We Should Not Get Rid of Incivility Online

Abstract

Incivility and toxicity have become concepts du jour in research about social media. The clear normative implication in much of this research is that incivility is bad and should be eliminated. Extensive research—including some that we’ve authored—has been dedicated to finding ways to reduce or eliminate incivility from online discussion spaces. In our work as part of the Civic Signals Initiative, we’ve been thinking carefully about what metrics should be adopted by social media platforms eager to create better spaces for their users. When we tell people about this project, removing incivility from the platforms frequently comes up as a suggested metric. In thinking about incivility, however, we’ve become less convinced that it is desirable, or even possible, for social media platforms to remove all uncivil content. In this short essay, we discuss research on incivility, our rationale for a more complicated normative stance regarding incivility, and what other orientations may be more useful. We conclude with a post mortem arguing that we should not abandon research on incivility altogether, but we should recognize the limitations of a concept that is difficult to universalize.

Incivility and toxicity have become concepts du jour in research about social media. The clear normative implication in much of this research is that incivility is bad and should be eliminated. Extensive research—including some that we’ve authored—has been dedicated to finding ways to reduce or eliminate incivility from online discussion spaces. In our work as part of the Civic Signals Initiative, we’ve been thinking carefully about what metrics should be adopted by social media platforms eager to create better spaces for their users. When we tell people about this project, removing incivility from the platforms frequently comes up as a suggested metric. In thinking about incivility, however, we’ve become less convinced that it is desirable, or even possible, for social media platforms to remove all uncivil content. In this short essay, we discuss research on incivility, our rationale for a more complicated normative stance regarding incivility, and what other orientations may be more useful. We conclude with a post mortem arguing that we should not abandon research on incivility altogether, but we should recognize the limitations of a concept that is difficult to universalize.

Research on incivility often makes two unstated assumptions: Incivility can be measured, and incivility is bad. We take exception to both assumptions. One of the main problems is that nobody really agrees on what incivility is. The public sees it as everything from impolite acts like name-calling and profanity to specific events, such as President Bill Clinton’s dalliance with Monica Lewinsky (Muddiman & Kearney, 2018).

Scholars offer a bit more conceptual clarity. But not much. A main cleavage is whether incivility and impoliteness are synonyms or separate concepts. Papacharissi (2004) argues persuasively that they are separate, asserting that profanity and name-calling are merely impolite, while more virulent speech—such as threats against democracy or stereotyping—are uncivil.

Muddiman (2017) asserts that person-level incivility is speech that could be civil if it didn’t include impoliteness. But public-level incivility is a distinct concept, including failure to compromise and lack of comity among politicians (Muddiman, 2017). Tromble (2018) distinguishes between incivility as “profanity or crude language” and intolerance, which “targets categories of people for discrimination, hate, abuse, etc.” Rossini (2019) strikes a related divide, positing that incivility is intense speech marred by foul language while intolerant language morally disrespects individuals or groups. Chen (2017) conceptualized incivility on a continuum that includes impoliteness on the more benign end and hate speech on the unacceptable end (see also Sydnor, 2018).

Operational definitions include profanity (Coe, Kenski & Rains, 2014; Stroud, Scacco, Muddiman, & Curry, 2015) the use of all capital letters (Gervais, 2015; Sobieraj & Berry, 2011), aspersions (Coe et al., 2014), accusations of lying (Coe et al., 2014), pejorative speech (Coe et al., 2014; Kenski, Filer, & Conway-Silva, 2018), ideologically extreme language ((Muddiman, 2017; Sobieraj & Berry, 2011; Stroud et al., 2015), exaggerated argument (Stroud et al., 2015), emotional language (Sobieraj & Berry, 2011), and misinformation (Muddiman, 2017).

Other approaches, like the one used in the development of Google’s Toxicity algorithm, also presume that detecting incivility is possible using a universal definition. For this algorithm, Google asks a sample of people to rate comments on a scale from “Very toxic” to “Very healthy.” They define toxic as “a rude, disrespectful, or unreasonable comment that is likely to make you leave a discussion.” These conceptual and operational definitions presuppose that incivility can be identified.

Despite well-intentioned efforts, research shows that incivility is in the eye of the beholder. Even work that has found substantial agreement in perceptions of incivility still presents some variance in understanding of the concept. For instance, while Stryker, Conway, and Danielson (2016) found that over 80% of the students in their study perceived slurs and threats of harm as very uncivil, interruptions, and character attacks prompted much more variance with less than 40% of students perceiving these behaviors as very uncivil.

In addition, individual characteristics matter in understanding incivility. People who are older (Ben-Porath, 2008), conflict avoidant (Mutz & Reeves, 2005), and who score high on the Big Five personality trait of agreeableness (Kenski, Coe, & Rains, 2017) may react more strongly to incivility than others. Group identity matters as well. Partisans are likely to perceive their out-group political party to be more uncivil than their in-group political party, even when the partisans are engaging in the exact same behaviors (Muddiman, 2017; Mutz, 2015). In addition, when individuals are asked to generate their own examples of politicians behaving uncivilly, they provide answers ranging from “personal attacks” to “empty promises” and “screaming and yelling” to “treason” (Muddiman, 2019). In sum, research indicates that there are a variety of approaches to incivility, and that individual and group differences influence reactions to potentially uncivil speech.

Given variation in what people believe is uncivil, any attempt to categorize incivility relies on a particular cultural understanding of what is uncivil. So-called uncivil terms can be entirely inappropriate when used by some people in some contexts. These same words and phrases, however, can be terms of endearment when used by the same people in other contexts, or by different people in the same context. And perceptions of incivility vary across people, making it difficult to know when incivility actually exists.

The second main problem is the assumption that underlies incivility research, which is usually that incivility is bad. Certainly, it can lead to some normatively negative outcomes. Incivility can boost aggressiveness (Gervais, 2015; Rösner, Winter, & Krämer, 2016) or lead people to retaliate (Chen & Lu, 2017), escalating nasty talk. Incivility in news comment sections can increase polarization (Anderson, Brossard, Scheufele, Xenos, & Ladwig, 2014) and bring about less favorable impressions of news articles and brands (Prochazka, Weber, & Schweiger, 2018; Tenenboim, Chen, & Lu, 2019). There also is a fear that repeated exposure to incivility in digital spaces will degrade what is normatively acceptable—leading to a dystopian society where even the most offensive speech is tolerated.

In other cases, the normative understanding of incivility is conveyed more subtly in research looking at ways to reduce incivility. For example, studies have found that using real names in online newspaper comments sections is associated with lower levels of incivility (Ksiazek, 2015; Santana, 2014 although see Chen, 2017; Rösner & Krämer, 2016). Stroud et al. (2015) showed that reporter involvement in the comments of a news organization’s Facebook page decreased the probability of uncivil comments. And work by Ksiazek (2015) suggested a number of policies and practices that could reduce incivility, including requiring on-site or third-party user registration, moderating comments before they appear on site or after they’ve been posted, and using reputation management systems that reward positive contributions via mechanisms such as likes, votes, and badges. All of these studies propose that there is a social benefit to curbing incivility.

Yet incivility can sometimes have a beneficial social purpose. Although profanity is entirely inappropriate in some contexts, it can be appropriate, and even an asset, in others. In some contexts, profanity is a way to bond with co-workers (e.g., Faulkner, 2009); female workers, for instance, have used profanity to establish their authority among male colleagues. Further, there’s an argument to be made that sometimes incivility is required for groups to get their point across. Research also suggests that incivility can draw attention (Mutz & Reeves, 2005), ignite interest in politics (Brooks & Geer, 2007), and prompt intentions to participate politically (Chen, 2017). As Herbst (2010) argues, “The line between passionate engagement and civility seems chronically fuzzy and arbitrary” (p. 3).

Critical work acknowledges the role of power in understanding incivility: Those in power can decide what is socially normative and then label as uncivil anything that pushes against this status quo. For example, women, and especially women of color, have historically been framed as uncivil when they step out of the private sphere to speak as public figures (Lozano-Reich & Cloud, 2009). Protesters are often covered by news media as deviant actors (Gitlin, 2003). Some scholars have, therefore, pushed against calls for civility, arguing that “civility and decorum . . . operate as border patrols” that make social mobility more difficult for disempowered groups (Young, Battaglia, & Cloud, 2010, p. 430) and even that “civility is the new censorship” (Bennett, 2011, p. 2). A general ban of incivility on platforms may constrain socially beneficial uses of incivility and cede more power to the already powerful.

When platforms and academics take it upon themselves to decide what is uncivil, they are imposing a particular definition of what counts and what doesn’t. And inevitably, these definitions may force a particular worldview.

One may reasonably retort that our argument is relativism at its worst, or that we are arguing that everything should be permitted in digital spaces, despite ample research showing that incivility can have detrimental consequences. This, however, is not our position. We do not quibble with the right of news organizations and social media platforms to prohibit uncivil speech or use algorithms to de-prioritize certain words and phrases. What we take issue with is an end goal of a sanitized space. One person’s incivility is another’s civility. And when an organization settles on a particular definition of incivility, they impose this view on others. Imperfect speech can still have value. We agree with Papacharissi (2004) who asserts, “Sanitized and controlled conversation does not fully capture the conditioned illogic of human thought” (p. 266). We wish for an online discourse that strikes the “sweet spot that is not so polite that it prohibits disagreement or discord but not so nasty that it makes rational speech impossible” (Chen, 2017, p. 177).

Some words and phrases arguably should be prohibited on social media, but not for the reason of preserving civility. It should be because the organization wants to create a particular culture for a specific audience, and the words and phrases chosen are inconsistent with that vision. To treat incivility as anything other than a norm agreed to by some and dismissed by others, however, would be misguided.

So When it Comes to Incivility, What can be Done?

The first possibility could be to allow communities to define for themselves what counts as civil or uncivil, rather than prescribing these standards as an outsider. An interesting initiative that acknowledged the diversity of perspectives about incivility is the now-defunct Civil Comments. The group created a commenting platform where commenters had to rate the civility of other people’s comments before their own comment could be posted. Each group adopting the platform would be responsible for creating its own standards for what counted as civil. In academic contexts, having participants in particular conversations report what is civil or uncivil would be a defensible way to detect incivility in situ.

A second approach could be to focus on outcomes beyond civility and incivility. The focus on incivility is not because certain words and phrases are, in and of themselves, universally bad. Rather, incivility is a focus because of the detrimental consequences it can cause, such as dehumanizing others and increasing polarization. Rather than focusing on incivility, platforms and scholars alike could focus on interventions targeting these factors that are ultimately of interest. Platforms could focus on quelling actual harassment—when people release personal and potentially damaging information about others or rally a crowd to attack a person or group indiscriminately. As another example, interventions designed to humanize others may have the desired effect of creating respect for others without the baggage of defining what is civil and uncivil.

Do we have an incivility problem in the digital public sphere? Many would suggest that we do. More than 84% of U.S. adults say they’ve experienced incivility in online or offline life, and people report an average of 5.4 uncivil online encounters every week (Weber Shandwick, Powell Tate, & KRC Research, 2018). In efforts to improve digital spaces, reducing incivility is an oft-referenced target. But is it the right one? We suggest that it is not. Incivility is relative and, as such, an inconsistent yardstick by which to judge whether a given comment is constructive or appropriate. An emphasis on incivility tends to privilege the groups that have the power to impose their definitions on a conversation, meaning marginalized groups may be more likely to be viewed as uncivil. Sometimes, incivility may actually be required for these groups to be heard. Our purpose in writing this essay is not to argue that incivility research is misplaced. Rather, we argue that research on civility should more bluntly wrestle with the perceptual element of incivility. Overall, we think there are better metrics by which to judge social media conversations and to strive for online debate that is meaningful, reasoned, and productive.