The Journal of Political Philosophy: Volume 17, Number 2, 2009, pp.

202227

Symposium on Conspiracy Theories

Conspiracy Theories: Causes and Cures*
Cass R. Sunstein
Law, Harvard University

and

Adrian Vermeule
Law, Harvard University

HE truth is out there:1 conspiracy theories are all around us. In August
2004, a poll by Zogby International showed that 49 percent of New York
City residents, with a margin of error of 3.5 percent, believed that officials
of the U.S. government knew in advance that attacks were planned on or
around September 11, 2001, and that they consciously failed to act.2 In a
Scripps-Howard Poll in 2006, some 36 percent of respondents assented to the
claim that federal officials either participated in the attacks on the World Trade
Center or took no action to stop them.3 Sixteen percent said that it was either
very likely or somewhat likely that the collapse of the twin towers in New York
was aided by explosives secretly planted in the two buildings.4
Conspiracy theories can easily be found all over the world. Among
sober-minded Canadians, a September 2006 poll found that 22 percent believed
that the attacks on the United States on September 11, 2001 had nothing to do
with Osama Bin Laden and were actually a plot by influential Americans.5 In a
poll conducted in seven Muslim countries, 78 percent of respondents said that
they do not believe the 9/11 attacks were carried out by Arabs.6 The most popular

*Thanks to Gabriella Blum, Mark Fenster, Don Herzog, Orin Kerr, Eric Posner, Andrei Shleifer,
Mark Tushnet, and this journals referees for valuable comments, and to Joel Peters-Fransen and
Elisabeth Theodore for excellent research assistance.
1
This slogan was popularized by the television show The X-Files, http://en.
wikipedia.org/wiki/The_X-Files. 9/11 conspiracy theorists often call themselves the 9/11 Truth
Movement; see http://www.911truth.org.
2
Zogby International, Half of New Yorkers believe US leaders had foreknowledge of impending
9-11 attacks and consciously failed to act, http://zogby.com/news/ReadNews.dbm?ID=855,
posted Aug. 30, 2004.
3
Thomas Hargrove and Guido H. Stempel III, A third of U.S. public believes 9/11 conspiracy
theory, Scripps Howard News Service, http://www.shns.com/shns/g_index2.cfm?action=
detail&pk=CONSPIRACY, posted Aug. 2, 2006.
4
Ibid.
5
One in 5 Canadians sees 9/11 as U.S. plot poll, Reuters, Sept. 11, 2006.
6
Matthew A. Gentzkow and Jesse M. Shapiro, Media, education and anti-Americanism in the
Muslim world, Journal of Economic Perspectives, 18 (2004), 117133.
2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd., 9600 Garsington Road,
Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.
doi: 10.1111/j.1467-9760.2008.00325.x

CONSPIRACY THEORIES

203

account, in these countries, is that 9/11 was the work of the U.S. or Israeli
governments.7 In China, a bestseller attributes various events (the rise of Hitler,
the Asian financial crisis of 19971998, and environmental destruction in the
developing world) to the Rothschild banking dynasty; the analysis has been read
and debated at high levels of business and government, and it appears to have
had an effect on discussions about currency policies.8 Throughout American
history, race-related violence has often been spurred by false rumors, generally
pointing to alleged conspiracies by one or another group.9
What causes such theories to arise and spread? Are they important and
perhaps even threatening, or merely trivial and even amusing? What can and
should government do about them? We aim here to sketch some psychological
and social mechanisms that produce, sustain, and spread these theories; to show
that some of them are quite important and should be taken seriously; and to offer
suggestions for governmental responses, both as a matter of policy and as a
matter of law.
Most of the academic literature directly involving conspiracy theories falls into
one of two classes: (1) work by analytic philosophers, especially in epistemology
and the philosophy of science, that explores a range of issues but mainly
asks what counts as a conspiracy theory and whether such theories are
methodologically suspect;10 (2) a smattering of work in sociology and Freudian
psychology on the causes of conspiracy theorizing.11 We offer some remarks on
the conceptual debates here, but we will generally proceed in pragmatic fashion
and mostly from the ground up, hewing close to real examples and the policy
problems they pose. To illuminate issues of policy, we draw upon literatures in
social psychology, economics, and other disciplines concerning informational
cascades, the spread of rumors, and the epistemology of groups and social
networks. We adapt the insights of these literatures by focusing on the features of
7

Ibid. at p. 120.
See Richard McGregor, Chinese buy into conspiracy theory, Financial Times, Sept. 25, 2007,
http://www.redicecreations.com/article.php?id=1907.
9
See Terry Ann Knopf, Rumors, Race and Riots, 2nd edn (New Brunswick, NJ: Transaction
Publishers, 2006).
10
See, e.g., David Coady, ed., Conspiracy Theories: The Philosophical Debate (Aldershot,
Hampshire, England: Ashgate, 2006) and Carl F. Graumann and Serge Moscovici, eds, Changing
Conceptions of Conspiracy (New York: Springer-Verlag, 1987).
11
There is also a body of work that collects many interesting examples of conspiracy theories, but
without any sustained analytic approach. See, e.g.: Michael Barkun, A Culture of Conspiracy
(Berkeley: University of California Press, 2003) and Daniel Pipes, Conspiracy (New York: Free Press,
1997). For a treatment of conspiracy theories from the standpoint of cultural studies, see Mark
Fenster, Conspiracy Theories (Minneapolis: University of Minnesota Press, 1999). A great deal of
literature exists on rumors; conspiracy theories can proliferate through rumors, although they need
not do so (consider the conspiracy theories introduced through bestselling books, described above).
For the classic treatment of rumor, see Gordon Allport and Leo Postman, The Psychology of Rumor
(New York: H. Holt and Company, 1947). Valuable recent discussions include: Nicholas Difonzo and
Prashant Bordia, Rumor Psychology: Social and Organizational Approaches (Washington, DC:
American Psychological Association, 2006); and Chip Heath and Veronique Campion-Vincent, eds,
Rumor Mills: The Social Impact of Rumor and Legend (New Brunswick, NJ: Aldine Transaction,
2005). On rumors, conspiracies, and racial violence, see Knopf, Rumors, Race and Riots.
8

204

CASS R. SUNSTEIN AND ADRIAN VERMEULE

false and harmful conspiracy theories that make them distinct from, and
sometimes more damaging than, other false and harmful beliefs.
Our running example involves conspiracy theories relating to terrorism,
especially theories that arose from and post-date the 9/11 attacks.
Terrorism-related theories are hardly the only ones of interest, but they provide
a crucial testing ground for the significance, causes, and policy implications
of widespread conspiracy theorizing. As we shall see, an understanding of
conspiracy theories illuminates the spread of information and beliefs more
generally. We shall also see, however, that because of their special characteristics,
conspiracy theories pose unique challenges.
Section I explores some definitional issues and lays out some of the
mechanisms that produce conspiracy theories and theorists. We begin by
narrowing our focus to conspiracy theories that are false, harmful, and
unjustified (in the epistemological sense), and by discussing different
understandings of the nature of such conspiracy theories and different accounts
of the kinds of errors made by those who hold them. Our primary claim is that
those who hold conspiracy theories of this distinctive sort typically do so not as
a result of a mental illness of any kind, or of simple irrationality, but as a result
of a crippled epistemology, in the form of a sharply limited number of
(relevant) informational sources. In that sense, acceptance of such theories may
not be irrational or unjustified from the standpoint of those who adhere to them
within epistemologically isolated groups or networks,12 although they are
unjustified relative to the information available in the wider society, especially if
it is an open one. There is a close connection, we suggest, between our claim on
this count and the empirical association between terrorist behavior and an
absence of civil rights and civil liberties.13 When civil rights and civil liberties are
absent, people lack multiple information sources, and they are more likely to
have reason to accept conspiracy theories.
Section II discusses government responses and legal issues. We address several
dilemmas of governmental response to false, harmful, and unjustified conspiracy
theories. Conspiracy theories turn out to be unusually hard to undermine or
dislodge; they have a self-sealing quality, rendering them particularly immune to
challenge. Our principal claim here involves the potential value of cognitive
infiltration of extremist groups, designed to introduce informational diversity
into such groups and to expose indefensible conspiracy theories as such.
12
For a valuable and analogous account of fundamentalist beliefs, see Michael Baurmann,
Rational fundamentalism? An explanatory model of fundamentalist beliefs, Episteme, 4 (2007),
15066.
13
See Alan Krueger, What Makes a Terrorist? (Princeton: Princeton University Press, 2007), pp.
7582. Krueger (p. 148) believes that low civil liberties cause terrorism, but acknowledges that his
data are also consistent with the hypothesis that terrorism causes governments to reduce civil liberties.
Of course, the two effects may both occur, in a mutually reinforcing pattern. Following Krueger, we
assume that low civil liberties tend to produce terrorism, a hypothesis that is supported by the
mechanisms we adduce.

CONSPIRACY THEORIES

205

I. DEFINITIONS AND MECHANISMS

A. DEFINITIONAL NOTES
There has been much discussion of what, exactly, counts as a conspiracy theory,
and about what, if anything, is wrong with those who hold one. Of course it
would be valuable to specify necessary and sufficient conditions for such theories,
in a way that would make it possible to make relevant distinctions. However, the
various views that people label conspiracy theories may well relate to each
other through a family-resemblance structure, such that necessary and sufficient
conditions cannot be given even in principle.
We bracket the most difficult conceptual questions here and suggest,
pragmatically, that a conspiracy theory can generally be counted as such if it is
an effort to explain some event or practice by reference to the machinations of
powerful people, who attempt to conceal their role (at least until their aims are
accomplished). While many conspiracy theories involve people who are not
especially powerful (friends, neighborhoods, fellow employees, family members,
and so forth), this account is the most useful for our particular purposes, and it
seems to capture the essence of the most prominent and influential conspiracy
theories about public affairs. Consider, for example, the view that the Central
Intelligence Agency was responsible for the assassination of President John F.
Kennedy; that doctors deliberately manufactured the AIDS virus; that the 1996
crash of TWA flight 800 was caused by a U.S. military missile; that the theory
of global warming is a deliberate fraud; that the Trilateral Commission is
responsible for important movements of the international economy; that Martin
Luther King, Jr. was killed by federal agents; that the plane crash that killed
Democrat Paul Wellstone was engineered by Republican politicians; that the
moon landing was staged and never actually occurred; that the Rothschilds and
other Jewish bankers are responsible for the deaths of presidents and for
economic distress in Asian nations; and that the Great Depression was a result of
a plot by wealthy people to reduce the wages of workers.14
14
See Mark Lane, Plausible Denial: Was the CIA Involved in the Assassination of JFK? (New York:
Thunders Mouth Press, 1991) (arguing that it was); Alan Cantwell, AIDS and the Doctors of Death:
An Inquiry into the Origins of the AIDS Epidemic (Los Angeles: Aries Rising Press, 1988) (suggesting
AIDS was the product of a biowarfare program targeting gay people); Don Phillips, Missile theory
haunts TWA investigation; despite lack of evidence and officials denials, some insist friendly fire
caused crash, Washington Post, Mar. 14, 1997, p. A03; Statement of Sen. Inhofe, Congressional
Record, 149, S10022 (daily ed. July 28, 2003) (With all the hysteria, all the fear, all the phony
science, could it be that manmade global warming is the greatest hoax ever perpetrated on the
American people? I believe it is.); David Mills, Beware the Trilateral Commission! The influential
world panel conspiracy theorists love to hate, Washington Post, Apr. 25, 1992, p. H1 (describing
various conspiracy theories about the Commission); William F. Pepper, An Act of State: The
Execution of Martin Luther King (New York: Verso, 2003) (arguing that the military, the CIA, and
others within the government conspired to kill King); Kevin Diaz, Findings dont slow conspiracy
theories on Wellstone crash; an official investigation has focused on pilot error and weather. Some
observers still have suggested a political plot, Star Tribune (Minn.), June 3, 2003, p. A1; Patty
Reinert, Apollo shrugged: hoax theories about moon landings persist, Houston Chronicle, Nov. 17,

206

CASS R. SUNSTEIN AND ADRIAN VERMEULE

Of course some conspiracy theories have turned out to be true, and under our
definition, they do not cease to be conspiracy theories for that reason.15 The
Watergate hotel room used by Democratic National Committee was, in fact,
bugged by Republican officials, operating at the behest of the White House. In the
1950s, the Central Intelligence Agency did, in fact, administer LSD and related
drugs under Project MKULTRA, in an effort to investigate the possibility of
mind control. Operation Northwoods, a rumored plan by the Department of
Defense to simulate acts of terrorism and to blame them on Cuba, really was
proposed by high-level officials (though the plan never went into effect).16 Our
focus throughout is on demonstrably false conspiracy theories, such as the
various 9/11 conspiracy theories, not ones that are true or whose truth is
undetermined. Our ultimate goal is to explore how public officials might
undermine such theories, and as a general rule, true accounts should not be
undermined.17
Within the set of false conspiracy theories, we also limit our focus to
potentially harmful theories. Consider the false conspiracy theory, held by many
of the younger members of our society, that the mysterious Santa Claus
distributes presents around the world on Christmas Eve. This theory turns out
to be false, but is itself instilled through a widespread conspiracy of the
powerfulparentswho conceal their role in the whole affair. It is an open
question whether most conspiracy theories are equally benign; we will suggest
that some are not benign at all.
Under this account, conspiracy theories are a subset of the larger category of false
beliefs, and also of the somewhat smaller category of beliefs that are both false and
harmful. Consider, for example, the beliefs that prolonged exposure to sunlight is
actually healthy, that cigarette smoking does not cause cancer, and that climate
change is neither occurring nor likely to occur. These beliefs are (in our view) both
false and dangerous, but as stated, they need not depend on, or posit, any kind of
conspiracy theory. We shall see that the mechanisms that account for conspiracy
theories overlap with those that account for false and dangerous beliefs of all sorts,
including those that fuel anger and hatred.18 But as we shall also see, conspiracy
2002, p. A1. On the Rothschilds, see McGregor, Chinese buy into conspiracy theory; on the Great
Depression, see Allport and Postman, The Psychology of Rumor, p. 517.
15
For the point that some conspiracy theories turn out to be true, and several attempts to explore
the philosophical implications of that fact, see Charles Pigden, Conspiracy theories and the
conventional wisdom, Episteme, 4 (2007), 219232 and Charles Pidgen, Complots of mischief,
Conspiracy Theories, ed. Coady, 13966.
16
See Bob Woodward and Carl Bernstein, All the Presidents Men (New York: Simon and Schuster,
1974); George Lardner, Jr., and John Jacobs, Lengthy mind-control research by CIA is detailed,
Washington Post, Aug. 3, 1977, p. A1; Memorandum from L.L. Lemnitzer, Chairman, Joint Chiefs
of Staff, to the Secretary of Defense, Justification for U.S. Military Intervention in Cuba,
http://www.gwu.edu/~nsarchiv/news/20010430/northwoods.pdf, posted Mar. 13, 1962.
17
We bracket the interesting question whether, on consequentialist grounds, it is ever appropriate
to undermine true conspiracy theories.
18
See Edward Glaeser, The political economy of hatred, Quarterly Journal of Economics, 120
(2005), 4586.

CONSPIRACY THEORIES

207

theories have distinctive features, above all because of their self-sealing quality;
the very arguments that give rise to them, and account for their plausibility, make
it more difficult for outsiders to rebut or even to question them.
Conspiracy theories often attribute extraordinary powers to certain agentsto
plan, to control others, to maintain secrets, and so forth. Those who believe that
those agents have such powers are especially unlikely to give respectful attention
to debunkers, who may, after all, be agents or dupes of those who are responsible
for the conspiracy in the first instance. It is comparatively easier for government
to dispel false and dangerous beliefs that rest, not on a self-sealing conspiracy
theory, but on simple misinformation or on an apparent or actual social
consensus that is fragile and easily tipped.19 The most direct governmental
technique for dispelling false (and also harmful) beliefsproviding credible
public informationdoes not work, in any straightforward way, for conspiracy
theories. This extra resistance to correction through simple techniques is what
makes conspiracy theories distinctively worrisome.
A further question about conspiracy theorieswhether true or false, harmful
or benignis whether they are justified. Justification and truth are different
issues, which is why pointing out that some conspiracy theories are true does not
show that it is rational to believe in those theories. A true belief may be
unjustified, and a justified belief may be untrue. I may believe, correctly, that
there are fires within the earths core, but if I believe that because the god Vulcan
revealed it to me in a dream, my belief is unwarranted. Conversely, the false belief
in Santa Claus is justified, because children generally have good reason to believe
what their parents tell them and follow a sensible heuristic (if my parents say it,
it is probably true); when children realize that Santa is the product of a
widespread conspiracy among parents, they have a justified and true belief that a
conspiracy has been at work.
Our final narrowing condition is that we are concerned only with (the many)
conspiracy theories that are false, harmful, and unjustified (not in the sense of
being irrationally held by those individuals who hold them, but from the
standpoint of the information available in the society as a whole). When and
under what conditions are conspiracy theories unjustified? Here there are
competing accounts and many controversies, in epistemology and analytic
philosophy.20 We need not opt for only one of these accounts, because they are
not mutually exclusive; each accounts for part of the terrain.
19
On the fragility of many cases of apparent social consensus, see Timur Kuran, Private Truths,
Public Lies (Cambridge, Mass: Harvard University Press, 1995). On the fragility of many cases of
actual social consensus, see David Hirshleifer, The blind leading the blind: social influence, fads, and
informational cascades, The New Economics of Human Behavior, ed. Mariano Tommasi and
Kathryn Ierulli (Cambridge: Cambridge University Press, 1995), pp. 188215.
20
For some of the latest philosophical treatments, compare David Coady, Are conspiracy
theorists irrational? Episteme, 4 (2007), 193204, with Neil Levy, Radically socialized knowledge
and conspiracy theories, Episteme, 4 (2007), 18192 and Pete Mandik, Shit happens, Episteme,
4 (2007), 20618. Roughly speaking, Coady denies that conspiracy theories are generally unjustified
and (thus) irrational, while Levy and Mandik affirm that they are.

208

CASS R. SUNSTEIN AND ADRIAN VERMEULE

Karl Popper famously argued that conspiracy theories overlook the pervasive
unintended consequences of political and social action; they assume that all
consequences must have been intended by someone.21 Many social effects,
including large movements in the economy, occur as a result of the acts and
omissions of many people, none of whom intended to cause those effects. The
appeal of some conspiracy theories, then, lies in the attribution of otherwise
inexplicable events to intentional action,22 and to an unwillingness to accept the
possibility that significant adverse consequences may be a product of invisible
hand mechanisms (such as market forces or evolutionary pressures) or of simple
chance,23 rather than of anyones plans.24
Popper captures an important feature of some conspiracy theories. There is a
pervasive human tendency to think that effects are caused by intentional action,
especially by those who stand to benefit (the cui bono? maxim), and for this
reason conspiracy theories have considerable but unwarranted appeal.25 On one
reading of Poppers account, those who accept conspiracy theories are following
a sensible heuristic, to the effect that consequences are intended; that heuristic
often works well, but it also produces systematic errors, especially in the context
of outcomes that are a product of complex interactions among numerous people.
More broadly, Popper is picking up on a general fact about human psychology,
which is that most people do not like to believe that significant events were
caused by bad (or good) luck, and much prefer simpler causal stories.26 In
particular, human minds protest against chaos, and people seek to extract a
meaning from a bewildering event or situation,27 a meaning that a conspiracy
may well supply.
Note, however, that the domain of Poppers explanation is quite limited. Many
conspiracy theories, including those involving political assassinations and the
attacks of 9/11, point to events that are indeed the result of intentional action,
and the conspiracy theorists go wrong not by positing intentional actors, but by
misidentifying them. (The theory that Al-Qaeda was responsible for 9/11 is thus
a justified and true conspiracy theory.)
Conspiracy theories that posit machinations by government officials typically
overestimate the competence and discretion of officials and bureaucracies, who
are assumed to be able to make and carry out sophisticated secret plans, despite
abundant evidence that in open societies government action does not usually
21
See Karl R. Popper, The conspiracy theory of society, Conspiracy Theories, ed. Coady, 1316;
see also Popper, The Open Society and Its Enemies, 5th edn (London: Routledge & K. Paul, 1966),
vol. 2.
22
See generally Mandik, Shit happens.
23
See Nassim Taleb, Fooled by Randomness (New York: Texere, 2001).
24
An illuminating discussion is Edna Ullmann-Margalit, The invisible hand and the cunning of
reason, Social Research, 64 (1997), 18198. We note that Poppers account has been criticized in
many places. See, for example, Pigden, Conspiracy theories and the conventional wisdom.
25
Ullmann-Margalit, The invisible hand and the cunning of reason.
26
See Taleb, Fooled by Randomness.
27
See Allport and Postman, Psychology of Rumor, p. 503.

CONSPIRACY THEORIES

209

remain secret for very long.28 Consider all the work that must be done to hide and
to cover up the governments role in producing a terrorist attack on its own
territory, or in arranging to kill political opponents.
In a closed society, secrets are far easier to keep, and distrust of official
accounts makes a great deal of sense. In such societies, conspiracy theories are
both more likely to be true and harder to show to be false in light of available
information. But when the press is free, and when checks and balances are in
force, it is harder for government to keep nefarious conspiracies hidden for long.
These points do not mean that it is logically impossible, even in free societies,
that conspiracy theories are true; sometimes they are. But it does mean that
institutional checks make it less likely, in such societies, that powerful groups can
keep dark secrets for extended periods, at least if those secrets involve illegal or
nefarious conduct. Of course conspiracy theories are widespread even in open
societies, including the United States, the United Kingdom, and France; the only
point is that such theories are less likely to be either true or justified in such
societies.
An especially useful account suggests that what makes unjustified conspiracy
theories unjustified is that those who accept them must also accept a kind
of spreading distrust of all knowledge-producing institutions, in a way that
makes it difficult to believe anything at all.29 To think, for example, that
U.S. government officials destroyed the World Trade Center and then covered
their tracks requires an ever-widening conspiracy theory, in which the 9/11
Commission, congressional leaders, the FBI, and the media were either
participants in or, at best, dupes of the conspiracy. But anyone who believed that
would undercut the grounds for many of their other beliefs, which are warranted
only by trust in the knowledge-producing institutions created by government and
society. As Robert Anton Wilson notes of the conspiracy theories advanced by
Holocaust deniers, a conspiracy that can deceive us about 6,000,000 deaths can
deceive us about anything, and [then] it takes a great leap of faith for Holocaust
Revisionists to believe World War II happened at all, or that Franklin Roosevelt
did serve as President from 1933 to 1945, or that Marilyn Monroe was more
real than King Kong or Donald Duck.30
This is not, and is not be intended to be, a general claim that conspiracy
theories are unjustified or unwarranted in all imaginable situations or societies.
Much depends on the background state of knowledge-producing institutions. If
those institutions are generally trustworthy, in part because they are embedded in
28
See, e.g., James Risen and Eric Lichtblau, Bush lets U.S. spy on callers without courts, New
York Times, Dec. 16, 2005, p. A1; Jane Mayer, The black sites: a rare look inside the C.I.A.s secret
interrogation program, New Yorker, Aug. 13, 2007, p. 46.
29
Brian L. Keeley, Of conspiracy theories, Conspiracy Theories, ed. Coady, pp. 4560 at pp. 46,
567. Keeleys argument has been the subject of much debate and controversy. For references, and a
nuanced defense-cum-critique of Keeleys theory, see Juha Rikk, On political conspiracy theories,
Journal of Political Philosophy, this issue.
30
Quoted in Keeley, Of conspiracy theories, p. 57.

210

CASS R. SUNSTEIN AND ADRIAN VERMEULE

an open society with a well-functioning marketplace of ideas and free flow of

information, and if it is difficult to dupe many diverse institutions simultaneously
(as the 9/11 conspiracy theories require), then conspiracy theories will usually
be unjustified. On the other hand, individuals in societies with systematically
malfunctioning or skewed institutions of knowledgesay, individuals who live in
an authoritarian regime lacking a free pressmay have good reason to distrust
all or most of the official denials they hear. For these individuals, conspiracy
theories will more often be warranted, whether or not true.
Likewise, individuals embedded in isolated groups or small, self-enclosed
networks who are exposed only to skewed information will more often hold
conspiracy theories that are justified, relative to their limited informational
environment.31 Holocaust denials might themselves be considered in this light.
When epistemologically isolated groups operate within a society that is both
wider and more open, their theories may be unjustified from the standpoint of the
wider society but justified from the standpoint of the individual or group. In these
situations, the problem for the wider society is to breach the informational
isolation of the small group or network (a problem we discuss below).
On our account, a central feature of conspiracy theories is that they are
extremely resistant to correction, certainly through direct denials or
counterspeech by government officials; apparently contrary evidence can usually
be shown to be a product of the conspiracy itself. Conspiracy theories often
display the characteristic features of a degenerating research program32 in
which contrary evidence is explained away by adding epicycles and resisting
falsification of key tenets.33 Some epistemologists argue that this resistance to
falsification is not objectionable if one also believes that there are conspirators
deliberately attempting to plant evidence that would falsify the conspiracy theory,
thereby covering their tracks.34 However that may be as a philosophical matter,
the self-sealing quality of conspiracy theories creates severe practical problems
for government; direct attempts to dispel the theory can usually be folded into the
theory itself, as just one more ploy by powerful conspiracy members.
So far we have discussed some epistemological features of conspiracy theories,
in the abstract; narrowed our focus to conspiracy theories that are false, harmful,
and unjustified from the standpoint of the wider society (although they may be
justified from the standpoint of individuals, given the information they have
received, or within the closed epistemological network of the conspiracy
theorists); and suggested institutional grounds for thinking that in a free and
31

Compare Baurman, Rational fundamentalism?

Imre Lakatos, Falsification and methodology of scientific research programmes, Criticism and
the Growth of Knowledge, ed. Imre Lakatos and Alan Musgrave (Cambridge: Cambridge University
Press, 1970), pp. 91196. Steve Clarke, Conspiracy theories and conspiracy theorizing, Conspiracy
Theories, ed. Coady, pp. 7792 at p. 78.
33
See Diana G. Tumminia, When Prophecy Never Fails: Myth and Reality in a Flying-Saucer
Group (Oxford: Oxford University Press, 2005).
34
Keeley, Of conspiracy theories, pp. 556.
32

CONSPIRACY THEORIES

211

open society, there is usually good reason to believe that most conspiracy theories
will lack adequate justification. We now turn to the sociology of conspiracy
theorizing, examining the mechanisms by which such theories arise and
expand.

B. HOW CONSPIRACY THEORIES ARISE AND SPREAD

i. Crippled Epistemologies
Why do people accept conspiracy theories that turn out to be false and for which
the evidence is weak or even nonexistent? It is tempting to answer in terms of
individual pathology, either literal or metaphorical.35 Perhaps conspiracy theories
are a product of mental illness, such as paranoia or narcissism, or of similar
conditions. And surely some people who accept conspiracy theories are mentally
ill and subject to delusions.36 But we have seen that in many communities and
even nations, such theories are widely held. It is not plausible to suggest that all
or most members of those communities are afflicted by mental illness; the
metaphor of mental illness itself obscures more than it clarifies. The most
important conspiracy theories are hardly limited to those who suffer from any
kind of pathology.
For our purposes, the most useful way to understand the pervasiveness of
conspiracy theories is to examine how people acquire their beliefs.37 For most of
what they believe that they know, human beings lack personal or direct
information; they must rely on what other people think. In some domains, people
suffer from a crippled epistemology, in the sense that they know very few
things, and what they know is wrong.38 Many extremists fall in this category;
their extremism stems not from irrationality, but from the fact that they have
little (relevant) information, and their extremist views are supported by what

35
See Richard Hofstadter, The paranoid style in American politics, The Paranoid Style in
American Politics and Other Essays (Chicago: University of Chicago Press, 1979); Robert S. Robins
and Jerold M. Post, Political Paranoia (New Haven: Yale University Press, 1997). Another common
idea treats conspiracy theories as a form of collective paranoid delusion. See, e.g., Deiter Groh, The
temptation of conspiracy theory, Changing Conceptions of Conspiracy, ed. Graumann and
Moscovici, pp. 138. Our suggestion is that the lens of psychopathology is not helpful, whether it is
interpreted in individual or collective terms.
36
See Erich Wulff, Paranoic conspiratory delusion, Changing Conceptions of Conspiracy, ed.
Graumann and Moscovici, pp. 17190.
37
There is an immense and growing literature on this question, not exploring conspiracy theories,
but with obvious relevance to them. For examples, with relevant citations, see Sendhil Mullainathan
and Andrei Shleifer, The market for news, The American Economic Review, 95 (2005), 103154;
Edward Glaeser and Cass R. Sunstein, Extremism and social learning, Journal of Legal Analysis,
forthcoming 2008.
38
Russell Hardin, The crippled epistemology of extremism, Political Extremism and
Rationality, ed. Albert Breton, Gianluigi Galeotti, Pierre Salmon, and Ronald Wintrobe (Cambridge:
Cambridge University Press, 2002), pp. 322 at pp. 16 ff. Of course we do not deny that some
extremism is justified and that the beliefs that underlie extremism may be true.

212

CASS R. SUNSTEIN AND ADRIAN VERMEULE

little they know.39 Conspiracy theorizing often has the same feature. Those who
believe that Israel was responsible for the attacks of 9/11, or that the Central
Intelligence Agency killed President Kennedy, may well be responding quite
rationally to the informational signals that they receive; in this sense, those beliefs
may well be justified from the standpoint of the individuals who hold them, even
if they are preposterous in light of the information available in the wider society.
Consider here the suggestive claim that terrorism is more likely to arise in
nations that lack civil rights and civil liberties.40 If this is so, it might be because
terrorism is not abstract violence but an extreme form of political protest, and
when people lack the usual outlets for registering their protest, they might resort
to violence.41 But consider another possibility: when civil rights and civil liberties
are restricted, little information is available, and what comes from government
cannot be trusted. If the most trustworthy or least untrustworthy information
justifies conspiracy theories and (therefore) extremism, and (therefore?) violence,
then terrorism is more likely to arise.
ii. Rumors and Speculation
Of course it is necessary to specify how, exactly, conspiracy theories begin. Some
such theories seem to bubble up spontaneously, appearing roughly
simultaneously in many different social networks; others are initiated and spread,
quite intentionally, by conspiracy entrepreneurs who profit directly or indirectly
from propagating their theories. One example in the latter category is the author
of the Chinese bestseller mentioned above;42 another is the French author Thierry
Meyssan, whose book 9/11: The Big Lie became a bestseller and a sensation for
its claims that the Pentagon explosion on 9/11 was caused by a missile, fired as
the opening salvo of a coup detat by the military-industrial complex, rather than
by American Airlines Flight 77.43
Some conspiracy entrepreneurs are entirely sincere. Others are interested in
money or power or in using the conspiracy theory to achieve some general social
goal. In the context of the AIDS virus, for example, a diverse set of people
initiated rumors, many involving conspiracies, and in view of the confusion and
fear surrounding that virus, several of those rumors spread widely.44 But even for

39
Ibid. See also Baurmann, Rational fundamentalism? It is also true that many extremists have
become extreme, or stayed extreme, after being exposed to a great deal of information on various
sides. Their refusal to change their views may or may not be justified, depending on the question and
the relevant information.
40
See Krueger, What Makes a Terrorist? pp. 7582.
41
Ibid., 8990.
42
See McGregor, Chinese buy into conspiracy theory. Consider the authors astonishing
statement: This book may be totally wrong, so before I write the next one, I have to make sure my
understanding is right.
43
See also James Fetzer, The 9/11 Conspiracy (Peru, Ill.: Catfeet Press, 2007) and Mathias
Broeckers, Conspiracies, Conspiracy Theories and the Secrets of 9/11 (Joshua Tree, Calif.: Progressive
Press, 2006). The latter book sold over 100,000 copies in Germany.
44
See Diane Goldstein, Once upon a Virus (Logan, Utah: Utah State University Press, 2004).

CONSPIRACY THEORIES

213

conspiracy theories put about by conspiracy entrepreneurs, the key question is

why some theories take hold while many more do not, and vanish into obscurity.
Whenever a bad event has occurred, rumors and speculation are inevitable.
Most people are not able to know, on the basis of personal or direct knowledge,
why an airplane crashed, or why a leader was assassinated, or why a terrorist
attack succeeded, or why many people stayed in an area despite what turned out
to be an imminent natural disaster. In the aftermath of such an event, numerous
speculations will be offered, and some of them will likely point to some kind of
conspiracy. To some people, those speculations will seem plausible, perhaps
because they provide a suitable outlet for outrage and blame, perhaps because the
speculation fits well with other deeply rooted beliefs that they hold. Terrible
events produce outrage, and when people are outraged, they are all the more
likely to seek causes that justify their emotional states, and also to attribute
those events to intentional action.45 Conspiracy theories, like rumors, may
simultaneously relieve a primary emotional urge and offer an explanation, to
those who accept the theory, of why they feel as they do; the theory rationalizes
while it relieves.46
In addition, antecedent beliefs are a key to the success or failure of conspiracy
theories. Within the United States, some people would find it impossibly jarring
to think that the Central Intelligence Agency (CIA) was responsible for the
assassination of a civil rights leader; that thought would unsettle too many of
their other judgments. Others would find those other judgments strongly
supported, even confirmed, by the suggestion that the CIA was responsible for
such an assassination. Compare the case of terrorist attacks. For most Americans,
a claim that the United States government attacked its own citizens, or condoned
such attacks, would make it impossible to hold onto a wide range of other
judgments. Clearly this point does not hold for many people in Islamic nations,
for whom it is far from jarring to believe that responsibility lies with the United
States (or Israel).
iii. Conspiracy Cascades: The Role of Information
To see how informational cascades work, imagine a group of people who are
trying to assign responsibility for some loss of life. Assume that the group
members are announcing their views in sequence. Each member attends,
reasonably enough, to the judgments of others. Andrews is the first to speak. He
suggests that the event was caused by a conspiracy of powerful people. Barnes
now knows Andrewss judgment; she should certainly go along with Andrews
account if she agrees independently with him. But if her independent judgment is
otherwise, she wouldif she trusts Andrews no more and no less than she trusts
herselfbe indifferent about what to do, and she might simply flip a coin.
45
46

See Leon Festinger, A Theory of Cognitive Dissonance (Evanston, Ill.: Row, Peterson, 1957).
Allport and Postman, Psychology of Rumor, p. 503.

214

CASS R. SUNSTEIN AND ADRIAN VERMEULE

Now turn to a third person, Charleton. Suppose that both Andrews and
Barnes have endorsed the conspiracy theory, but that Charletons own view,
based on limited information, suggests that they are probably wrong. In that
event, Charleton might well ignore what he knows and follow Andrews and
Barnes. It is likely, after all, that both Andrews and Barnes had evidence for their
conclusion, and unless Charleton thinks that his own information is better than
theirs, he should follow their lead. If he does, Charleton is in a cascade. Of course
Charleton will resist if he has sufficient grounds to think that Andrews and
Barnes are being foolish. But if he lacks those grounds, he is likely to go along
with them. This may happen even if Andrews initially speculated in a way that
does not fit the facts. That initial speculation, in this example, can start a process
by which a number of people are led to participate in a cascade, accepting a
conspiracy theory whose factual foundations are fragile.
Of course the example is highly stylized; conspiracy cascades arise through
more complex processes, in which diverse thresholds are crucial. In a standard
pattern, the conspiracy theory is initially accepted by people with low thresholds
for its acceptance. Perhaps the theory is limited, in its acceptance, to those with
such thresholds. But sometimes the informational pressure builds, to the point
where many people, with somewhat higher thresholds, begin to accept the theory
too. And when many people hold that belief, those with even higher thresholds
may come to accept the theory, leading to widespread acceptance of falsehoods.
In theory, a conspiracy theory might be justifiably held by many even though it
is false and harmful, and even though only a few early movers suggested a strong
commitment to it. As a real-world example of a conspiracy cascade, consider the
existence of certain judgments about the origins and causes of AIDS, with some
groups believing, implausibly, that the virus was produced in government
laboratories.47 These and other views about AIDS are a product of social
interactions, and in particular of cascade effects.
iv. Conspiracy Cascades: The Role of Reputation
Conspiracy theories do not take hold only because of information. Sometimes
people profess belief in a conspiracy theory, or at least suppress their doubts,
because they seek to curry favor.48 Reputational pressures help account for
conspiracy theories, and they feed conspiracy cascades. In a reputational cascade,
people think that they know what is right, or what is likely to be right, but they
nonetheless go along with the crowd in order to maintain the good opinion of
others.
Suppose that Albert suggests that the Central Intelligence Agency was
responsible for the assassination of President Kennedy, and that Barbara concurs
47
See Fabio Lorenzi-Cioldi and Alain Clmence, Group processes and the construction of social
representations, Group Processes, ed. Michael A. Hogg and R. Scott Tindale (Malden, Mass.:
Blackwell Publishers, 2001), pp. 31133 at pp. 31517.
48
For a vivid illustration in an analogous context, see Festinger, Theory of Cognitive Dissonance.

CONSPIRACY THEORIES

215

with Albert, not because she actually thinks that Albert is right, but because she
does not wish to seem, to Albert, to be some kind of dupe. It should be easy to
see how this process might generate a cascade. Once Albert, Barbara, and
Cynthia offer a united front on the issue, their friend David might be reluctant to
contradict them even if he believes that they are wrong. In real-world conspiracy
theories, reputational pressures often play a large role, leading people to squelch
their own doubts in order to avoid social sanctions.
v. Conspiracy Cascades: The Role of Availability
Informational and reputational cascades can occur without any particular
triggering event. But a distinctive kind of cascade arises when such an event is
highly salient or cognitively available.49 In the context of many risks, such as
those associated with terrorism, nuclear power, and abandoned hazardous waste
dumps, a particular event initiates a cascade, and it stands as a trigger or a symbol
justifying public concern, whether or not that concern is warranted.50 Often
political actors, both self-interested and altruistic, work hard to produce such
cascades.
Conspiracy theories are often driven through the same mechanisms. A
particular event becomes available,51 and conspiracy theories are invoked both in
explaining it and using it as a symbol for broader social forces and large
narratives about political life, casting doubt on accepted wisdom in many
domains. Within certain nations and groups, the claim that the United States or
Israel was responsible for the attacks of 9/11 fits well within a general narrative
about who is the aggressor, and the liar, in a series of disputesand the view that
Al Qaeda was responsible raises questions about that same narrative.
vi. Conspiracy Cascades: The Role of Emotions
Thus far our account has been purely cognitive: conspiracy theories circulate in
the same way that other beliefs circulate, as people give weight to the views of
others and attend to their own reputations. But it is clear that affective factors,
and not mere information, play a large role in the circulation of rumors of all
kinds. Many rumors persist and spread because they serve to justify or to
rationalize an antecedent emotional state produced by some important event,
such as a disaster or a war.52 When people are especially angry or fearful, they are

49
In the context of race relations, rumors that amount to conspiracy theories have often had this
feature, sometimes producing violence. See Knopf, Rumors, Race and Riots.
50
See Robert Repetto, ed., Punctuated Equilibrium and the Dynamics of U.S. Environmental
Policy (New Haven: Yale University Press, 2006); Timur Kuran and Cass R. Sunstein, Availability
cascades and risk regulation, Stanford Law Review, 51 (1999), 683768.
51
See Knopf, Rumors, Race and Riots.
52
See Allport and Postman, Psychology of Rumor, pp. 503504; Festinger, Theory of Cognitive
Dissonance; Frederick Koenig, Rumor in the Market Place (Dover, Mass.: Auburn House, 1985),
p. 33.

216

CASS R. SUNSTEIN AND ADRIAN VERMEULE

more likely to focus on particular sorts of rumors and to spread them to others.
And when rumors trigger intense feelings, they are far more likely to be
circulated.
Experimental evidence strongly supports this speculation in the analogous
context of urban legends.53 When urban legendsinvolving, for example, a
decapitated motorcycle rider, a rat in a soda bottle, or cat food mislabelled as
tunaare devised so as to trigger strong emotions (such as disgust), people are
more likely to pass them along. Perhaps the most revealing of these experiments
involved actual spreading of urban legends on the Internet.54 The conclusion is
that in the marketplace of ideas, emotional selection plays a significant role,
and it helps to explain such diverse phenomena as moral panics about deviant
behavior, hysteria about child abuse, and media attention to relatively small
sources of risk such as road rage and flesh eating bacteria.55 A particular
problem involves emotional snowballingrunaway selection for emotional
content rather than for information.56
The applications to conspiracy theories should not be obscure. When a terrible
event has occurred, acceptance of such theories may justify or rationalize the
affective state produced by that event; consider conspiracy theories in response to
political assassinations.57 In addition, such theories typically involve accounts, or
rumors, that create intense emotions, such as indignation, thus producing a kind
of emotional selection that will spread beliefs from one person to another.58 Of
course evidence matters, and so long as there is some kind of process for meeting
falsehoods with truth, mistaken beliefs can be corrected. But sometimes the
conditions for correction are not present.
vii. Group Polarization
There are clear links between cascades and the well-established phenomenon of
group polarization, by which members of a deliberating group typically end up
in a more extreme position in line with their tendencies before deliberation
began.59 Group polarization has been found in hundreds of studies involving over
a dozen countries.60 Belief in conspiracy theories is often fueled by group
polarization.61
Consider, as the clearest experimental example, the finding that those who
disapprove of the United States, and are suspicious of its intentions, will increase
53
See Chip Heath, Chris Bell and Emily Sternberg, Emotional selection in memes: the case of
urban legends, Journal of Personality and Social Psychology, 81 (2001), 102841.
54
Ibid., pp. 10379.
55
Ibid., p. 1039.
56
Ibid., p. 1040.
57
See Festinger, Theory of Cognitive Dissonance.
58
In the racial context, see Knopf, Rumors, Race and Riots.
59
See Roger Brown, Social Psychology, 2nd edn (New York: Free Press, 1986), pp. 20226.
60
See ibid., p. 204.
61
For a number of examples, see Jonathan Vankin, Conspiracies, Cover-Ups and Crimes (New
York: Paragon House, 1991).

CONSPIRACY THEORIES

217

their disapproval and suspicion if they exchange points of view. There is specific
evidence of this phenomenon among citizens of France: with respect to foreign
aid, they trust the United States a great deal less, and suspect its intentions a great
deal more, after they talk with one another.62 It should be easy to see how similar
effects could occur for conspiracy theories. Those who tend to think that Israel
was responsible for the attacks of 9/11, and who speak with one another, will end
up with a greater commitment to that belief.63 One result of group polarization
is that different groups may end up with radically different attitudes toward
conspiracy theories in general and in particular. Speaking with like-minded
others, some people may come to find such a theory irresistible and others may
come to find it preposterous.64
Group polarization occurs for reasons that parallel the mechanisms that
produce cascades.65 Informational influences play a large role. In any group with
some initial inclination, the views of most people in the group will inevitably be
skewed in the direction of that inclination. As a result of hearing the various
arguments, social interactions will lead people toward a more extreme point in
line with what group members initially believed. Reputational factors matter as
well. Once people hear what others believe, some will adjust their positions
at least slightly in the direction of the dominant position. Relatedly, group
polarization can occur through positional jockeying; if, for example, several
members of a group want to be the second-most-extreme supporter of the view
held in common by the group, the distribution of views within the group can shift
to become more extreme overall.
For purposes of understanding the spread of conspiracy theories, it is
especially important to note that group polarization is particularly likely, and
particularly pronounced, when people have a shared sense of identity and are
connected by bonds of solidarity.66 These are circumstances in which arguments
by outsiders, unconnected with the group, will lack much credibility, and fail to
have much of an effect in reducing polarization.
viii. Selection Effects
A crippled epistemology can arise not only from informational and reputational
dynamics within a given group, but also from self-selection of members into and
out of groups with extreme views.67 Once polarization occurs or cascades arise,
and the groups median view begins to move in a certain direction, doubters and
halfway-believers will tend to depart while intense believers remain. The overall
62

Brown, Social Psychology, pp. 2234.

See Glaeser and Sunstein, Extremism and social learning.
See the racial disparities catalogues in Knopf, Rumors, Race and Riots.
65
See ibid., pp. 21222, 22645; Robert S. Baron and Norbert L. Kerr, Group Processes, Group
Decision, Group Action, 2nd edn (Buckingham, UK: Open University Press, 2001), p. 540.
66
See Cass R. Sunstein, Why Societies Need Dissent (Cambridge, Mass.: Harvard University Press,
2003).
67
Hardin, The crippled epistemology of extremism, pp. 912.
63
64

218

CASS R. SUNSTEIN AND ADRIAN VERMEULE

size of the group may shrink, but the group may also pick up new believers who
are even more committed, and in any event the remaining members will, by
self-selection, display more fanaticism. Group members may engage in a kind of
double-think, segregating themselves, in a physical or informational sense, in
order to protect their beliefs from challenge by outsiders.68 Even if the rank and
file cannot coherently do this, group leaders may enforce segregation in order to
insulate the rank and file from information or arguments that would undermine
the leaders hold on the group. As a result, group polarization will likely intensify.
Members of informationally and socially isolated groups become increasingly
distrustful or suspicious of the motives of others or of the larger society, falling
into a sinister attribution error.69 This error occurs when people feel that they
are under pervasive scrutiny, and hence they attribute personalistic motives to
outsiders and overestimate the amount of attention they receive. Benign actions
that happen to disadvantage the group are taken as purposeful plots, intended to
harm.70 Although these conditions resemble individual-level pathologies such as
paranoid cognition, they arise from the social and informational structure of the
group, especially those operating in enclosed or closely knit networks, and are
not usefully understood as a form of mental illness, not even in a metaphorical
sense.
II. GOVERNMENTAL RESPONSES
What can the government do about conspiracy theories, and what should it do?
(1) Government might ban conspiracy theories, somehow defined. (2)
Government might impose some kind of tax, financial or otherwise, on those who
disseminate such theories. (3) Government might itself engage in counterspeech,
marshaling arguments to discredit conspiracy theories. (4) Government might
formally hire credible private parties to engage in counterspeech. (5) Government
might engage in informal communication with such parties, encouraging them to
help. Each instrument has a distinctive set of potential effects, or costs and
benefits, and each will have a place under imaginable conditions. Our main
policy claim here is that government should engage in cognitive infiltration of the
groups that produce conspiracy theories, which involves a mix of (3), (4), and (5).
The first-line response to conspiracy theories is to maintain an open society, in
which those who might be tempted to subscribe to such theories are unlikely to
distrust all knowledge-creating institutions, and are exposed to evidence and
corrections. Nongovernmental organizations, including the media, can and do
work hard to respond to such theories. As an ambitious example, consider an
Internet site, www.snopes.com, which researches rumors and conspiracy theories
68

Ibid., p.10.
Roderick M. Kramer, The sinister attribution error: paranoid cognition and collective distrust
in organizations, Motivation and Emotion, 18 (1994), 199230.
70
Ibid.
69

CONSPIRACY THEORIES

219

and reports on their truth or falsity. (Another is www.counterknowledge.com).

For those concerned about the proliferation of conspiracy theorizing on the
Internet, this site provides a reliable and helpful reality check. It would be easy to
imagine other ventures, small and large, in this vein, for the Internet provides not
only a mechanism by which to spread conspiracy theories, but also a range of
corrective tools. The more general point is that in free societies, conspiracy
theories are generally dislodged by the media and other non-governmental actors.
But we have seen that even in open societies, conspiracy theories have some
traction.71 The Internet itself has ambiguous effects; it is no panacea, even if it is
entirely free. Reduced information costs may make it easier for private monitors
to rebut conspiracy theories,72 yet by the same token that very reduction in
information costs also makes it easier for conspiracy theorists to generate and
spread their theories in the first place. The overall effect of new technology is
unclear, as is the ability of nongovernmental monitors to police conspiracy
theories. In part because this is so, an official response will sometimes be
essential, at least in important cases. And in all cases, the government must
necessarily decide whether to deny, confirm, or ignore conspiracy theories that
come to the attention of officials.
Here we suggest two concrete ideas for government officials attempting to
fashion a response to such theories. First, responding to more rather than fewer
conspiracy theories has a kind of synergy benefit: it reduces the legitimating effect
of responding to any one of them, because it dilutes the contrast with unrebutted
theories. Second, we suggest a distinctive tactic for breaking up the hard core of
extremists who supply conspiracy theories: cognitive infiltration of extremist
groups, whereby government agents or their allies (acting either virtually or in
real space, and either openly or anonymously) will undermine the crippled
epistemology of believers by planting doubts about the theories and stylized facts
that circulate within such groups, thereby introducing beneficial cognitive
diversity.
Throughout, we assume a well-motivated government that aims to eliminate
conspiracy theories, or draw their poison, if and only if social welfare is improved
by doing so. (We do not offer a particular account of social welfare, taking the
term instead as a placeholder for the right account.) This is a standard
assumption in policy analysis, although real-world governments can themselves
be purveyors of conspiracy theories, as when the Bush administration suggested
that Saddam Hussein had conspired with Al Qaeda to support the 9/11 attacks.73

71
A range of examples can be found in Vankin, Conspiracies, Cover-ups and Crimes and Knopf,
Rumors, Race and Riots.
72
As argued in Steve Clarke, Conspiracy theories and the internet: controlled demolition and
arrested development, Episteme 4 (2007), 7792.
73
See, e.g., Frank Rich, Editorial: dishonest, reprehensible, corrupt, New York Times, Nov. 27,
2005, p. 11.

220

CASS R. SUNSTEIN AND ADRIAN VERMEULE

A. ARE CONSPIRACY THEORIES CONSEQUENTIAL?

One line of thinking denies that conspiracy theories matter.74 There are several
possible reasons to think so. First, conspiracy theories may be held by only a tiny
fraction of the relevant population. Second, even if a particular conspiracy theory
is widely held in the sense that many people will confess to it when polled,
conspiracy theories may be held as quasi-beliefsbeliefs that are not costly
and possibly even fun to hold, like a belief in UFOs, and that do not form a
premise for action.75 Perhaps those who seem to accept such theories have soft
beliefs, in a way that leads them generally to keep quiet, and rarely to act on what
they tend to think.
It is true that many people do not in fact take any action on the basis of their
mistaken beliefs. But in the racial context, a belief in conspiracies has often
played a significant role in producing violence; conspiracy theories have had large
effects on behavior.76 And even if only a small fraction of adherents to a particular
conspiracy theory act on the basis of their beliefs, that small fraction may be
enough to cause serious harms. Consider the Oklahoma City bombing, whose
perpetrators shared a complex of conspiratorial beliefs about the federal
government. Many who shared their beliefs did not act on them, but a few actors
did, with terrifying consequences. James Fearon and others argue that
technological change has driven down the costs of delivering attacks with
weapons of mass destruction, to the point where even a small group can pose a
significant threat.77 If so, and if only a tiny fraction of believers act on their
beliefs, then as the total population with conspiratorial beliefs grows, it becomes
nearly inevitable that action will ensue.
In other, perhaps more common, cases the conspiracy theory will be of a
different nature and will not directly produce such action. However, such theories
can still have pernicious effects from the governments point of view, either by
inducing unjustifiably widespread public skepticism about the governments
assertions, or by dampening public mobilization and participation in
government-led efforts, or both.
B. DILEMMAS AND RESPONSES
Imagine a government facing a population in which a particular conspiracy
theory is becoming widespread. We will identify two basic dilemmas that recur,
74
See Clarke, Conspiracy theories and the internet, p. 91 (noting that few [conspiracy theories]
are actually harmful).
75
Jon Elster, Explaining Social Behavior (Cambridge: Cambridge University Press, 2007); Bryan
Caplan, The Myth of the Rational Voter (Princeton: Princeton University Press, 2007).
76
See Knopf, Rumors, Race and Riots. In the international context, conspiracy theories
undoubtedly play a role in reputational cascades. See Timur Kuran, Ethnic norms and their
transformation through reputational cascades, Journal of Legal Studies, 27 (1998), 62360.
77
James D. Fearon, Catastrophic terrorism and civil liberties, http://www.
stanford.edu/~jfearon/papers/civlibs.doc.

CONSPIRACY THEORIES

221

and consider how government should respond. The first dilemma is whether to
ignore or rebut the theory; the second is whether to address the supply side of
conspiracy theorizing by attempting to debias or disable its purveyors, to address
the demand side by attempting to immunize third-party audiences from the
theorys effects, or to do both (if resource constraints permit).
In both cases, the underlying structure of the problem is that conspiracy
theorizing is a multi-party game. Government is faced with suppliers of
conspiracy theories, and might aim at least in part to persuade, debias, or silence
those suppliers. However, those two players are competing for the hearts and
minds of third parties, especially the mass audience of the uncommitted.78
Expanding the cast further, one may see the game as involving four players:
government officials, conspiracy theorists, mass audiences, and independent
expertssuch as mainstream scientistswhom government attempts to enlist to
give credibility to its rebuttal efforts.
i. Ignore or Rebut?
The first dilemma is that either ignoring or rebutting a conspiracy theory has
distinctive risks and costs.79 Ignoring the theory allows its proponents to draw
ominous inferences from the governments silence. If the theory stands
unrebutted, people may pay less attention to it, and even when they notice it, a
natural inference from the governments silence is that the theory is too ludicrous
to need rebuttal. But another possibility is that the government is silent because
it cannot offer relevant evidence to the contrary. The suppliers of the conspiracy
theories will propose the second inference. On this view, all misinformation (the
initial conspiracy theory) should be met with countermisinformation.
On the other hand, to rebut the theory may be to legitimate it, moving the
theory from the zone of claims too ludicrous to be discussed to the zone of claims
that, whether or not true, are in some sense worth discussing. This legitimation
effect can arise in one of two ways. First of all, third-party audiences may infer
from the governments rebuttal efforts that the government itself estimates
the conspiracy theory to be plausible and fears that the third parties will be
persuaded. Here one risk is that the very act of rebuttal squarely focuses the
audience on the conspiracy theory itself, in a way that may increase its salience
and also its plausibility. Those who might reject the theory, or in any case not
think about it, may take the rebuttal as a reason to give it serious consideration.
A second possible source of the legitimation effect is that some members of the
audience may infer that many other members of the audience must believe the
conspiracy theory, or government would not be taking the trouble to rebut it.
78

For relevant discussion, see Glaeser, The Political Economy of Hatred.

For discussion of the somewhat analogous problem whether firms or governments should deny
and rebut circulating rumors, or instead ignore them, see DiFonzo and Bordia, Rumor Psychology,
ch. 9. The problems are not identical, because rumor is only one mechanism by which conspiracy
theories may arise and spread.
79

222

CASS R. SUNSTEIN AND ADRIAN VERMEULE

Consider circumstances of pluralistic ignorance, in which citizens are unsure

what other citizens believe.80 Citizens may take the fact of rebuttal itself as
supplying information about the beliefs of other citizens, and may even use this
information in forming their own beliefs. If the number who follow this cognitive
strategy and thus adopt a belief in the conspiracy theory exceeds the number who
are persuaded by the rebuttal, the perverse result of the rebuttal may then be to
increase the number of believers.
In a typical pattern, government plays a wait-and-see strategy: ignore the
conspiracy theory until it reaches some ill-defined threshold level of widespread
popularity, and then rebut. There is a straightforward logic to this strategy. When
the government ignores the theory, either the relevant audiences will draw an
inference that the theory is silly, or else will infer that the government cannot
effectively deny it. If the conspiracy theory does not spread despite the
governments silence, the former inference is probably dominant, and response is
unnecessary. There is also an option value to the strategy of ignoring the theory:
a public rebuttal now is costly or impossible to undo, but maintaining silence
now leaves government with the option to rebut later, if it chooses to do so.
Finally and most generally, it seems silly and infeasible to chase after and rebut
every conspiracy theory that comes to governments attention.
However, the concern that rebuttal will inadvertently legitimate a conspiracy
theory overlooks an important synergistic gain: rebutting many conspiracy
theories can reduce the legitimating effect of rebutting any one of them. When
government rebuts a particular theory while ignoring most others, the
legitimating effect arises at least in part because of a contrast between the
foreground and the background: the inference is that government has picked the
theory it is rebutting out of the larger set because this theory, unlike the others,
is inherently plausible or is gaining traction among some sectors of the mass
audience. The more theories government rebuts, the weaker is the implicit
legitimating signal sent by the very fact of rebuttal.
It is impossible to say, in the abstract, how great this synergistic gain may be.
However, the implication is that government should rebut more conspiracy
theories than it would otherwise choose, if assessing the expected costs and
benefits of rebuttal on a theory-by-theory basis. Because of synergy effects,
government action considered over an array or range of cases may have different
total costs and benefits than when those cases are considered one by one.
ii. Which Audience?
Should governmental responses be addressed to the suppliers, with a view to
persuading or silencing them, or rather be addressed to the mass audience, with
a view to inoculating them from pernicious theories? Of course these two
strategies are not mutually exclusive; perhaps the best approach is to straddle the
80

See Kuran, Private Truths, Public Lies.

CONSPIRACY THEORIES

223

two audiences with a single response or simply to provide multiple responses.

However, if there are resource constraints or intrinsic tradeoffs, such that the
arguments that appeal to one audience tend to alienate the other, government will
face a choice about what mix of responses is optimal
The basic problem with pitching governmental responses to the suppliers of
conspiracy theories is that, as we have noted, those theories have a self-sealing
quality. They are (1) resistant and in extreme cases invulnerable to contrary
evidence,81 and (2) especially resistant to contrary evidence offered by the
government, because the government rebuttal is folded into the conspiracy theory
itself. If conspiracy theorists are responding to the informational signals given by
those whom they trust, then the governments effort at rebuttal seems unlikely to
be effective, and might serve to fortify rather than to undermine the original
belief. After 9/11, one complex of conspiracy theories involved American Airlines
Flight 77, which hijackers crashed into the Pentagon. Even those conspiracists
who were persuaded that the Flight 77 conspiracy theories were wrong folded
that view into a larger conspiracy theory. The problem with the theory that no
plane hit the Pentagon, they said, is that the theory was too transparently false,
disproved by multiple witnesses and much physical evidence. Thus the theory
must have been a straw man initially planted by the government, in order to
discredit other conspiracy theories and theorists by association.82
Government can partially circumvent these problems if it enlists credible
independent experts in the effort to rebut the theories. There is a tradeoff between
credibility and control, however. The price of credibility is that government
cannot be seen to control the independent experts. Although government can
supply these independent experts with information and perhaps prod them into
action from behind the scenes, too close a connection will prove self-defeating
if it is exposedas witnessed in the humiliating disclosures showing that
apparently independent opinions on scientific and regulatory questions were in
fact paid for by think-tanks with ties to the Bush administration.83 Even apart
from this tradeoff, conspiracy theorists may still fold independent third-party
rebuttals into their theory by making conspiratorial claims of connection between
the third party and the government. When the magazine Popular Mechanics
offered a rebuttal of 9/11 conspiracy theories, conspiracists claimed that one of
the magazines reporters, Ben Chertoff, was the cousin of Homeland Security
Secretary Michael Chertoff and was spreading disinformation at the latters
behest.84
81

Cf. Festinger, Theory of Cognitive Dissonance.

See, e.g., Jim Hoffman, Video of the Pentagon attack: what is the government hiding,
http://911research.com/essays/pentagon/video.html.
83
See Ian Sample, Scientists offered cash to dispute climate study, Guardian, Feb. 2, 2007, p. 1
(noting that a thinktank with close links to the Bush administration had paid scientists to challenge
a report on global warming).
84
In fact, the two may be distant relatives, but had never met. Will Sullivan, Viewing 9/11 from
a grassy knoll, U.S. News & World Report, Sept. 11, 2006.
82

224

CASS R. SUNSTEIN AND ADRIAN VERMEULE

Because of these difficulties, many officials dismiss direct responses to the

suppliers of conspiracy theorists as an exercise in futility. Those with strong
commitments often engage in biased assimilation of evidence,85 and conspiracy
theorists are likely to be especially biased assimilators. Thus officials address their
responses to the third-party mass audience, hoping to stem the spread of
conspiracy theories by dampening the demand rather than by reducing the
supply. When the National Institute of Standards and Technology issued a fact
sheet to disprove the theory that the World Trade Center was brought down by
a controlled demolition, the government spokesman stated that [w]e realize this
fact sheet wont convince those who hold to the alternative theories that our
findings are sound. In fact, the fact sheet was never intended for them. It is for the
masses who have seen or heard the alternative theory claims and want balance.86
The problem with this line of argument, however, is that there are intrinsic
costs to the strategy of giving up on the hard core of conspiracy theorists and
directing government efforts solely towards inoculating the mass audience. For
one thing, the hard core may itself provide the most serious threat. For another,
a response geared to a mass audience (whether or not nominally pitched as a
response to the conspiracy theorists) will lead some to embrace rather than reject
the conspiracy theory the government is trying to rebut. This is the legitimation
dilemma again: to begin a program of inoculation is to signal that the disease is
already widespread and threatening. Under pluralistic ignorance, the perverse
result may actually be to spread the conspiracy theory further.
iii. Cognitive Infiltration and Persuasion
Rather than taking the continued existence of the hard core as a constraint, and
addressing itself solely to the third-party mass audience, government might
undertake (legal) tactics for breaking up the tight cognitive clusters of extremist
theories, arguments and rhetoric that are produced by the hard core and reinforce
it in turn. One potentially promising tactic is cognitive infiltration of extremist
groups. By this we do not mean 1960s-style infiltration with a view to
surveillance and collecting information, possibly for use in future prosecutions.
Rather, we mean that government efforts might succeed in weakening or even
breaking up the epistemological complexes that constitute these networks and
groups.
Recall that extremist networks and groups, including the groups that purvey
conspiracy theories, typically suffer from a kind of crippled epistemology. We
suggest a role for government efforts, and agents, in introducing cognitive
diversity. Government agents (and their allies) might enter chat rooms, online
social networks, or even real-space groups and attempt to undermine percolating
85
For evidence of why this might be so, see Charles G. Lord, Lee Ross, and Mark R. Lepper,
Biased assimilation and attitude polarization: the effects of prior theories on subsequently
considered evidence, Journal of Personality and Social Psychology, 37 (1979), 2098109.
86
Jim Dwyer, U.S. counters 9/11 theories of conspiracy, New York Times, Sept. 2, 2006, p. B1.

CONSPIRACY THEORIES

225

conspiracy theories by raising doubts about their factual premises, causal logic,
or implications for action, political or otherwise.
In one variant, government agents would openly proclaim, or at least make
no effort to conceal, their institutional affiliations. A recent newspaper story
recounts that Arabic-speaking Muslim officials from the State Department have
participated in dialogues at radical Islamist chat rooms and websites in order to
ventilate arguments not usually heard among the groups that cluster around
those sites, with some success.87 In another variant, government officials would
participate anonymously or even with false identities. Each approach has distinct
costs and benefits; the second risks perverse results but potentially brings higher
returns.
In the former case, where government officials participate openly as
such, hard-core members of the relevant networks, communities, and
conspiracy-minded organizations may entirely discount what the officials say,
right from the beginning. Because conspiracy theorists are likely to approach
evidence and arguments in a biased way,88 they are not likely to respond well, or
even logically, to the claims of public officials. Of course government agents offer
arguments and evidence against the conspiracy theory; perhaps their efforts are
merely additional proof that the theory is correct. But the self-sealing quality of
conspiracy theories (we should emphasize) is a matter of degree. Those who hold
such theories may not be totally impervious to contrary evidence, even if it comes
from those who are thought to have a stake in persuasion.
The risk with tactics of anonymous participation is that those tactics may be
discovered or disclosed, with possibly perverse results. If the tactic becomes
known, the conspiracy theory may become further entrenched, and any genuine
member of the relevant groups who raises doubts may be suspected of
government connections. And as we have emphasized throughout, in an open
society it is difficult to conceal government conspiracies, even the sort of
conspiratorial tactic we have suggested, whose aim is to undermine false and
harmful conspiracy theorizing.
If disclosure of the tactic does occur, however, the perverse results are just a
possible cost, whose risk and magnitude is unclear. Another possibility is that
disclosure of the governments tactics will sow uncertainty and distrust within
conspiratorial groups and among their members; new recruits will be suspect and
participants in the groups virtual networks will doubt each others bona fides. To
the extent that these effects raise the costs of organization and communication
for, and within, conspiratorial groups, the effects are desirable, not perverse.
(And both sets of effects might occur simultaneously). So the two forms of

87
Neil MacFarquhar, At State Dept., blog team joins Muslim debate, New York Times, Sept. 22,
2007, p. A1.
88
Cf. Lord et al., Biased assimilation and attitude polarization (showing biased assimilation on
the part of those with strong political commitments).

226

CASS R. SUNSTEIN AND ADRIAN VERMEULE

cognitive infiltration offer evidently different risk-reward mixes. And despite the
dangers, both are potentially useful instruments.
There is a similar tradeoff along another dimension: whether the infiltration
should occur in the real world, through physical penetration of conspiracist
groups by undercover agents, or instead should occur strictly in cyberspace. The
latter is safer, but potentially less productive. The former will sometimes be
indispensable, where the groups that purvey conspiracy theories (and perhaps
themselves formulate conspiracies) formulate their views through real-space
informational networks rather than virtual networks. Infiltration of any kind
poses well-known risks: Perhaps agents will be asked to perform criminal acts to
prove their bona fides, or (less plausibly) will themselves become persuaded by
the conspiratorial views they are supposed to be undermining; perhaps agents
will be unmasked and harmed by the infiltrated group. But the risks are generally
greater for real-world infiltration, where the agent is exposed to more serious
harms. Our main suggestion is just that, whatever the tactical details, there would
seem to be ample reason for government efforts to introduce some cognitive
diversity into the groups that generate conspiracy theories.89
CONCLUSION
Our goal here has been to understand the sources of false and harmful conspiracy
theories and to examine potential government responses. Most people lack direct
or personal information about the explanations for terrible events, and they are
often tempted to attribute such events to some nefarious actor, in part because of
their outrage. The temptation is least likely to be resisted if others are making the
same attributions. Conspiracy cascades arise through the same processes that fuel
many kinds of social errors. What makes such cascades most distinctive, and
relevantly different from other cascades involving beliefs that are also both false
and harmful, is their self-insulating quality. The very statements and facts that
might dissolve conspiracy cascades can be taken as further evidence on their
behalf. These points make it especially difficult for outsiders, including
governments, to debunk them.
Some false conspiracy theories create serious risks. They do not merely
undermine democratic debate; in extreme cases, they create or fuel violence. If
government can dispel such theories, it should do so. One problem is that its
efforts might be counterproductive, because efforts to rebut conspiracy theories

89
There are also hard questions about how, exactly, to introduce cognitive diversity into a group
of people strongly committed to a conspiracy theory. Although our claims do not depend upon the
tactical details, we note a growing body of research indicating that if the goal is to dislodge a
particular belief of an individual or group, the best approach is to begin by affirming other beliefs, or
at least the competence and character, of that individual or group. See the overview in David K.
Sherman and Geoffrey Cohen, The psychology of self-defense: self-affirmation theory, Advances in
Experimental Social Psychology, 38 (2006), 183342.

CONSPIRACY THEORIES

227

also legitimate them. We have suggested, however, that government can minimize
this effect by rebutting more rather than fewer theories, by enlisting independent
groups to supply rebuttals, and by cognitive infiltration designed to break up
the crippled epistemology of conspiracy-minded groups and informationally
isolated social networks.