Department of Informatics Magister thesis, 15 hp
Master’s Program in Human-Computer Interaction and Social Media SPM 2020.13
ONLINE SHAMING Ethical Tools for
Human-Computer Interaction Designers
Erik Campano
Dedication to Kathleen
Content Warning
This paper discusses the topics of racism, transphobia, and gender-based-violence.
1
Abstract
A set of tools – concepts, guidelines, and engineering solutions – are proposed to help human-computer interaction designers build systems that are ethical with regards to online shaming. Online shaming’s ethics are unsolved in the literature, and the phenomenon can have devastating consequences, as well as serve social justice. Kantian ethics, as interpreted by Christine Korsgaard, provide our analytical methodology. Her meta-ethics invokes Wittgenstein’s private language argument, which also models relevant concepts in human-computer interaction theory. Empirical studies and other ethicists’ views on online shaming are presented. Korsgaard’s Kantian methodology is used to evaluate the other ethicists’ views’ moral acceptability, and guidelines are drawn from that analysis. These guidelines permit shaming, with strong constraints. Technical engineering solutions to ethical problems in online shaming are discussed. All these results are situated in the public dialogue on online shaming, and future research from other ethical traditions is suggested.
Keywords: online shaming, ethics, human-computer interaction
1. Introduction
1.1 Three Examples of Online Shaming
We begin with three examples of online shaming. I will be using these examples throughout this paper.
Justine Sacco: Justine Sacco was an American communications executive at a large media and internet holding company. She had a history of fighting for causes endorsing racial equality. She also had an “ascerbic” sense of humor (Ronson, 2015). On December 20, 2013, on the way to visit family in South Africa, just before she boarded a plane in London, she tweeted: “Going to Africa. Hope I don’t get AIDS. Just kidding. I’m white!”
[Fig. 1] She says she intended this as a joke, an ironic, self-reflective critique of white privilege. (ibid). The tweet was republished quickly on the blog of technology journalist Sam Biddle. Readers, who did not know Sacco, did not appreciate the comment as an ironic joke, but rather saw it as serious racism. During her 11 hour flight to South Africa – when she was offline and could not explain herself – her tweet went viral. Sacco became the target of online anti-racism vigilantes, who posted tens of thousands of threatening and abusive messages. The number one worldwide trending Twitter hashtag became
#HasJustineLandedYet. “We are about to watch this @JustineSacco bitch get fired. In REAL time. Before she even KNOWS she’s getting fired,” tweeted one critic. Indeed, when she landed, she had been fired (Billingham and Parr, 2019) and branded a racist on social media and search engines.
Natalie Wynn: Natalie Wynn is a transwoman, former philosophy graduate student and YouTube vlogger who since 2017 has published movie-length video essays about ethics, politics, race, and gender, on a channel called ContraPoints. In October 2019, in a video,
2
she included a 10 second clip voiced over by transsexual pornographic actor Buck Angel, whom some transgender rights activists criticize as a transmedicalist – that is to say, someone who believes that people who do not experience gender dysphoria, or undergo medical transition, are not truly transgender. Wynn says she is not a transmedicalist.
Nonetheless, transmedicalism critics barraged Wynn with hatred on social media for having used Angel’s voice. “Pee pee, poo poo, I'm ContraPoints and I can't stop being a transmedicalist and sh*tting myself in public bwuh,” wrote one commenter. The critics called for Wynn to be “cancelled” – for her channel to be boycotted. "Deplatform her just like every other racist, sexist, bigoted content creator,” a critic wrote. Wynn disappeared from public view for about a month, and then resurfaced with a widely-praised video called Canceling. In it, she details the process of online shaming, and also movingly describes the negative effects it had on her – and her colleagues. “I'm becoming a burden to anyone who associates with me. … I'm working on this totally alone because I'm aware of how radioactive I am right now. And I don't want to contaminate anyone else,” she says (Wynn, 2020). Canceling has received almost two million views.
Brock Turner: Brock Turner was a 20-year old Stanford undergraduate. In January 2015, after a party in which he and a young woman named Chanel Miller drank heavily, he was discovered by two Swedish graduate students outdoors, on top of Miller while she was unconscious from the alcohol. Turner tried to run, but the students caught up and tackled him. Stanford banned Turner from the campus for life. He was convicted of sexual assault with intent to rape an intoxicated person, and two other charges. Judge Aaron Persky sentenced Turner to six months in prison, three years’ probation, and lifetime registration as a sex offender. Anti-gender-based-violence advocates, led by Stanford law professor Michele Dauber, felt that this sentence was too lenient, and launched an online campaign to have Persky recalled from the bench. Countless memes and websites labeled Turner him a rapist; an introductory law textbook even used his picture next to the definition of rape [Figure 2.] (Rennison, 2017). After Dauber’s campaign, county voters recalled Persky from the bench – the first time that a judge in California had been recalled in 80 years. Persky took a new job as a high school tennis instructor, but was fired after a Change.org petition accused that school of “explicitly and ignorantly allowing rape culture to ensue.” (BBC, 2019) Dauber herself was the target of an online counter-campaign, which she says led to offline threats, such as two stalkers, and an envelope sent to her house with white powder in it. (Deruy, 2018) In September 2019, Miller told her story in a bestselling book, Know My Name: A Memoir. Turner still faces online death threats and calls for his execution. On May 14, 2020 one user tweeted, “my homies wanna kill brock turner”. [Fig. 3]
There are uncountably many identified cases of online shaming, but for the purposes of this essay we shall focus on these three. Each has a certain extreme quality that makes it particularly useful for our analysis:
Justine Sacco’s case was extreme in the relative small scale of the trigger (a single 12-word tweet from a private, largely unknown person) versus the large scale of the reaction (worldwide media coverage).
Natalie Wynn’s case is extreme in how successfully the target defended herself, and in her detail of description of the dynamics and personal effects of online shaming.
3
Brock Turner’s case is extreme, as popular commentators have called it one of the most morally justified examples of online shaming. An influential anti-shaming book, Shame Nation: The Global Epidemic of Online Hate, says about Turner: “We raged, we blogged, we vented, but finally, all that was left was to enact vigilante justice.” (Scheff et al., 2017, p. 13) The Turner case is also important for our purposes, because many extant academic arguments for online shaming see it as a tool to combat gender-based-violence.
1.2 A Working Definition of Online Shaming
Online shaming, as defined in the literature, has many types. These include:
• doxing – publishing private information about an individual, such as his address or workplace, in order to encourage others to harass him
• revenge pornography – publishing sexual content without a subject’s permission
• harassment – for example, online stalking, blackmail, threats, and fake profiles
• bullying – for example, children writing hurtful things about another child online
• bigotry – publishing hate speech about a group based on race, gender, sexual orientation, religion, ethnic origin, or other identity status
• gossiping – sharing false personal information carelessly online (Laidlaw, 2017) Popular and academic literature broadly concede that these types are ethically unacceptable. We will therefore skip them in this paper. Rather, we will explore one other, specific type, which, as will become clear, raises the ethical issues about which we are concerned. For our purposes, shaming is: a large group of people publishing many individual posts on social media, intending to punish a person extrajudicially for doing something they perceive as unjust, for the ostensible purpose of societal modification.
This type is sometimes called online vigilantism, although many definitions of that term have been proposed in the academic literature (Billingham and Parr, 2019; Gallardo, 2017; Chandler, 2016; Chang and Poon, 2016; Hou, 2017). In this paper, we shall use online shaming and online vigilantism almost interchangeably. The former emphasizes the act of punishing, whilst the latter, its extrajudicial nature.
How large is large, and how many posts are necessary, for punishment to count as shaming? There is no number. As I shall argue, the threshold is crossed when moral accountability becomes distributed among many shamers.
1.3 Knowledge Assumed by this Paper
This paper is written for human-computer interaction designers. Therefore, it assumes knowledge of certain human-computer interaction concepts, specifically: distributed cognition, embodied interaction, persuasive technologies, metaphor, blended space, and the spiral of silence. This paper assumes no knowledge of academic philosophy, particularly ideas from Korsgaard’s Kantianism, Wittgenstein’s private language argument, social contract theory, and feminist ethics. These will be explained in due course.
4
2. Research Question and Aim
Our research question is then: what tools can we develop to help human-computer interaction designers build systems that are ethical with regards to online shaming?
By tools, I mean three different things:
• a set of concepts, originating in academic ethics and human-computer interaction theory, that designers can use to evaluate the moral dimension of online shaming
• a set of ethical guidelines that designers should follow
• practical engineering solutions for preventing unethical online shaming
The aim of the paper is, then, to provide these tools to designers of any systems that might enable shaming, which, now at least, are primarily social media platforms and blogs.
What justifies this enquiry? There are three main answers to this question. First: online shaming is a sociologically dangerous phenomenon. Its outcomes include suicides (Sommer, 2015), public calls for sexual or physical violence against individuals (Gallardo, 2017), and punishment of innocent parties (Sprecher, 2016). Second: the ethics of online shaming are not a solved problem in the academic literature. Authors from the philosophical ethical traditions of social contract theory, and some feminist ethicists, argue directly against online shaming. Others – primarily feminist ethicists – argue in favor of it, directly or indirectly. Third: no code of ethics has been published for human-computer interaction designers with regards to online shaming. They have no commonly agreed-upon professional best practices to deal with the phenomenon.
This paper addresses all three of these answers head-on. It develops a set of conceptual tools allowing the reader to do her own ethical analysis of online shaming. It suggests guidelines which could be primordial bases for a future code of ethics. It then explores technologies to implement these guidelines. My primary goal is not to derive a comprehensive, permanent set of ethical principles about online shaming for designers;
that would be impossible to do in 29 pages. The goal is rather to offer original insights on moral issues, surrounding the phenomenon, which will inspire the reader to think through her own opinions about its ethics.
3. Research Methodology
*3.1 Overview of Research Methodology
We define our analytical framework as the Kantian ethical test called the Categorical Imperative, as interpreted by prominent modern Kantian scholar Christine Korsgaard (1996). We will intertwine the Categorical Imperative with concepts from human-computer interaction theory listed above. In a moment, I will justify why I made this choice. Before we apply the method, we will need, in the Related Research section below, to establish:
• a more complete empirical understanding of the characteristics of online shaming
* In the assignment template for this thesis, Related Research comes before Research Methodology.
However, in my paper, I have reversed this order. An understanding of our methodology makes it much easier to understand related research.
5
• the various arguments for and against online vigilantism from the academic philosophical literature, and in a couple of cases, from particularly insightful popular philosophical literature
We will then use Korsgaard’s Kantianism’s method to directly analyze the existing arguments for and against online shaming. Each separate analysis should produce one or more ethical guidelines. Once those concepts are compiled, we will suggest engineering solutions to help avoid enabling potentially unwanted online shaming.
At a surface level, Kantian methodology is a simple process. Kant’s universal test, the Categorical Imperative, is used to determine whether an action is morally acceptable or not.
However, the devil is in the details. Kantianism is particularly useful to human-computer interaction designers, if we describe the Categorical Imperative’s meta-ethical background using concepts that also underlie human-computer interaction theory. That is to say, the semantic, epistemological, and metaphysical theories out of which Korsgaard’s Kantianism can be built can also be used to model distributed cognition, embodied interaction, and so forth. The Categorical Imperative is not just a framework we apply around the phenomenon of online shaming; rather, the Categorical Imperative raises questions, as we shall see, about specific techno-social choices made by online shamers, as those choices are described by human-computer interaction theory.
As a human-computer interaction designer, you may be used to reading short method sections in papers, which you might even skip over because many research methods are standardized in the discipline. In philosophy papers, however, method sections are often long, and this one is no exception. This is because much original thought in a philosophy paper occurs because the author has created a new methodology for analyzing perhaps an old problem. In fact, some of your first tools – your set of concepts – for evaluating the ethics of shaming in your interaction design are here, in the methodology section. To keep things clear, key tools borne out of the research methodology are underlined.
3.2 Why This Methodology?
Korsgaard’s Kantianism is justified as a research method for three reasons.
First: online vigilantism has already received analysis from two other major schools of Western normative academic ethics: social contract theory (as understood by John Locke) and feminist ethics. Furthermore, the most well-founded feminist argument for shaming – Goguen’s – is implicitly utilitarian, as I hope to show when we describe it later. Since there are four major normative Western ethical traditions (Driver, 2007)…
1. deontology, prominently represented by Kantianism and social contract theory 2. consequentialism, prominently represented by utilitarianism
3. feminist ethics, borne in part of critical theory, and
4. virtue ethics, originating in ancient Greek philosophy and arguably finding on modern form in existentialism (Davenport, 2001, p. 265 ff.)
…the gap in the literature is in Kantian and virtue ethics. We must pick between the two, and since virtue ethics is the minority opinion among academic philosophers, we default to Kantianism, leaving virtue ethics for later research (Hursthouse & Pettigrove, 2018). Kant
6
is also a good choice because he is often considered the central figure of post-medieval philosophy (Rohlf, 2020). His ethical theory dates from the late 1700s.
Second: the Categorial Imperative judges an action on whether or not it is universalizable and whether it treats people as ends and not mere means, that is to say whether it puts value on the dignity of each individual person. In online vigilantism, actions are, more or less, universalized. A single tweet by Sacco was retweeted almost five thousand times and seen by at least tens of thousands of people (Vingiano, 2013). Meanwhile, online vigilantism intends to use the target’s punishment as a means of enacting social change.
The target’s own personal improvement is, sometimes, not an end in itself. For example, Wynn’s critics called not for her to be reformed, but rather deplatformed (that is, silenced online) via what Wynn called a “circular firing squad” among her audience, which is generally sympathetic to LGBTQ+ social justice issues. [Fig 4.] Online vigilantism can (often intentionally) ruin its target’s dignity – and that of her associates.
Third: Korsgaard’s Kantianism is grounded in a meta-ethics which can also ground much of the relevant human-computer interaction theory. This meta-ethics is Wittgenstein’s account of linguistic obligation and language games which are central to his private language argument. This paper ties ethics into human-computer interaction theory in a both relevant and profound way via Wittegenstein’s meta-ethics.
Korsgaard’s Kantianism, like all moral theories, is debatable. For this paper, we will apply what philosophers call the principle of generosity. We will use the best interpretation of Korsgaard’s arguments. We will assume Korsgaard’s Kantianism to be correct until the end of the paper, when we address its weaknesses and suggest avenues for future research.
3.3 Building the Methodology
In this section, we describe Korsgaard’s Kantianism, how it relates to human-computer interaction theory, and what methodology evolves from their combination. This section assumes no knowledge of Kant. We will not attempt to explain every aspect of the derivation or meta-ethics of Korsgaard’s Kantianism (which could, and has been, debated endlessly by an untold number of authors), but rather just its parts relevant to human-computer interaction theory. To do this I am indebted to Guyer (2006) and Scruton (1982; 1996).
It is easiest to begin with a quick description of the Categorical Imperative. It has three formulations, which Kant believed were equivalent in meaning, but different in wording and emphasis. For the moment, we state only the first formulation. Kant instructs you to follow this rule: I should always act so that I could will my action’s maxim to be a universal law (my translation; Kant 1785). A maxim is the reason you do the action – boiled down to its ultimate aim (Timmermann, 2000). Therefore, our methodology includes testing whether the maxim behind a shamer’s action (retweet, post, injurious message, etc.) could be willed to be a universal law.
So, for a non-shaming but hopefully very clear example: you probably avoid driving on the left side of the road not only because Swedish (or US, or any right-side driving country’s) law tells you to; you do it because you do not want to crash and kill yourself; that is your real maxim. Imagine, however, for a moment, that you were in a traffic jam, and so you drove on the left side of the road with a different maxim: “I want to go much faster”. If this
7
maxim were a universal law, then everyone in the traffic jam would drive on the left side of the road for the same reason – to go much faster. In this case, the left side of the road would be crowded, and no one could go much faster. The maxim, “I drive on the left side of the road in a traffic jam because I want to go much faster” therefore, would yield a logical contradiction – this is the key point. Logical contradictions are impossible in reality. No human can possibly believe X and not X to be true; it is a rule of logic that we simply know a priori, just like 1+1=2. So, this maxim could not be willed to be a universal law. Kantians believe that the Categorical Imperative is baked into the structure of cognition itself – and by extension, distributed cognition.
Kant’s starting point, which we will not derive because the derivation is irrelevant to our analysis, is that the most valuable thing in the world is a person’s good will, that is, her desire to act in a manner consistent with a free conscience, and consistent with the duties laid out by the Categorical Imperative. (Korsgaard, 2016b, p. 12) Dignity is exercise of the good will. (Kant, 1785). The Categorical Imperative is categorical, because it applies to all persons at all times. It is an imperative, because it commands you how to act. You might notice that it resembles, in form, the Christian Golden Rule: do unto others as you would have them do unto you (Matthew 7:12). The Categorical Imperative differs from the Golden Rule, in that if a person would have others hurt him, Kant does not prescribe him hurting others. Kant believed that the Categorical Imperative is a form of moral rationality understood by all humans a priori – like mathematics, or the laws of logic (Kant, 1785) – or image schemas in distributed cognition, as described by Benyon (2019, p. 558).
If human beings have dignity inherent in their value as persons, then to deny them this dignity by constraining their moral choices would be to destroy something of moral value, which is the definition of doing a bad thing. Therefore, we have an obligation always to allow people dignity. This is what Kant calls treating people as ends rather than as means.
So, you are obligated to “act in such a way that you always treat people – whether yourself, or someone else – always as an end in themselves, and never merely as a means.”
(my translation; Kant 1785) This is the Categorical Imperative’s second formulation. So, when evaluating the ethics of a case of shaming, we have to consider whether the target (or perhaps shamer) is being treated as an end in herself, or as a mere means; that is to say, whether the target’s dignity is being preserved during the shaming.
The moral law is “a piece of universal legislation, which binds rational beings equally”
(Scruton, 1982). Maximizing morality involves everyone acting in accordance with these formulations of the Categorical Imperative. When everyone is treated as an end, and never as a mere means, we have created a community which is a “Kingdom of Ends”, in Kant’s terms. “Therefore, a reasonable being must act as if through its maxims, it is always a law-making member of a general Kingdom of Ends.” (my translation; Kant 1785) This is the Categorical Imperative’s third formulation. Although Kant probably never envisioned the Internet, I will argue that it is a community which can be a Kingdom of Ends. When evaluating the ethics of a case of shaming, we have to consider whether all shamers and targets are following rules that everyone could follow, furthering a Kingdom of Ends.
Now that we understand the three formulations of the Categorical Imperative, we can dive more deeply into the meta-ethics behind them. You may have noticed that Kant has
8
made a conceptual jump that neither he nor I have explained very clearly. The jump is that the Categorial Imperative is not just a law that we must adhere to, but a universal law, that is to say, one which applies to all people at all times. This is where Korsgaard comes in, to explain this jump.
Here is how she argues it. Korsgaard (1996) first assumes that the idea that the law is not universal -- that is to say, it does not apply to all people at all times, but rather to a limited group of people with whom one “identifies”, say, all men or all women, or all members of a religion, or all Umeå University affiliates. These identities are all contingent.
One could, in theory, stop identifying as a man or woman, or could leave one’s religion, or Umeå University. A law that only applies to a contingent identity would not fulfil Kant’s project, which is to find a moral rule binding on all people at all times.
There is one identity, however, that is absolutely not contingent: you are a human being. You cannot stop being a human being, unless you die. Human beings, crucially, are self-reflective. We can turn our perception to our own mental activity. We can think about our thinking. This means that we can question whether any possible action is right.
...I desire and I find myself with a powerful impulse to act. But I back up and bring that impulse into view and then I have a certain distance. Now the impulse doesn’t dominate me and now I have a problem. Shall I act? Is this desire really a reason to act? (Korsgaard, 1996, p. 93).
Because you are human, Korgaard then argues, you need to have some practical conception of your identity, or you would have no reason for action. If you did not – because your consciousness is reflective – then you would then never act at all.
Since you cannot act without reasons and your humanity is the source of all your reasons, you must value your humanity if you are to act at all. It follows from this argument that human beings are valuable. Enlightenment morality is true (p.
83) (emphasis mine).
What, in particular, is valuable about human beings is their good will. Therefore, any case of shaming that causes the target (or shamer) to lose his dignity is morally suspect.
Reflectivity is crucial to acting according to conscience. External circumstances can threaten this. For example, the spiral of silence can someone to lose the ability to act according to conscience. “[P]eople’s natural fear of isolation from a social group often leads them to outwardly agree with a dominant public opinion, statement, or norm. This is true even if the same individual rejects the opinion internally.” (Huffman, 2016, p. 9) If a target’s associates are punished along with the target, then any sign of solidarity with the target will induce punishment. The mass media, according to Noelle-Neumann (1993), “more than any court, take it upon themselves” to select which issues matter to the general public (p.
153). “Today, the mass media includes social media services,” suggests Huffman (2016, p.
9). Those caught in the spiral of silence cannot express the rational reflection of their conscience. This is both psychologically distressing (producing cognitive dissonance) and
9
immoral, from a Kantian perspective, as they might be lying, which is always wrong.* Kant sees shaming-induced disgrace as socially forced silence. Sussman (2008) clarifies his position.
When we are disgraced, our freedom is indeed diminished, in that we lose a certain power to act regardless of our own “inner” attitudes or acceptance.
Unlike the merely insulted, the disgraced are no longer able to do something that they may be sometimes morally required to do, viz protest against the disrespectful treatment they receive from others.
It follows that any case of shaming that deprives someone of her ability to act according to conscience is morally suspect.
A shaming target can also lose the ability to act according to conscience when surrounded by an overwhelming topology and volatility in blended spaces, which “bring together physical, digital, informational, and social spaces and establish new conceptual spaces where people form intentions, make meanings, have sensations and feelings, and take actions.” (Benyon, 2019. p. 214) In a moment, we will have more to say about blended space. For now, we can understand the philosophical underpinnings of blended space, and many other human-computer interaction concepts, in Korsgaard’s defense of the Categorical Imperative against an objection from the skeptic-about-other-minds. This objection – repeated for the last 250 years – holds that Kant is not clear about why valuing your humanity means valuing that of others. That is because scholars have not come to agreement that Kant succeeds in proving that other minds exist at all. One prominent critic here is Aune (1991), who says that Kant’s model of consciousness is “essentially solipsistic.”
So, Korsgaard turns to Wittgenstein’s famous private language argument, which, although hotly debated, is one of the premiere arguments for the existence of other minds in contemporary philosophy (Scruton, 1996, p.53).
Wittgenstein laid out his private language argument in the mid-20th century. He begins by defining a private language as one which refers to a sensation a person experiences, but which is otherwise incommunicable. Imagine a sensation that a person calls “S”. There is no way she can describe S using a public language, for example as “painful” or “red” or
“common”. It is completely unique from all publicly discussed sensations. Wittgenstein’s argument is that it is impossible for her to meaningfully call that sensation “S”. As Korsgaard reads Wittgenstein, language requires two minds: one who defines X to mean Y, and another mind to obey that rule. Otherwise, language has no meaning.
If what you call S is just that sensation which makes you feel like saying “S”, and it cannot be identified in any other way, then you cannot be wrong. … I impress upon myself that “S” will be the name for this sensation. [This] can only mean:
this process brings it about that I remember the connection right in the future.
But in the present case I have no criterion of correctness. One would like to say:
* For space reasons, I will not include Kant’s argument for the immorality of lying. See Carson (2010).
10
whatever is going to seem right to me is right. And that only means that here we cannot talk about “right” (from Korsgaard, 1996, p. 137).
If private languages are impossible, then language is a public phenomenon. This means a public must exist – that is to say, other minds share the language.
If you are having trouble understanding the private language argument immediately, that is normal. Many philosophers agree that it needs to be re-read or re-explained a few times before it clicks (Scruton, 1996, p. 52). This is also my experience.
In any case, if the private language argument is correct, then reasons for action are also public. “[T]o say that R is a reason for A is to say that one would do A because of R; and this requires two, a legislator to lay it down, and a citizen to obey” (Korsgaard, 1996, p. 138). A shamer cannot simply say, “I am targeting this person for my own private reason R, impossible to explain to others”. Reasons exist in a web, stretched out over time.
Furthermore, when we speak (or write), we “obligate”, in Korsgaard’s words, others to think something. “If I say to you, ‘Picture a yellow spot!’ you will,” she notes. We can also obligate others to consider reasons – maxims – for action.
If I call out your name, I make you stop in your tracks. … [I]f you walk on, you will be ignoring me and slighting me. … By calling out your name, I have obligated you. I have given you a reason to stop. Of course that’s overstated: you don’t have to stop. … But … in ordinary circumstances, you will feel like giving me [a reason] back. “Sorry, I must run, I’m late for an appointment.” (p. 140)
As we trade information, reasons, and responsibility, we engage in what Wittgenstein calls
“language games”, which are “social shared linguistic practices ‘consisting of language and the actions into which it is woven.’” (Dourish, 2004, p. 123) Linguistic obligation and language games can be used to model much relevant human-computer interaction theory.
In the following paragraphs we will illuminate Korsgaard and Wittgenstein’s contributions to distributed cognition, embodied interaction, metaphors, persuasive technologies, and, finally again, blended spaces. Each of these has an implication for our ethical analysis.
During distributed cognition, a web of minds, artefacts, and devices obligate each other to think and reason (a simple definition of cognition from Benyon (2019, p. 546). If a shamer has a reason, then it somehow must be semantically related to a web of reasons communicated among fellow shamers. Since reasons are tied to moral responsibility, online shamers share moral responsibility. Therefore: to analyze the morality of a case of shaming, we must take into consideration how responsibility is distributed amongst the shamers.
During embodied interaction, a computing environment is “grounded in and emerging out of everyday, mundane experience” (Dourish, 2004, p.125). The verbal is the embodied.
The shamer who says, “feel X”, obligates the target to feel X, or some other kind of reaction – in any case, an emotion, which is felt in the body. Online punishment therefore leads to physical manifestation. As Wynn (2020) described the feeling of being cancelled:
I'm not allowed to get angry. I'm not allowed to show pain. I'm not allowed to get defensive, and I'm not allowed to lash out. All I'm allowed to do is go totally numb on the inside as I try to frantically calculate the ideal public relations
11
response that pays due deference to the valid concerns of these poor marginalized people, all the while ignoring the tsunami of verbal abuse that's crashing over me A shaming campaign affects the target’s mundane experience. Sacco and Persky, for example, could not keep their jobs. Therefore: the actual emotional and physical experiences caused by a case of shaming matter to its moral value. I am not saying here that a case of shaming can be judged by its consequences; that would be consequentialism, which Kant strenuously argued against. What I am arguing is that shaming is a process, not a single instance of action. It plays out over time and involves many agents, each of whom is affected in an embodied way and responds to that effect. Cases of shaming set precedents for future cases. When evaluating the morality of a case of shaming, you must try to apply the Categorical Imperative to each step in the sequence of events surrounding it, and think about the precedent it sets.
Metaphors as they appear in human-computer interaction theory are a language game played between the computer’s designer and the user. Social media branding contains much metaphor; Facebook presents itself as an online version of pre-internet printed school facebooks, which contained a picture of all students along with information about them. Twitter’s metaphor is a tweeting bird among a crowd of birds. A predecessor to social media is the bulletin board system, a metaphor for a place to post messages online. Social media platform creators and users are accepting rules to a game. The game is that the platform’s branding linguistically obligates the user to think about the platform as an intimate metaphor, rather than an Internet megaphone. Human-computer interaction philosopher Blackwell (2006) raises alarms about metaphors in user interfaces, pointing out numerous similar examples that “cast doubt on the common engineering perception that an instructional metaphor might be retro-fitted as a communication channel that explains existing system functionality, rather than being the core of the design.”
The presentation of social media as metaphor therefore obscures the actual impact of posting. If the user confuses the metaphor with what it actually represents, he may think that his publication is less visible than it is. Facebook and Twitter are accessible to the entire world, not only to the people on a college campus who own a physical facebook, or to the small area of a forest where a bird’s tweet might be heard. Most social media companies – nowadays, at least – have built virality into their business model. There is a profit incentive to encourage shaming. Sam Biddle openly acknowledged that as his motivation, as we shall see. One way that the companies drive the virality is by employing deceptive metaphors.
So: if a shamer insults the target in a real-life conversation over a dinner table with one listener, that shamer linguistically obligates only his dinner partner to experience the insult. However, if the shamer posts the insult on social media or a website, he obligates every person who sees it to experience the insult. In major shaming cases – like Sacco’s, where #HasJustineLandedYet became the number one hashtag on Twitter – literally millions of people can be obligated to experience a punisher’s message. To call it a “tweet”
is to use a poor metaphor; a better metaphor is speaking into a microphone at a huge, packed stadium. Shamers may know this intellectually, but the benign metaphorical presentation of Twitter or Facebook provides cover for their power – even, perhaps, to them. Although after his conviction Turner was linguistically obligated to being presented
12
as evil in the minds of probably millions of people, “it appeared to the students, and to many in the public, that Turner was being treated with unearned deference” (emphasis mine) because his prison sentence was for six months (Rennison, 2017). I am not making a claim that his shaming and his sentence were morally unjustified. What I am saying is that the common statement that Turner was not punished very severely is open to question.
Although his prison sentence was short, his online shaming is uncountably vast.
The target can himself become a metaphor; Turner has become a metaphor for rape.
His picture stands beside the definition of rape in a criminology textbook. Therefore: to properly evaluate the morality of a shaming case, we must take into account how the metaphor the social media employs does (or does not) deceive users into misunderstanding the obligative power of their posts. Furthermore, we may need to evaluate if and how the target has become a public metaphor for the perceived injustice.
Persuasive technologies, as intended by one of their father figures Fogg, engender mass interpersonal persuasion, without coercion (Fogg, 2008). They can linguistically obligate a user to reconsider an online action. When we discuss engineering solutions, we will include the option of a persuasive technology offering shaming warnings. In analyzing a shaming case, we can ask if technology was used to try to combat the shaming.
Blended spaces obligate a target to re-experience shaming throughout physical space.
A physical place is “deliberately integrated in a close-knit way with a digital space” to create a conceptual space which is very much a real user experience (Benyon, 2019). “The spatial geometries of the local space continue coherently across the distributed boundary into the remote site, providing the illusion of a single unified space,” explain O’hara et al. (2011).
Online vigilantes exact punishment by putting insults and threats into as ubiquitous a physical space as possible, even to the point of, for example, encouraging people to visit the homes of the targets. The target’s whole environment can become a blended space of online and offline punishment, impossible to escape. In Benyon’s (2019) terminology,
“topologically” the target’s punishment surrounds and overwhelms him, and is “volatile”
and ever-changing. For example: Stanford University erected a memorial of the Turner case on the spot of the assault, a metal plaque bearing a quote from Miller’s online victim impact statement, which calls for Turner’s punishment, and which itself had gone viral online (Bae, 2020). [Fig. 5] Miller’s book is sold in stores throughout the nation. It is not just Turner who is overwhelmed. Such a high-profile case of gender-based violence, publicized ubiquitously, risks retraumatizing victim-survivors. Therefore, our moral tools include considering how shaming is embedded in the blended space of physical and online environment. Ubiquitous embedding can emotionally overwhelm a target, or victim- survivor, so that he has trouble reasoning morally, and therefore becomes unable to act according to good will.
This concludes our presentation of Korsgaard’s Kantian ethical theory. We looked at how it is related to human-computer interaction concepts via Wittgenstein’s private language argument. From this, we derived useful conceptual tools for designers to evaluate online shaming. In Section 5, we will use these tools to evaluate arguments from other traditions against and for online shaming. First, however, we need to learn what exactly those arguments are, and also more about the empirical data we have on shaming.
13
4. Related Research
For most of this paper, I have been primarily using the term online shaming, but in this section, I will mostly be using online vigilantism. That is because the authors I cite here are discussing vigilantism specifically, emphasizing its extrajudicial nature.
How is our research question positioned with regards to this related research? Our research question is: what tools can we develop to help human-computer interaction designers build systems that are ethical with regards to online shaming? Our related research provides data to answer this question directly. First, we present empirical findings about characteristics of online shaming. Then, we look at how academic philosophers so far have evaluated the ethics of online shaming. All this data will then be ripe for application of our research methodology, Korsgaard’s Kantianism.
4.1 Empirical Observations about Online Shaming
4.1.1 Norm Enforcement, Anonymity, and Virtue Signaling
Gallardo and Trottier explain how online vigilantism can involve norm enforcement, anonymity, and virtue signaling. Norm enforcement is the vigilante’s ostensible attempt to punish the target in order to promote a social norm that the target has broken.
“[I]ndividuals seek to correct behavior that does not comply with the perceived norm, or rule of conduct in society” (Gallardo, 2016, p. 726). Trottier (2016) describes a similar concept to norm enforcement, but with a moral layer: online vigilantism “seeks to identify and shame unmitigated evil in embodied contexts … [it] is similar to cyber-bullying [but]
framed in terms of a moral compass.” The norm is a just world that the vigilante imagines.
Vigilantes can use anonymity to escape being held responsible for threatening behavior. “The anonymity that the Internet provides enables some individuals to dodge accountability and any kind of tangible fallout from derisive statements they make about other individuals online.” (Gallardo, 2016, p. 728) Is not just assignment of responsibility, but real-world backlash, that the vigilante may try to avoid through anonymity.
Virtue signaling, in contrast, is participating in online vigilantism un-anonymously, so that other internet users think the vigilante stands up for a generally-accepted social norm.
Gallardo does not actually use the term virtue signaling, but it is widely used elsewhere.
(Jordan & Rand, 2019) “By reprimanding offensive behavior, individuals can effectively and easily signal to others that they are trustworthy because they punished a norm violator”
(Gallardo, 2016).
4.1.2 Psychological Characteristics of Online Vigilantes
Chang and Poon (2016) attempt to describe some of the psychological characteristics of people likely to engage in online vigilantism. A survey of Chinese online vigilantes showed that they tend to: 1) perceive the criminal justice system as ineffective; 2) possess high “self- efficacy” – that is, “online disinhibition”, wherein they are “more likely to express negative emotions, harsh criticism, anger, hatred, and threats” online than in real life; and
3) perceive online vigilantism to be a means to social justice.
14
Chang and Poon hypothesize that online vigilantes possess strong psychological empowerment: a “personal belief in control” combined with “involvement in community activities”. This leads to a snowballing cycle of achieving a goal, perception of control, and further participation in online vigilantism.
4.2 Other Philosophers’ View of Online Shaming
There is very little academic philosophical literature on the ethics of online shaming.
Billingham and Parr provide the only real comprehensive analysis from a normative ethical tradition, social contract theory. We want to draw from as much published wisdom as possible, though, so we will consider a few cogent arguments from non-peer-reviewed philosophical sources, namely Wynn, R.Dave, and Goguen. Their analyses stand on their own merits, and are worthy of attention.
4.2.1 The Stages of Canceling
Wynn (2020) breaks down the various stages of the canceling version of online vigilantism, all of which she finds ethically problematic. All quotes that follow in this section are from Canceling.
1) Presumption of guilt. Online vigilantism skips the process of hearing the target’s side of the story. Wynn locates presumption of guilt in politically progressive circles where the slogan “believe victims” has currency. “It’s a norm that was put into place in progressive spaces because out in the world at large, people generally don't believe victims.” Online vigilantism ignores what in law is known as due process for the accused.
2) Abstraction. Online vigilantes abstract out broadly negative general claims about the target from specific details. Wynn points out “beauty guru” James Charles, who was accused of “trying to trick straight men into thinking they’re gay.” Regardless of this claim’s unlikeliness, an online vigilante quickly rewrote it as “James Charles is a sexual predator.”
The changing meme involves a “linguistic shift”, says Wynn, wherein “the verb in the sentence is is or to be. … it’s not now his actions we’re criticizing, but his personality.”
Abstraction usually falsely attributes an undesirable general character trait to the target, and is usually thus a form of lying.
3) Essentialism. Nuances of the target’s life and character are lost, and he is reduced to symbolizing an essential vice. Wynn thinks that “sexual predator” is not a fair summary of the original, already doubtful, accusation against Charles. Essentialism “happened instantaneously” and it “dominated the conversation in a community of millions of people for weeks.” Charles was defenseless against it. With full humanity denied, targets become a metaphor for a certain type of norm violator.
4) Pseudo-Moralism or Pseudo-Intellectualism. Some online vigilantes might say that they just want an apology, criticism, or for the person to be reformed, but this is a smoke screen for attacking a person. “[A] lot of my YouTube friends were like, ‘Did you see James Charles was canceled? L-O-L-O-L-O-L-O-L,” Wynn recalls.
5) No forgiveness. Charles posted an apology which Wynn says “could not be better”, but “[c]ancellers will often dismiss an apology as insincere, no matter how convincingly …
15
delivered. … [as] further proof of what a Machiavellian psychopath you really are.” (Note:
not all apologies are, of course, sincere.)
6) The Transitive Property of Cancellation. A target’s associates are punished along with the target. Wynn’s observed that she became “radioactive” to her colleagues.
4.2.2 A Kantian View (R.Dave)
In a thorough review of the literature, I found only one single Kantian analysis of online vigilantism. This analysis appears in a blog comment on a feminist philosophy website. It is not peer-reviewed, but it is literally all we have, and it raises an important argument.
The commenter, with the handle R.Dave, sees online vigilantism as treating persons as a mere means, and so violates formulation two of the Categorical Imperative. Vigilantism uses targets as mere means “by punishing them to a greater extent than their individual actions warrant in order to achieve a broader goal of societal modification.” (Goguen, 2013) 4.2.3 A View from Social Contract Theory
Billingham and Parr’s (2019) analysis is the most detailed ethical breakdown of online vigilantism in the academic literature, and is drawn from Locke’s social contract theory.
They find that Locke can come down in favor of online vigilantism, but with many complications. Shaming makes an offender aware he has violated a social norm; deters others from violating it; demonstrates solidarity with those wronged; and promotes public expression of what is right. “Ideally,” Billingham and Parr write, the target will “recognize that she has acted wrongly, feel remorse, apologize, seek to make appropriate amends, and commit to complying with the norm in future.” That is the ideal. As we have seen from our three case studies, however, such outcomes sometimes fail to occur.
Locke (as did Kant) argued that anyone who punishes another should always be accountable. The shamer must therefore also be able to held to account. “The decentralized nature of online public shaming,” however, “means that no particular participant can be held to account for the full extent of an instance of shaming” (Billigham & Parr, 2019). Since with distributed cognition comes distributed responsibility, an accountability threshold is eventually crossed. Punishment by individuals scales up to shaming when the sheer mass of shamers makes determining anyone’s relative accountability practically impossible.
Here are Billingham and Parr’s (2019) three complications.
1. A group of potential vigilantes can disagree about whether a social norm has been violated. “This is because the nature of an individual’s conduct is sometimes unclear, such as in the case of Sacco, whose comments were intended as ironic.”
2. Informal punishment (sometimes like formal) tends to favor socially advantaged individuals. It is “likely to reflect racist, sexist, and classist biases, and so be more targeted against members of marginalized groups” who do not have the resources to defend themselves.
3. Disproportionate punishment. The pain inflicted by online shaming could be much greater than deserved. “We hope that an increased awareness of the risks of disproportionality will lead people to be more hesitant to engage in online public shaming. (ibid)”
16 4.2.4 Feminist Views
Feminist views are crucial to our analysis, because the primary ethical defense of online vigilantism can be constructed from a feminist theory of pre-emptive self-defense.
However, no academic work of feminist writing addresses the ethics of online vigilantism as directly and thoroughly as, for example, Billingham and Parr do. We have to piece together the feminist analysis from various sources.
4.2.4.1 For Online Vigilantism
The most direct feminist philosophical defense of online vigilantism is in a blog post by the feminist philosopher Goguen (2013). She sees societal modification as extremely important. Problems like racism, transphobia, or lack of justice for gender-based-violence survivors are massive, she argues, and a person who contributes to these problems – even in a small way – is contributing to a great evil. Therefore, responding to them with massive online vigilantism is actually a proportionate response. She uses this analogy:
If I’m in a boat with another person, and they are bored and carelessly poke a single hole in the boat’s hull, I might be justified in getting mad, but throwing that person overboard (which could possibly do real harm to them) would not be warranted. But if we are in a boat where there are already lots of holes in the hull, and we are in serious danger of sinking, and they poke another hole in the boat (also: they don’t think we will actually sink, and even if I yell at them they will probably do it again), I think I am warranted in throwing them overboard in order to safeguard myself (Goguen, 2013).
The person who writes a racist tweet, or a YouTuber who associates with a transmedicalist, or a sexual assailant who does not apologize, or the judge who under-sentences him, may be perpetuating a great evil, according to this argument. Therefore, that person deserves to be removed from the community, in order to protect it from that evil developing further.
This is, in essence, a utilitarian-feminist stance. Utilitarianism seeks to maximize the greatest utility for the greatest number. The boat hole-poker, or any norm violator, must be removed “in order to safeguard” the punisher and her cohort. This improves the chances of survival – and therefore, utility – for the other sailor or sailors. Not throwing the hole-poker overboard means a bad outcome for everyone.*
D’Amore (2017) takes a related stand, in favor of – or, at least, not opposed to and can be read as romanticizing – what she calls vigilante feminism:
the performance of vigilantism by girls and women who have undertaken their own protection, and the protection of others, against violence—such as sexual assault, abduction, abuse, and trauma—because they have been otherwise failed
* One of my correspondents, Kalle Grill, points out that Goguen’s argument also has a deontological flavor, in that “those who think that morality is about not harming others, rather than about maximizing the good, typically make exceptions to the rule against harming, with self-defense a central exception.”
17
Vigilante feminism finds theoretical backing in some feminists’ approval of preemptive self-defense, which permits someone at danger of domestic violence to kill her partner, 1) even if the threat is not imminent, 2) if she has no means of leaving the relationship, and 3) by using overproportionate violence, like a weapon against an unarmed man. (Hartline, 1997, p. 166-167) D’Amore identifies vigilante feminism in fairy tales and movie characters, who use extrajudicial, often violent, means to punish men who commit gender-based- violence. They “track and kill serial rapist/murderers because law enforcement failed to act on behalf of the victims.” Vigilante feminism “does not dismantle patriarchy, but rather uses patriarchal means to undergrid its own takeover. (D’Amore, 2017)”
D’Amore defends this violence by illustrating how serious the problem of gender- based-violence is. She quotes feminist popular author Valenti (2017), who describes:
a constant fear of rape (conscious or not), [when] women do things throughout the day to protect themselves. Whether it’s carrying our keys in our hands as we walk home, locking our car doors as soon as we get in, or not walking down certain streets, we take precautions.... It’s essentially like living in a prison—all the time (Valenti, 2017, p. 63, quoted in D’Amore, 2017)
Responding to this ever-pervasive fear of rape warrants justice “by any means necessary”
(White & Rastogi, 2009).
Women using vigilante violence to right the wrong of gender-based-violence does not only occur in fiction. Jane (2016) points to the case of female gangs in India. “[T]he major targets … are corrupt officials and violent, immoral husbands: ‘their activities range from beating up men who abuse their wives to shaming officials with whatever weapons are available including walking sticks, iron rods, axes, and even cricket bats’” (White and Rastogi, 2009, p.313) Jane says that this is “not representative of ‘irrational, spontaneous mob violence’ but of ‘grassroots feminist sensibilities’ which offer ‘psychological, social, and justice-related assistance to … women who have been failed … by the judiciary system.’”
4.2.3.2 Against Online Vigilantism
The major feminist argument directly against online vigilantism comes from Powell (2015).
She defends gender-based violence victim-survivors telling their stories on social media, without necessarily naming abusers. Powell calls for “technosociality”, whereby informal justice can be carried out online. “[T]echnologies become embedded in our experiences of the social world at the same time as they contribute to new social and cultural practices.”
Along with the usual concerns about due process, attacks on dignity, and disproportionate punishment, Powell points out the potential for catastrophic outcomes from online vigilantism.
High profile cases of rape victim suicides following abuse on social media (such as Rehtaeh Parsons and Audrie Pott) are tragic examples of the extent of the additional harm and trauma experienced by victims when the evidence of an assault never goes away—and when the response online via social media and the public sphere is all too often negative and victim-blaming (Powell, 2015).
18
Justice does not involve “focused action against accused perpetrators”, but rather victim- survivors “voicing personal narratives of sexual violence to be acknowledged by a trusted audience”. Justice means “information, participation, voice, validation, control, and offender accountability,” and Powell hopes that “social media, blogs, and other online communications increasingly [mediate] informal justice for rape.” This would require more research into how informal justice functions. “[T]here is a comparative underdevelopment within theoretical criminology regarding the nature and mechanisms of such practices.”
In 2019, Miller showed the power of the personal narrative view of justice, when she went public as the victim-survivor in the Turner trial. She, like Goguen, saw her case as one instantiation of a wide-ranging problem, and therefore continued to feel that Turner’s sentence was too short, and that Persky’s recall was justified. “I just want people to know that it’s not happening in a vacuum, in the neat perimeter of college campuses. It is not young people being too silly and too reckless,” she told CNN. “There are greater patterns of male sexual entitlement playing out.” (Sangal, 2019)
Miller does not defend shaming, or vigilantism, as such. However, going public and publishing a bestseller about the case necessarily had ramifications for Turner and Persky.
It turned a harsh public lens toward them once more. Meanwhile, however, Miller also took control of the narrative of the case. Instead of public discussion of who Turner and Persky are, what they did, and whether or not they were properly punished, her book addresses who Miller is, what she experienced, and how gender-based-violence can be prevented and how victim-survivors can be supported. She has thereby raised awareness of gender-based- violence. Is this vigilantism? Not necessarily, according to our definition – Miller may not intend to further punish Turner and Persky in order to achieve societal modification. Their punishment is a by-product, although one might argue that they punished themselves in their original actions. (Kantians will have something to say about this later.)
5. Data Analysis
5.1 Kantian Responses to the Various Arguments For and Against Online Shaming
We now look at the various arguments against and for online shaming in 4.2 above, and discuss how Kantians might respond. We will use that response to try to develop a set of ethical guidelines for human-computer interaction designers to follow. As before, these new tools for you will be in underlined. In this section, I present a radically new conception of the moral obligations of the human-computer interaction designer – obligations that may seem impossible to fulfil. Until today, the platforms where shaming take place, such as Twitter and Facebook, engage in very little content regulation. They might defend this as a free speech policy, but Kantians would retort that by not stopping some online shaming, these platforms are complicit in its spread. Bear with me, because later, I will suggest some technologies that might make fulfilling these obligations possible.
19 5.1.1 Response to Wynn
Wynn (2020) described six morally problematic stages in online vigilantism: presumption of guilt, abstraction, essentialism, pseudo-moralism or pseudo-intellectualism, no forgiveness, and the transitive property of cancellation. Let us address each.
Presumption of guilt. Presuming a person’s guilt – not giving them some kind of due process in justice – involves failing to consider the whole context of their perceived unjust action. It denies the target the chance to comment on the case, and perhaps establish her innocence. Every justice system has its flaws, and due process is meant to try to ensure that innocent people are not punished, whether that punishment is capricious, abusive, or just scapegoating. Presuming guilt therefore uses the target as a means to an end, and violates the second formulation of the Categorical Imperative. It follows for human-computer interaction designers that any shaming you allow should only be targeted at a person whose guilt has been proven. How much proof is required? That is a topic for future research.
Abstraction. As pointed out above, abstraction is usually a form of lying, which Kant forbids. Therefore, your platform as a designer should find a way to prevent people from abstracting out false character traits about a person based on a few details about them.
Essentialism. Essentializing a target turns her into a metaphor for a particular kind of norm violator. The person’s name, picture, and so forth are then being used as vocabulary elements in a language game (Sacco for racist, Wynn for transmedicalist, Turner for rapist).
If someone consents to being used as a metaphor, this poses no ethical problem because it is a free choice. For example, Liszt might have consented to be a metaphor for “virtuoso piano player” (which he is, among pianists). Without consent, however, the person is being used as a means to an end, and we have a violation of the second formulation. It follows that people should only be used in your system as a metaphor when they have given consent to be used that way.
Pseudo-moralism or pseudo-intellectualism. Another form of lying. You need to find a way for your platform to prevent fake demands for apologies, criticism, or reform.
No forgiveness. Kant was clear that he believed repentant wrongdoers should be forgiven (Kant, 1797). When we forgive a wrongdoer – when we overcome negative emotions toward her, such as vindictiveness, anger, and resentment – we cease obligating those emotions in our interaction with that person. When the wrongdoer does not feel those emotions, she is more likely to be able to reflect rationally on her actions and thereby grow in moral conscience. Maintaining negative emotions in this case would, conversely, deprive the target of the ability to act according to conscience. Therefore, your platform must have some mechanism for preventing automatic dismissals of a target’s apologies – if they are genuine apologies, that is to say, not lies. (Victim-survivors of gender-based-violence report that they commonly receive insincere apologies from assailants (Marsh & Wager, 2015).)
The Transitive Property of Cancellation. We have seen how we must evaluate the how shaming is embedded in a physical environment. When the transitive property applies, associates of the target are shamed. The emotion of shame is obligated across a community.
The transitive property is really the trickiest of Wynn’s concepts to analyze, because in real-life cases, people have felt very justified in shaming associates of targets. The Turner case provides a clear example. During the trial, 39 people wrote character statements for
20
the court supporting Turner. Two of them, childhood friend Leslie Rasmussen and high school guidance counselor Kelly Owens, faced public backlash for their letters.
Rassmussen’s rock band was subject to a “viral mob” (Cuevas, 2016) and subsequently dropped from a New York music festival (Levin, 2016). Both women retracted their statements.
If responsibility is distributed among shamers, it is also distributed among norm violators. Meanwhile, if a target becomes a metaphor for a perceived injustice, then so can associates. So, on one hand we have an argument for blaming associates, and another saying that they should only be used as a metaphor with their consent. Our Kantian analysis leaves us in a bind here as to the moral acceptability of the transitive property of cancellation. I will leave this here as a limitation of Kantianism.
This section contained a number of guidelines which seem to require constant human monitoring to enforce, and so are unrealistic for large platforms like Twitter or Facebook.
However, Powell (2015) will proffer a partially automated solution for large platforms.
5.1.2 Response to R.Dave
R.Dave argued that online vigilantism treats the target as a mere means to the end of societal modification, and therefore violates the second formulation of the categorical imperative. Here, I will argue that this is not always the case; that there may be a tipping point at which too many ill-intentioned vigilantes can overwhelm well-meaning ones who are respecting dignity; and whether the target actually turns out being treated as an end in herself is unpredictable and may only be determinable long after the vigilantism stops.
We begin with the question of whether targets are indeed only used as mere means.
This would imply that in no cases of online vigilantism, targets are ends in themselves. To treat someone as an end in herself means to support her freedom to make moral decisions according to her conscience. Is it ever the case that online shamers are doing this?
It is not impossible. Without a doubt, in Sacco, Wynn, and Turner’s cases, some, if not most, shamers were trying simply to restrict the target’s freedoms – freedom from emotional pain, freedom of speech, even freedom of movement or from physical pain. Some vigilantes may, however, have genuinely been attempting to raise the targets’ moral awareness – treating them as an end in itself.
We have defined shaming as an attempt to punish a person. The goal – and effect -- of punishment can absolutely be to raise a person’s moral awareness. Offenders are separated from society in prisons, in part, so they can use the time to reflect upon their actions and realize how they were violations of moral law. So, when a vigilante calls for a target to be ostracized, they may be doing so in the hopes that the target will reform his thinking.
Online vigilantism involves, however, a large amount of minds. A single genuine attempt to raise moral awareness may be surrounded by a dozen messages intended only to attack, linguistically obligating emotional and therefore bodily impacts on the target, preventing the target from reflecting rationally upon his actions. So, any well-meaning vigilante may be drowned out by others who are punishing the target as a mere means.
In any given instance of online vigilantism, then, there is a balance, a tipping point, at which too many vigilantes would be punishing with ill intent, disabling the target from
