As Twitter attempts to fix its struggling fortunes by positioning itself as a counter revolutionary institution of the extant socialist order, it has come out that Twitter will be adopting Susan Benesch's dangerous framework for evaluating speech. Benesch's dangerous framework provides a methodology for censors to grade speech utterances on their likelihood of affecting change in the world so that they might be censored while innocuous utterances of little impact may be permitted in order to create a simulacrum of free speech occurring. At the core of Benesch's framework which purports to be motivated by principles assuming equality among persons is a concession to the reality that speakers are fundamentally unequal. For interested entomologists the full text of a working paper by Benesch produced for the "World Policy Institute" in 2012 on her framework is submitted in full plaintext below:
By Susan Benesch
Senior Fellow, World Policy Institute
January 12, 2012.
Dangerous Speech: A Proposal to Prevent Group Violence
This paper presents draft guidelines for monitoring speech and evaluating its dangerousness, i.e.
the capacity to catalyze violence by one group against another. Inflammatory speech is of special interest
for atrocity prevention since it tends to rise dramatically before outbreaks of mass violence, suggesting
that it may serve as a basis for efforts to prevent such violence, including genocide. However such efforts
must not infringe upon freedom of expression, a fundamental right whose exercise can, itself, prevent
There are at least three ways in which inflammatory speech 1presents an opportunity for violence
prevention, and therefore three distinct applications for the guidelines. First, such speech can serve as a
key indicator for early warning, since it is often a precursor – if not also a prerequisite – for mass
violence.2 Second, it may be possible to limit violence by finding ways to limit such speech or its
dangerousness. Third, speakers may be held accountable for speech that constitutes crime.
1. The Importance of Dangerous Speech
The guidelines are designed to identify a subset of hate speech, which I have termed Dangerous
Speech. Hate speech is variously defined in law and in common parlance, but is generally understood to
mean speech that denigrates people on the basis of their membership in a group, such as an ethnic or
religious group. This category of speech is too broad for successful early warning of mass atrocities, for
two related reasons. First, hate speech is common in many societies, unfortunately, including those at
minimal risk of genocide. Second, some hate speech does not appreciably increase the risk of mass
violence, although it may cause serious emotional and psychological damage.
In other words, speech can harm directly or indirectly, or both. It may directly offend, denigrate,
humiliate or frighten the people it purports to describe – such as when a racist shouts at a person of color.
Speech can also bring about harm indirectly – and with equal or even greater brutality – by motivating
others to think and act against members of the group in question.
‘Speech‟ includes any form of expression, including images such as drawings or photographs, films, etc.
It has long been widely assumed that inflammatory speech can catalyze and magnify mass violence, but the
inference was unsupported by data until just recently. In what is apparently the first quantitative evidence of a link
between speech and mass violence, David Yanagizawa‟s 2009 statistical study of the effects of the virulent
propaganda outlet Radio Television Libre des Milles Collines (RTLM) in Rwanda indicates that killings were 6577% higher in Rwandan villages that received the RTLM signal, compared with those that did not (for exogenous
reasons such as topography) receive the signal. See David Yanagizawa, Propaganda and Conflict: Theory and
Evidence from the Rwandan Genocide, available at http://people.su.se/~daya0852/Rwanda_jmp.pdf
When an act of speech has a reasonable chance of catalyzing or amplifying violence by one group
against another, given the circumstances in which it was made or disseminated, it is Dangerous Speech.3
To my knowledge there is no other framework or methodology designed to identify this type of speech,
either in the literature or in use in the field. 4
2. Five Defining Variables
The guidelines are based on five variables which affect the dangerousness of a particular speech
act in the time and place in which it was made or disseminated: the speaker, the audience, the speech act
itself, the social and historical context, and the mode of dissemination.
The most dangerous speech act, or ideal type of dangerous speech, would be one for which all
five variables are maximized:
1. a powerful speaker with a high degree of influence over the audience,
2. the audience has grievances and fear that the speaker can cultivate
3. a speech act that is clearly understood as a call to violence
4. a social or historical context that is propitious for violence, for any of a variety of reasons,
including longstanding competition between groups for resources, lack of efforts to solve
grievances, or previous episodes of violence
5. a means of dissemination that is influential in itself, for example because it is the sole or
primary source of news for the relevant audience
Each speech act should be evaluated in terms of the information available, regarding the full set
of variables. As in the case of the OSAPG‟s own Analysis Framework for gauging the risk of genocide,
the variables are not ranked. Nor are they weighted equally across cases: in many circumstances, one or
more variables will „weigh‟ more than others. For example an especially outrageous or frightening speech
act may be more important (i.e. more dangerous) than other factors in a particular instance. In other cases,
an audience may be especially vulnerable to incitement, or the overweening influence of a speaker may
contribute most to the force of his or her speech.
In law, speech intended to provoke another to commit a crime is incitement, or instigation. Dangerous Speech can
also be understood, therefore, as incitement to mass violence, which is reasonably likely to be successful. We do not
limit “mass violence” to genocide, since inflammatory speech is a precursor to outbreaks of mass violence
including, but not limited to, genocide.
For further information and related research, see Susan Benesch, Vile Crime or Inalienable Right; Defining
Incitement to Genocide, Va J. Int‟l L, 48 Va. J. Int‟l. L. 485 (2008), available at
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1121926 and Susan Benesch, The New Law of Incitement: A
Critique and a Proposal, available at http://www.ushmm.org/genocide/spv/pdf/benesch_susan.pdf and
www.voicesthanpoison.wordpress.com For an address on the topic, given as a side event to the UNGA in October
2010, see http://www.unmultimedia.org/tv/webcast/2010/10/special-event-dialogue-on-dangerous-speech-on-theroad-to-genocide.html
Analysis may be conducted with varying degrees of information: the more information available,
the more fine-grained the analysis will be. Since speech can only be well understood (and its influence
gauged) in the psycho-social, historical, and cultural context in which it was made or disseminated, the
analysis is best carried out with knowledge of the relevant language, culture, and social conditions – or
with assistance from advisors who have such knowledge.
3. Guidelines for Evaluative Monitoring of Inflammatory Speech
The persuasiveness or force of speech has long been studied, albeit usually not in the context of
mass violence. The first three of our variables were identified in a different form by Aristotle, who
observed, in his Rhetoric, that the means of persuasion reside in the character of the speaker, the
emotional state of the audience, and in the argument or logos, itself. Also searching for the keys to
persuasiveness in their pioneering studies of propaganda in the aftermath of World War II, Carl Hovland
and other social psychologists at Yale University focused on the same three variables to construct the socalled Yale attitude change model.
To evaluate a speech act‟s capacity to move one group to attack another – the capacity not only
to persuade but to inspire action – in the contemporary world, I have added two key variables to the first
three: the historical and social context in which a speech act is made or disseminated, and the medium or
means of dissemination. Social context is an essential variable, since mass violence is unlikely in certain
contexts, irrespective of other factors such as the intent and rhetorical skill of the speaker, and more likely
in others. Finally, in the present day, the means of dissemination (limited in Aristotle‟s time to speech
delivered in person) are increasingly varied, and can be influential in and of themselves. A message may
be more compelling because it is delivered over a particular radio station, for example, or via SMS.
Any number of factors may affect the dangerousness of speech, but the most relevant factors are
all related to the five variables that I have identified, based on fieldwork and research in literature
including genocide studies, social psychology, and the philosophy of language. It is therefore possible to
conduct a relatively streamlined analysis, taking each of the variables in turn. I have listed the variables
together with exemplary factors presented as questions, since those would be most easily incorporated
into a monitoring plan, and some explanatory notes.
1. THE SPEAKER
It should be noted that some cases present more than one speaker, such as when a radio talk-show
host interviews a public figure.
a. Did the speaker have authority, power, or influence over the audience? Influence
or authority need not derive from a de jure political post; cultural and religious
figures and entertainers often have more influence over an audience than political
officials. Some speakers control resources needed by an audience, or can deploy
force. Any of these factors can render their speech more dangerous.
b. Was the speaker charismatic or popular? A speaker may be seen as popular only
by a subset of the audience, but those listeners may also be most likely to react to the
2. THE AUDIENCE
a. Who was the audience most likely to react to the speech at issue? In many cases,
the audience may be large or somewhat indeterminate, e.g. “the public” or all
listeners of a radio station. The analysis should focus on the audience that is most
likely to react with violence in response to the speech.
b. Was the speech directed primarily at members of the group it purported to
describe, i.e. victims, or at members of the speaker’s own group, or both? If the
latter is true, the speech is more likely to be dangerous. The primary audience is often
indicated by the language or venue in which the speech was delivered.
c. Did the audience have the means or capacity to commit violence against the
group targeted in the speech? If an audience is unable to commit mass violence,
incitement cannot succeed, and is not Dangerous Speech.
d. Was the audience suffering economic insecurity, e.g. lacking in food, shelter,
employment, especially in comparison with its recent past? 5
e. Is the audience characterized by excessive respect for authority? This would
make an audience more vulnerable to incitement.
Was the audience fearful? Fear might be objectively reasonable or not; its impact
may be equally large, and equally well exploited by a compelling speaker.
3. THE SPEECH ACT
Speech may be made in any number of forms, and disseminated by myriad means: a shouted
command, a song broadcast at a rally, a newspaper editorial, a tweet, poster, web page, SMS blast, leaflet,
film or photograph.
a. Was the speech understood by the audience as a call to violence? Inflammatory
speech is often expressed in elliptical, indirect language, which can be variously
interpreted. For this analysis, the only relevant meaning is the way in which the
speech was understood by the audience most likely to react, at the time when it was
made or disseminated.
b. Did the speech describe the victims-to-be as other than human, e.g. as vermin,
pests, insects or animals? This is a rhetorical hallmark6 of incitement to genocide,
and to violence, since it dehumanizes the victim or victims to be.
This factor, like the next two, is emphasized by the genocide scholar Ervin Staub, who has carried out extensive
research into the causes of mass violence. See, e.g. Overcoming Evil; Genocide, Violent Conflict, and Terrorism, p.
208 et seq.
These hallmarks are discussed at greater length in Vile Crime or Inalianable Right, supra note 3.
c. Did the speech assert that the audience faced serious danger from the victim
group? Another hallmark of incitement, this technique is known as “accusation in a
mirror.” Just as self-defense is an ironclad defense to murder, collective self-defense
gives a psychological justification for group violence, even if the claim of selfdefense is spurious.
d. Did the speech contain phrases, words, or coded language that has taken on a
special loaded meaning, in the understanding of the speaker and audience?
Such coded language is typical of Dangerous Speech. It bonds the speaker and
audience more tightly together. Familiar examples of this are the phrase “go to
work,” used as code for killing during the Rwandan genocide, or the word “inyenzi”
(Kinyarwanda for “cockroach”), used to refer to Tutsi or even to non-Tutsi who
sympathized with Tutsi.
4. SOCIO-HISTORICAL CONTEXT
a. Were there underlying or previous conflicts between relevant groups?
b. Were there recent outbreaks of violence following other examples of hate
speech? This would put both speaker and audience on notice that such speech can
indeed lead to violence, thereby increasing the dangerousness of the speech.
c. Were other risk factors for mass violence present, such as weak democratic
structures and rule of law, and structural inequalities and discrimination against a
group or groups? These risk factors have been described in greater detail elsewhere,
e.g. in the OSAPG‟s Analysis Framework.
5. MODE OF TRANSMISSION
a. Was the speech transmitted in a way that would reinforce its capacity to
persuade, e.g. via a media outlet with particular influence or without
competitors? Other modes of transmission can be compelling in and of themselves,
such as new media platforms that make members of an audience feel that they are
part of a select and privileged group. Music can also increase the force or influence
of a message.
b. Was the audience exposed to, or did it have access to, alternate views or sources
of information? Where there are no alternative sources of information, the impact of
speech is much greater.
c. Was the speech frequently repeated, in similar form or content? Repetition
magnifies the impact of a message.