As the work cited in this dissertation shows, philosophers have been writing about subordinating speech for a few decades now. And while the internet have been around almost as long, much of the philosophical work on hate speech, propaganda, and subordinating speech in
167
general has focused on offline life.182 In-person hate speech like what you might see in public
spaces (Maitra 2012; Langton 2018a; 2018b; McGowan 2012), propaganda as it is disseminated in print or on the radio (Stanley 2015; Tirrell 2012; Smith 2012), and more recently
microaggressions as they occur in say, a workplace or college classroom (Liebow 2017; Rini 2018; Saul 2018b), are the main examples. This has remained the case even as more and more of our lives have migrated online.
These ‘real world’ phenomena are still worthy of philosophical analysis. But online speech raises many new issues for social philosophers and philosophers of language, and these are only beginning to be explored. Fundamental questions like who or what should count as a ‘speaker,’ or how retweets, ‘likes,’ ‘favs,’ and emojis fit into an account of utterances, all need to be re-examined. As does my current question: how has the internet changed subordinating speech?
How much the internet has changed the way we use language is subject to disagreement. One tempting answer is to say that the internet has radically changed everything! And there is at least some truth to this, even for more familiar speech acts, like apologies, retractions, quotes, etc.183 We may think this is especially likely of subordinating speech—in part, since if you give the oppressor a new tool, usually they’ll use it to oppress. And the internet offers a lot of new tools.
182 For a quick and non-decisive example, consider that the index for the (2012) anthology Speech & Harm: Controversies over Free speech, edited by Ishani Maitra and Mary Kate McGowan has no entries for the terms
‘internet,’ ‘website,’ ‘online,’ or other specifically online communication mediums. A few noteworthy exceptions include: Barney (2016), Cherry (2015a), Frost-Arnold (2016), Herbert (2018b), Levmore (2010), Manne (2017), Nguyen (2018), Nussbaum (2010), Rini (2018b), and Technau (2018).
168
2.1. Epistemic Problems
Take propaganda. One initial thought might be that all that the internet has done is made it easier to spread hateful propaganda to more people, more quickly.184 And if this were all it did, that would be problem enough. However, recent analysis suggests things are more complex, and more insidious, than that. Writing for Wired, Issie Lapowsky notes that “[t]he long past of propaganda blended with the communication channels of the present and future form a toxic mix” (2017).185
The reach and speed of the internet is but one concern. These, along with other new technologies that have their home online point to a difference in kind, rather than only degree. In addition to potentially reaching millions of people in mere seconds, photo-, video-, and audio- editing tools offer new tools in the propagandist's toolbox. These have profound effects on the epistemic dimension of propaganda, with the potential to make otherwise unbelievable things frustratingly credible. For example, researchers at Stanford University have developed software capable of “manipulat[ing] video footage of public figures to allow a second person to put words in their mouth—in real time” (Solon 2017; and see BBC News 2017 for a video demonstration). More generally, “deepfakes,” that is, videos created using machine learning to create the illusion that someone has said or done something they never did, are becoming more widely available as the technology improves.186
184 See Ellul (1965) for an argument as to why propaganda will “leave no part of the intellectual or emotional life
alone. [and will] surround [one] on all sides […] by all possible routes” (quoted in Lederer 1995, 133).
185 On this toxic potential, two philosophers quoted in the Wired article seem to agree: “‘I think this is real
dangerous shit,’ says David Livingstone Smith, a professor of philosophy at the University of New England, who specializes in the history of dehumanization and who authored a book on the topic called Less Than Human. […] ‘This is scary shit,’ echoes Jason Stanley, a professor at Yale and author of the book How Propaganda Works, whose father fled Nazi Germany in 1939 (Lapowsky 2017).
186 As Rini (2019) describes them: “Deepfakes are fabricated video or audio recordings created through machine
169
The above capture cases where someone is willing to put the effort in to create an elaborate fake; often, this is unneeded. While the norms here are still evolving, many of us are still disposed to think that images, and especially video, give us a less-filtered version of reality than mere testimony. And this has clear repercussions for the type of propaganda produced. As Lapowsky notes:
It’s not simply cartoons and phony headlines filling people’s minds. Doctored photos and misrepresentations of real footage, like the video the President shared [of supposed ‘Muslim violence’], are a dangerous new development in the history of propaganda, experts say. “Everyone knows caricatures exaggerate,” says Claudia Koonz, a historian at Duke University and author of The Nazi Conscience, “but gullible viewers, including probably Trump, see videos as reality.” (Lapowsky 2017)
In other words, sometimes even small changes to photos and videos—like the simple technique of slowing video down to make speech sound slurred, as occurred for House Majority Leader Nancy Pelosi (see Harwell 2019)—can be just as effective as an elaborate fake.187
The epistemic problems online go deeper when we consider how testimony functions on social media. Even as testimony, it is, as Regina Rini (2017) sees it, a “bent” form of testimony whose features exacerbate the pre-existing problems posed by something like fake news, which is plausibly an old sub-branch of propaganda. For Rini, fake news is not limited to online communications, but there is, as she says, “a strong contingent relationship between fake news and social media” (2017, 45), making the one ripe for the other. As she says:
I suspect that the two bent features of social media testimony are related to one another. Perhaps people are less inclined to subject ridiculous stories to scrutiny because we have unstable testimonial norms on social media. A friend posts a ridiculous story, without comment, and maybe they don’t really mean it. But then other friends ‘like’ the story, or
characteristics of a person, then superimposes this onto recordings of another person. The effect is an apparent recording of a well-known person doing or saying something they never did.”
187 It is worth acknowledging that a common and effective technique to suppress dissident views is simply to flood
the information channels with irrelevant and sometimes contradictory content. The thought being that it is easier to drown out ‘unwanted’ information than to censure it. Zeynep Tufekci (2017) describes this aspect in relation to how networked communication has changed things for both revolutionary social movements and repressive regimes. The philosophy Michael P. Lynch (2016) makes a similar point.
170
comment with earnest revulsion, or share it themselves. Each of these individual
communicative acts involves some ambiguity in the speaker’s testimonial intentions. But, when all appear summed together, this ambiguity seems to wash away. Perhaps the implicit thought is like this: could it really be that all these people aren’t really testifying to this? A thought like that might overwhelm ordinary skepticism about ridiculous testimony. (2017, 49)188
Rini’s analysis shows how fake news can spread originally, given the unstable norms of social media (as they relate to testimony, in particular). This means that little to no malicious intent is needed. But, as Zeynep Tufekci (2017, 241) notes, “social media’s business model financed by ads paid out based on number of pageviews makes it not just possible but even financially lucrative to spread misinformation, propaganda, or distorted partisan content that can go viral in algorithmically entrenched echo chambers.” And, of course, there is little incentive for the private companies that own and operate these social network platforms to eliminate this type of content.
In sum, the internet—and social media in particular—not only amplify the reach and speed of propaganda, but also increase its credibility in several distinctive ways.
2.2. Non-Epistemic Problems
All the above cases all involve deception, though perhaps sometimes unwitting. But another development flips the direction of fit and instead aims to make the world as the propagandist wants it, and then captures that for distribution. ProPublica reported on a new white-supremacist group called RAM (the Rise Above Movement) responsible for some of the violent assaults in Charlottesville, Virginia, in August 2017. They note how RAM members have
188 Chloé Bakalar (2018) offers a different analysis of the epistemic features of social media testimony that focuses
171
teamed up with Vincent James Foxx, “a 31-year-old video blogger and livestreamer with a fondness for white supremacists and radical right-wing politics” who operates as their unofficial propagandist (Thompson, Winston, and BondGraham 2017). Writing about one violent rally in Berkeley, California, they note that:
Foxx wasn’t just documenting the violence at the rally—he was inciting it. On video Foxx posted to YouTube, he can be heard repeatedly encouraging RAM members and others to assault people. “Get that fucking cuck!” he screamed when a RAM member and four or five other men grabbed a counter-protester and began beating him. “Charge!” Foxx yelled as a mob of right-wingers went on the offensive. (Thompson, Winston, and BondGraham 2017, emphasis added)
Here, rather than affecting the reach, speed, or credibility of propaganda, social media is itself providing the impetus for physical violence—the ability to document violence for the purposes of spreading propaganda online served as motivating cause of his incitement. I call attention to this since violence is usually thought of as the downstream ‘mentally mediated’ consequence of propaganda. But here, rather than propaganda begetting violence, we have violence begetting propaganda. At least in a sense.
Indeed, the crossover between online hate and real-life violence is hard to deny. After the New Zealand Mosque shootings, in a piece titled, “Mass Shootings Have Become a Sickening Meme,” New York Times writer Charlie Warzel (2019) wrote:
it’s becoming increasingly difficult to ignore how online hatred and message board screeds are bleeding into the physical world—and how social platforms can act as an accelerant for terroristic behavior. The internet, it seems, has imprinted itself on modern hate crimes, giving its most unstable residents a theater for unspeakable acts—and an amplification system for an ideology of white supremacy that only recently was relegated to the shadows.
And the connections aren’t only there for one-off events, but also for broader cultural shifts. For example, New York Times reporter Kevin Roose speculates that the violent events in Charlottesville in 2017 wouldn’t have occurred had Discord—a chat app meant for gamers but
172
also a home for the ‘alt-right’—and other platforms shut them down earlier (Barbaro 2017). And some take the combined presence of Twitter and Facebook to have been necessary conditions for President Trump’s 2016 election win (for one example, see Sherr and Carson 2017).189 Whether
or not this is true, it has led to much soul-searching among the tech giants, as demonstrated by Facebook’s recent “Hard Questions” series (see Harbath 2018).
In sum, these issues lead many to think, reasonably, that online hate is substantially different from what we’ve seen before, and it changes the game compared to more traditional forms of propaganda. In short, there is therefore good reason to think the rise of the internet— and social media in particular—make online hate speech different from what we’ve seen before and is a cause for alarm. The worldwide concern for fake news demonstrates this well. We are only now beginning to see the effects of these developments, and some analyses are quite damning. Violence arguably caused by online propaganda and fake news has been reported in countries ranging from the US, Myanmar, Germany, India, Canada, and others.190
It is noteworthy, however, that the bulk of this analysis focusses on subordinating speech as it functions in its propagandistic mode—as outreach—rather than cases where it is directly targeting and harming particular individuals.191 This has led to discussions of the ‘potential’ harms of online hate and have focused on abstract values ‘democracy’ as its main victim. But this ignores those who have already been victimized by online hate and minimizes the harm they’ve experienced—this is no surprise, as most of these victims are not white men. In what
189 That is, how a fringe online movement comprised of angry gamers and angry racists—what some have called
‘the Chanterculture’ after the notorious online forum 4chan—played a key role in elevating Trump to the presidency.
190 Indeed, as I mentioned in Chapter One, the Quebec City Mosque shooter was partly inspired by online extremist
content, and then the final straw was a tweet by Prime Minister Trudeau. See Cecco (2018) for details.
191 Note that a single speech act can play both roles at once, though. I discussed these differences and where they
173
follows, I will examine online abuse as a topic worthy of serious philosophical investigation. I consider it as a structural problem, but I want to draw attention to how it is experienced by those targeted by online abuse.