-
mation, which markets appropriately took into account.13