From higher education to workplaces, institutions are increasingly adopting data-driven and semiautomated technologies to facilitate, manage, and arbitrate sexual affairs. These largely US-based systems, which I term “technologies of sexual governance,” are encoded with and reify particular ideologies about sexual (mis)conduct, and thus call for a critical feminist inquiry about their cultural, political, and moral implications for advancing a feminist sexual politics. Drawing from Halley et al.’s “governance feminism” framework, this article makes the case that a critical feminist inquiry into technologies of sexual governance must take into account the co-constitutive nature of feminist sexual politics and technology. Specifically, I argue that critical inquiries must begin by interrogating which feminist ideologies about sex and power gain purchase with and through particular computational logics and form. To demonstrate this approach, I offer two ways of reading feminist scholarly and popular responses to “antirape technologies” that capture both readings’ shortcomings, and I propose a third approach that captures the cultural work that particular feminist ideologies and technologies mutually perform. This article concludes by demonstrating how the third approach can advance a feminist analysis of workplace misconduct management softwares.
On November 1, 2018, more than twenty thousand Google workers participated in a walkout across fifty cities to protest the company’s mishandling of sexual harassment allegations (Statt 2018). The walkout was catalyzed by the New York Times article published two weeks prior on Andy Rubin’s exit: in spite of several allegations of sexual misconduct against him, the creator of Android left Google with a generous exit package (Wakabayashi and Benner 2018). The revelations about Rubin, according to the walkout organizers, exemplified a number of systemic issues of sexism and racism that created a toxic workplace culture at Google (Stapleton et al. 2018). The stories shared during the walkout had a common thread of how the company dismissed and silenced misconduct complaints and funneled complainants into opaque, arbitrary, and demanding grievance procedures that would eventually exhaust and disincentivize them (Google Walkout for Real Change 2018).
Informed by these experiences, the walkout organizers’ demands centered on the need for an equal and fair adjudication process, including anonymous reporting, a uniform grievance procedure, and an end to forced arbitration (Stapleton et al. 2018). Following the organizers’ meeting with Google leadership, CEO Sundar Pichai announced a number of policy changes and made a commitment to take sexual harassment issues more seriously (Pichai 2018). One of the policy changes involved the creation of an anonymous and secure reporting platform called Alphabet helpline for employees. Google parent company Alphabet Inc. saw the helpline, developed by a third-party company called NAVEX Global, Inc., specializing in compliance management softwares, as an opportunity to “foster a culture of integrity and ethical decision-making” (Alphabet helpline 2020). The implementation of confidential “open channels of communication” by a third-party entity “completely separate from your organization” would “promote a positive work environment and maximize productivity” (Alphabet helpline 2020).
The walkout organizers’ demand for an anonymous reporting process and the company’s software solution are not new (Sim 2018). The digital helpline is just the latest technological instantiation of what I describe as technologies of sexual governance—from the so-called “antirape technologies” to reporting chatbots, they employ data-driven and automated technologies to document, facilitate, and mediate sexual conduct in public, private, and intimate spheres. They engage heavily with some of the most prevailing feminist ideas about sex and power, including “affirmative consent” and “workplace misconduct.” They operate in political, social, and legal institutions to categorize misconduct, create and communicate codes of conduct, and facilitate reporting and grievance procedures.
As networked and communicative technologies increasingly mediate how users, including those who experience sexual violence, access resources and seek redress, there is a need for an analytical framework that captures the co-constitutive nature of feminist sexual politics and technology. How do technologies of sexual governance come into being? What precisely do these technologies aim to govern and through which logics? To answer this question is to interrogate how particular feminist ideas about sexual governance gain purchase in technological logics and forms. In this article, I make the case that a critical feminist inquiry of technologies of sexual governance must take into consideration both the “feminist” and the “technology” of “feminist technology” as mutually constitutive. I make the case that current popular and scholarly framings of such technologies are inadequate because they imply a one-directional interaction. Taking antirape technologies as an example, I offer two ways of reading feminist responses that capture what is inadequate about both readings, and I propose another analytical approach to challenge and complement both readings. The remaining section of the article applies the third approach to misconduct management technologies.
TOWARD A CRITICAL FEMINIST INQUIRY OF TECHNOLOGIES OF SEXUAL GOVERNANCE
A “cottage industry” of antirape technology
Alphabet Inc. is not alone in datafying and automating workplace grievance procedure for sexual misconduct. From do-it-yourself rape kits (Nashrulla 2019) and wearable and smartphone panic buttons to date rape drug-detecting nail polish (LaVito 2018) and machine learning–driven reporting platforms (Galang 2017), technologies designed to “combat”1 sexual violence have been proliferating in the market for personal safety for nearly a decade. The makers and users range from individuals to organizations across various sectors, including government, finance, technology, and nonprofits. Dubbed “antirape technology,” the examples discussed thus far illustrate the conjoining of data-driven and automated technologies with sexual politics. While they do not all have an explicit focus on rape, I use “antirape technology” to refer to a wide array of data- and technology-driven efforts at sexual violence prevention and response. It is an imperfect, if not crude, term, as if a piece of technology could ever be “antirape,” but I use it for two reasons. First and foremost, I want to underscore how the social imaginaries of rape and technology intertwine to give shape to the design, implementation, and use of such devices and softwares. Second, I use the term because it has been used in popular and scholarly discourses.
This “cottage industry” (New 2014) of personal safety technologies employs various forms of technology and spans jurisdictions. Take, for example, a slew of smartphone reporting apps like LiveSafe and #NotMe that connect users to human resources, private security, or the local police to report a range of safety concerns, including sexual harassment. Chatbots like Botler.ai (https://botler.ai) and Spot (https://talktospot.com) use machine learning to facilitate disclosure, walk through possible legal and nonlegal remedies, and direct users to relevant resources. Other systems look like Callisto (https://mycallisto.org),2 a veteran in the market for sexual harassment reporting systems founded by a survivor of campus sexual assault (TED 2016). Initially designed for use by colleges and universities, its enterprise-facing software similarly employs a matching algorithm to identify a repeat offender, if there is one,3 from submitted reports. More recently, Madison Campbell, a twenty-three-year-old survivor of campus sexual assault, created a do-it-yourself rape kit called #MeTooKit (www.metookit.com)4 and garnered much public scrutiny for taking a misdirected approach to sexual assault response (North 2019).
These systems have traction beyond the United States. Vault Platform (https://vaultplatform.com), for example, is a London-based company that shares much of Callisto’s design and approach, though its focus rests explicitly on workplace misconduct, including non-gender-based forms (Hernbroth 2019). Maru (https://maruchatbot.co/) is a chatbot created by feminist artists, advocates, and engineers from Ghana, Cameroon, Nepal, South Africa, and Germany to offer guidance to those witnessing or experiencing online harassment (Cooper 2020). Omdena (https://omdena.com/projects/sexual-harassment/), created by data scientists, takes a cartographic approach to create an “AI-driven heatmap” of sexual harassment in major Indian cities that identifies safer routes. Some, more than others, engage with the local and national context of the system’s intended user base, but many draw heavily from US-specific language and framing of sexual violence. As critical legal scholars have remarked, the exporting of American legalese, specifically the focus on bureaucratizing sexual harassment adjudication and a carceral approach to remedies, raises concerns as such ideas circulate and gain purchase in different legal, political, and social landscapes of sexual politics.
Two ways of reading feminist responses: Cooptation and inclusion
Feminist reactions to the existence, circulation, and proliferation of these technologies have been varied. In this section, I lay out the two ways of reading feminist reactions to antirape technologies, primarily in the United States. My objective here is to add precision to what exactly is being discussed when we speak of feminist technologies. This precision is important because a nuanced description of the conditions (Davis 2020) and imaginaries (Jasanoff and Kim 2015) through which sex, power, and technology interplay proffers a more robust road map for identifying how and why different bodies are differentially serviced or harmed by emergent technologies, and for challenging norms and policies. It also allows the analysis to be freed from the constricting binaries of technological determinism and technological neutrality, “good” and “bad” feminism. I take this call for precision from feminist legal scholars Halley, Kotiswaran, Rebouché, and Shamir, who insist that critical feminist inquiry must disaggregate describing and assessing “feminist engagement with the law…and its sister institutions” (Halley et al. 2018). Assessing feminist engagement with technologies of sexual governance, then, must begin with a nuanced and precise description of its engagement.
The debate about antirape technologies provides a useful case study here. I offer two ways of reading feminist responses that are intended to provide an analytical scaffold that underscores how feminist and technological discourses conceptualize and articulate the relationship between sex, power, and technology. Specifically, my analysis of both readings aims to underline how each lens conceives the “feminist” and the “technology” of “feminist technology”: the cooptation reading accentuates how technologies of sexual governance control women’s bodies; the inclusion reading seeks an opportunity for feminist intervention in technology design. Both readings offer an important analytical framework, but I argue that they are inadequate because they are principally concerned with the “technology” of feminist technology and thus offer an inadequate critical inquiry of the co-constitutive nature of feminist technology.
Before I introduce the two readings, I should first clarify how I am using “feminist” and “feminism” as descriptors. For purposes of explicating the two readings, I use the term “feminist” politics and “feminism” to refer to various ideologies that share the core concern about how patriarchy, as a system of power, structures institutional, social, and intimate lives and that have a fundamental commitment to collective emancipation from patriarchy. This liberal use of “feminist” is intended to trace what kinds of ideas are put forward in service of feminist critique and world-making.
One way of reading feminist response to antirape technologies is through the cooptation paradigm, which sees such artifacts as emblematic of neoliberal cooptation of feminist ideals. The cooptation paradigm captures the sentiment that the “cottage industry” of antirape technologies reflects the commercial interests of technology makers (Chemaly 2014; Culp-Ressler 2014; Hess 2014) and their fervent belief that technology is a uniquely positioned to solve social ills, which Broussard terms “technochauvinism” (Broussard 2019). Antirape technologies thus emerge from gendered power relations of (male) inventors and (female) users, producers and consumers, saviors and victims, and reveal the makers’ troubling assumptions about sexual violence as a regrettable happenstance of not being vigilant enough. For example, sociologists Rees and White (2012) regard antirape devices as a vehicle of risk negotiation for women against the threat of rape. White and McMillan (2020) echo this point when they “suggest that as these technological responses are situated within, and highly compatible with, neoliberal culture and capitalism, they serve to further privatize and atomize the problem of sexual violence” (1136).
Reading these responses through the cooptation paradigm illuminates important assumptions about sexual politics and technology. In regarding technological instantiations of rape resistance as a regrettable example of neoliberal cooptation of feminism, the cooptation paradigm’s interlocutors imply that feminist interest in rape resistance is being disserved, if not tainted, by the commercial and saviorist motivations of technology makers. This criticism echoes feminist scholars’ critiques of the neoliberal state’s cooptation of feminist resistance against sexual violence (Bumiller 2008), which creates what Nancy Fraser (2013) terms “feminism’s uncanny double.” Following this, technology is a vehicle through which neoliberal and patriarchal systems exert control over women’s bodies and access to public spaces. The cooptation reading thus analyzes the relationship between feminism and technology as one in which the former is being misrepresented by the latter. While this reading is helpful in accentuating the gendered power relations embedded in antirape technologies, its analytical focus is on technology production as a manifestation of the cooptation of feminist sexual politics. In doing so, the cooptation paradigm does not adequately engage with how and why institutions and individuals do give purchase to antirape technologies, and the potentially emancipatory possibilities of technologies are largely overlooked.
The second reading is the inclusion paradigm. This paradigm shares the cooptation paradigm’s diagnosis of antirape technologies as artifacts inculcated in power structures but locates its critique in technology design. The interlocutors of the inclusion reading center their analysis on diversifying and democratizing technology-making as an intervention against the shortcomings of antirape technologies.5 Antirape technologies may be “tone-deaf,” but they are an unsurprising outcome of institutions and people, processes, and artifacts structured by social relations of power. The problem site also becomes an intervention site: had more women and other marginalized subjects been involved in the design of these technologies, antirape technologies would not have been so flagrantly misguided. To the interlocutors of the inclusion paradigm, democratizing, empowering, and improving technology design do raise concerns about legitimating technosolutionism, but, as Lindtner, Bardzell, and Bardzell (2016) suggest, taking a “reflexive-interventionist approach” to challenging technology-making beyond solutionism may offer an opportunity for meaningful world-making. For example, human-computer interaction (HCI) research projects that examine survivors and other vulnerable users’ safety practices identify potential opportunities for practitioners to design for (Lindsay et al. 2013; Matthews et al. 2017). In their survey of antirape apps, communication scholars Bivens and Hasinoff (2017) conclude, “we are hopeful that evidence-based anti-rape apps might have the potential to genuinely help reduce sexual violence” (13). Beaton (2015) cautions that, while antirape technologies have the potential to map onto women’s existing safety practices, the process of making, especially under institutional pressure and commercial motives, may fail to “replicate” such safety practices.
The two readings explained thus far are helpful in amplifying different considerations for analyzing the interplay of sex, power, and technology. They ultimately share the conclusion that positions antirape technologies as artifacts embedded in the underlying capitalistic structures of raced, heteronormative patriarchy. Following this, they see antirape technologies as a well-intentioned, if not misdirected, attempt at solving a fundamentally social problem, and they eschew evaluating the devices’ impact since any potential positive benefits of a technological fix are minimal and individual at best. They arrive at this conclusion by amplifying different points of considerations: the cooptation reading accentuates how technologies of sexual governance appropriate feminist sexual politics; the inclusion reading reframes technology design as an opportunity to meaningfully operationalize feminist sexual politics. Both readings critically analyze the “technology” of feminist technology.
Toward a critical feminist inquiry of technologies of sexual governance
What remains undertheorized is what the “feminist” of feminist technology is doing when sexual politics are instantiated technologically. How do antirape technologies come into being? What are the conditions through which feminist ideas are made tractable to emergent technologies? Which feminist ideas gain purchase in particular computational logics and forms? I take these questions from feminist sociotechnical inquiries like Rentschler’s (2018) examination of white feminist epistemology in harassment prevention app Hollaback (https://www.ihollaback.org/take-action/get-app/) or Quinlan’s (2020) analysis of how the forensics industry gains purchase through sexual assault victim’s advocates and their “techno-optimism.” To pose these questions is to take seriously both the “feminist” and the “technology” of “feminist technology” as objects of analysis. Each of the two readings suggests that feminist sexual politics are being misrepresented by the people (cooptation) and the process (inclusion) of technology-making. The underpinning assumption here is that there exists a self-evident and universally shared feminist sexual politics; “feminism” remains intact as it is being undermined or adulterated by technological appropriations.
This is a troubling assumption, as feminist legal scholars Halley et al. (2018) point out. Their introduction of “governance feminism” lends a useful heuristic for challenging and extending the analytical scaffolds offered by the two readings. Feminist ideas increasingly gain purchase in the political and social mainstream, and Halley and her coauthors observe a troubling development in how particular feminists and feminist ideas “exert a governing will within human affairs” (Halley et al. 2018, ix). Feminist ideas may be rooted in oppositional and emancipatory politics, but something happens when and as they enter and constitute the political, economic, and social mainstream, and these transformations must be reflected on (15). Take, for example, how the initial call to take sexual violence seriously evolved into the carceral agenda (Kim 2020) of antitrafficking campaigns (Bernstein 2010) and intimate partner violence–related policies (Goodmark 2018; Mazingo 2014) that not only fail to underserve those most affected by sexual violence but also give legitimacy to further state violence in the name of feminism and women’s empowerment. As Halley and her coauthors ask, “How do we, as feminists, understand our feminism once we acknowledge our immersion in the uncontainably complex dynamics of law and its sister institutions?” (2018, xvi).
Both readings overlook which feminist ideologies are considered and how they intertwine with antirape technologies; this overlooking mischaracterizes and obfuscates the complex role certain women play in building, using, and legitimizing certain antirape technologies. It may be that earlier iterations of antirape technologies embodied gendered binaries of makers/users, producers/consumers, saviors/victims. However, their lasting endurance and popularity rests in the fact that they are developed for and by women, many self-identifying as victims and survivors of sexual violence. Behind products like Callisto, #MeTooKit, and Botler.ai, among others, are women, often white, highly educated, and entrepreneurial, and with access to financial and social capital to start a company. In all three of these examples’ company storytelling, their founders’ experiences of sexual harassment feature prominently to establish their product’s legitimacy and trustworthiness.6 That these systems and their duplicates gain traction in other parts of the world further challenge existing analytical framework to take into account the role of US imperialism in setting norms and standards for sexual governance through technological innovation (Halley, Kotiswaran, and Rebouché 2019).
In the remainder of this section, I apply the analytical frameworks discussed thus far to the case of LegalFling, a blockchain-based consent app, with the hope that it can concretize earlier discussions. After briefly introducing the app, I describe what the two readings amplify and overlook. I argue that the binary logic of affirmative consent maps easily onto the app’s computational binary rhetoric. Analyzing a particular feminist conception of consent and the internal logic of a blockchain system as mutually reinforced by each other illuminates why advocates and scholars at the intersection of gender and technology should be cautious of quantifying, datafying, and automating intimate affairs.
The short-lived history of a blockchain-based consent app called LegalFling (https://legalfling.io/) is a helpful example of the two paradigms at play. In early 2018, Dutch startup LegalThings launched a website for its prototype app that uses blockchain technology to “secure” and “protect” consent between intimate partners (Grant 2018). After much public outrage, driven by feminist advocates and technologists, the app was eventually removed from the Apple AppStore and GooglePlay due to “mature content.”7 The two readings are helpful in making sense of feminist reactions. The cooptation reading captures outright rejection of LegalFling as “morally wrong” (Powell 2018), an “incredibl[e] exploitatio[n] of women’s fears…some capitalist-commerce fix to the problem of sexual violence” (Ramanathan 2018). Feminists in the cooptation camp condemned the creators’ tone-deaf understanding of consent as a finite object that cannot be rescinded or changed (Cole 2018; Salam 2018). The inclusion paradigm equally rejected LegalFling’s corruption of affirmative consent but attributed it to the vendor’s lack of diversity. In one telling open letter written by a woman (but signed “all of us”), she connected LegalFling’s problematic understanding of consent to its white male creators and concluded her letter: “for the love of god, hire some women” (Vadnal 2018).
The two readings are helpful in identifying the objectification and commercialization of women’s fear (cooptation argument) and the embodied relationship between the process and people of technology design (inclusion argument). But neither fully captures what is so troubling about LegalFling as a technology designed in the name of affirmative consent—its binary articulation of consent as a means to adjudicate desirable sexual conduct. To examine this issue, it is integral to take the app seriously and at face value as a technological instantiation of a particular thread of feminist discourse on consent. Framing the app in this way not only clarifies which feminist ideas of consent are being operationalized but also expands the list of responsible actors beyond creators and users.
It is first important to contextualize affirmative consent, LegalFling’s conceptualization and expression of sexual consent. The concept of affirmative consent emerged as a feminist intervention against the “negative” model of consent that narrowly defined sexual violence as sexual conduct happening with an explicit rejection. Whereas unwanted sexual interaction in intimate partnership or the “gray” scenario of a drunken hookup would qualify as acceptable, if not unlawful, in the traditional model, the affirmative consent standard challenged the incumbent paradigm by foregrounding women’s sexual agency and pleasure (Friedman and Valenti 2019). Today, affirmative consent and its accompanying heuristics (Fischel 2017) of “enthusiastic consent” (to foreground sexual pleasure) and “consent is easy” (to emphasize consent as a learned social practice) have gained legal, political, and social currency as an alternative progressive standard for sexual governance, including in higher education institutions8 and some states.9 Culturally, affirmative consent discourses, as expressed in consent education programs (Appleton and Stiritz 2016), offer a helpful framework for guiding young people’s development that is focused on bodily autonomy, pleasure, and communication. However, some, like Fischel, caution against the affirmative consent discourse for its “conceptual thinness” that conflates “bad sex’” with sexual assault and flattens the important distinction between consent, desire, and pleasure (Fischel 2019). Sociologists Hirsch and Khan similarly caution against the isolation of affirmative consent from the sociocultural conditions from which people might deliberate on and seek sexual interactions, which they term “sexual project” (Hirsch and Khan 2020).
Affirmative consent, for LegalFling, is an object to be obtained and insured. From what remains of LegalFling’s website emerge two rhetorical threads. The app simplifies sexual decision-making into a set of dos and don’ts that demystify “the rules of engagement as well as personal preferences” through “explicit” communication. Complementing simplicity is the messaging around certainty: the app claims to “store, timestamp, guarantee” secured “transactions” that will “remove misunderstandings.” That this rhetoric of simplicity and certainty is expressed through blockchain application is significant. For blockchain applications, as communications scholars Woodall and Ringel (2020) explain, trust is their crucial justification and discursive landscape. Expressed through blockchain, sexual decision-making is reduced to a set of discrete preferences that are “anchored in the properties of the record as evidence rather than subjective judgments of the individual” (2206). LegalFling’s intervention, then, is to simplify and insure sexual decision-making to “remove misunderstandings and prevent unintentional bad situations.”
When this objective is situated in the context of the discourse on affirmative consent that seeks to simplify consent (i.e., “consent is easy”), LegalFling’s instantiation of consent is not surprising. Rather, it shows how feminist efforts to simplify and operationalize sexual decision-making into the affirmative consent framework find an ally in blockchain technology, with the ease and assurance it affords. This reading, when placed together with the cooptation and inclusion readings, shows not only that blockchain technology is an inapt vehicle for sexual decision-making but also that affirmative consent is an inapt framework for sexual project.
A CRITICAL FEMINIST INQUIRY OF HARASSMENT MANAGEMENT SYSTEMS
In this final section, I take complaint management softwares for workplace harassment as a case study to demonstrate how to critically analyze technologies of sexual governance. I begin by situating these systems in the legal and organizational contexts of workplace misconduct management. I apply the two ways of reading these systems and argue that neither offers a satisfactory analysis of the datafication and automation of misconduct management. I argue that these systems embody and stabilize “a top/down, bottom/up model” (Halley et al. 2018, 38) of sexual harassment, which Halley et al. (2018) term “dominance feminism”: they exceptionalizes sexual harm as indicative of male dominance and proffer legal recourse as a technique of rectifying female subordination.
By workplace harassment management systems, I am referring to a wide array of data-driven systems for collecting, organizing, and managing misconduct complaints in the workplace. Traditionally, workplace misconduct management falls under the remit of employers’ human resources (HR) division for both formal and informal procedures and remedies. Digitization and automation of misconduct management introduces third-party vendors to this structure, thereby complicating the expectations of anonymity and liability. Softwares for complaint management facilitate and organize communication between employers (likely case managers from HR) and employees, where the former interact with a web-based dashboard and the latter with web- or smartphone app-based interfaces. To the system vendor, the two parties are not of equal standing: the former is the client, and the latter is the user.10 Unlike the software industry’s familiar “b2b2c model” in which the supplier sells a product to a business that, in turn, sells it to customers, system vendors for misconduct management softwares must negotiate the demands and expectations of both the client and the user.
As the case of Alphabet helpline suggests, reporting systems are intended to serve both the interests of employers and employees, but their incentives and expectations can be at odds. Consider the tension between anonymity and accountability. Anonymity during the reporting procedure is an attractive and a crucial element for both employers and employees. Companies in creative (Griffin 2018), entertainment (Hutchinson 2018), financial (Dishman 2017), and tech industries (Solon 2017) have a vested interest in taking the discourse around #MeToo and the “toxic workplace” seriously as it informs their recruitment. In conjunction with diversity and inclusion initiatives (Panchadar 2018), bias and antiharassment trainings (Thorp-Lancaster 2019), and interest groups for marginalized employees (Tiku 2019) in these industries, complaint management systems represent the proliferation of digitally mediated efforts to rectify the toxic workplace. It is equally attractive for employees. Perceived neutrality and impartiality of data-driven systems created by independent suppliers position reporting platforms as a desirable alternative to going to the HR department. Vendors thus emphasize trust as a key affordance of implementing reporting systems. For example, the CEO of Vault Platform alleges “a 40% uptick in reports in clients” since adoption (Elsesser 2020). A client of Spot, a bot-driven misconduct management system, claimed a 60 percent increase in employee reengagement with cases compared to a hotline (Spot 2020; Moran 2019).
However, the motivations for desiring anonymity may be odds. Complainant employees may desire anonymity for a number of reasons, including self-protection from unsympathetic peers, offending employees, and reputational damage. Some employees may file a complaint to comment on the wider workplace culture, rather than to seek to discipline specific individuals, and find their identity irrelevant to the complaint. Whatever the reason, anonymity reduces some degree of burden complainants face in deliberating whether to report and in seeking help (Waldman 2019). For employers, anonymity can contribute positively by increasing the number or quality of complaints (Liu 2018), but it also obfuscates their responsibility to anonymous complainants (Henry and Powell 2015). Anonymity, then, is not a fixed value; it instead moves differently—and importantly so—through the conditions, motivations, and aspirations of the complaint.11
It is in this complicated web of procedures and expectations that misconduct management systems operate. The two ways of interpreting such technologies apply here again. The cooptation reading accentuates how automation and digitization enable employers to evade responsibility by offering a seemingly novel reporting option that, on a surface level, increases the quantity and quality of complaints. The inclusion reading amplifies how reporting softwares do equip employees with greater control over what grievances they communicate to employers and how they do so. In some cases, where systems explicitly align with employees, the control and resources systems offer to employees can be substantive. Yet neither reading adequately analyzes how misconduct management systems come into being and their differential impact on employers and employees, neither of which are monolithic.
What is lacking in the two readings’ analyses is their critical engagement with the underpinning logics of reporting systems. I bring together literature on sexual governance in conversation with scholarship on media and communication to investigate how assumptions about sex, violence, and power are encoded within and expressed through data-driven systems’ computational logics. I argue that the fixation on formal adjudication of sexual harassment and the visibility politics of information systems co-create reporting softwares that favor legal remedies and foreclose others. In addition, the logic of categorization maps easily onto a binary understanding of workplace harassment as violence done by men against women.
A note on method
The softwares discussed in this article are far from comprehensive. My initial attempt at a systematic survey of harassment management systems had to contend with the reality that many such tools are not publicly accessible due to the caution around discussing sexual harassment on the part of both the vendor and the employer. Seaver (2017) sees these restrictions as an opportunity for ethnographic researchers of technology to become “scavenging ethnographer[s]” (7) as they negotiate and navigate access to various sites of technology production and use. As I am primarily interested in how reporting tools embody and express dominance feminist ideas of sexual harassment, I restricted myself to third-party products with explicit mention of sexual harassment.
Overreliance on legal remedies
Management softwares largely offer two reporting options for workplace misconduct complaints: internally through HR or externally through law enforcement. The conditions and consequences of these recourses are rarely explained. The provision of these two—and only these two—options, when situated in the context of feminist campaigns against sexual harassment, makes much more sense. Since the 1980s, feminist resistance to sexual violence, including workplace harassment, took a decidedly legal characteristic by instrumentalizing the “law and order agenda” (Bernstein 2010) to mobilize victims (Reynolds 2019), punish offenders (Kim 2020), and institutionalize standards and procedures for sexual governance (Halley et al. 2018). Contemporary recognition of sexual violence as a legitimate harm owes much to this history, but it has also contributed to exceptionalizing sexual harm in isolation from other forms of violence (Fischel 2016) and prioritizing legal redress with punitive sanctions as an intervention (Tambe 2018).
Data-driven systems that prefigure employees into two legal reporting options thus embody the broader history of mainstream feminist resistance of sexual violence. Options that do not appear, such as directories of local mental health professionals and contact information of union representatives, further suggest that a more robust understanding of remedies exists outside their purview.
To better understand how vendors position themselves in relation to the two channels of redress, I provide a typology of the softwares discussed so far in table 1.
|Type .||Example .||Features for end users .||Remit .|
|Complaint carrier||Alphabet helpline; iSight; Lighthouse Services||All workplace concerns (e.g., ethics compliance, fraud detection, health and safety violation, sexual misconduct, etc.)|
|Complaint counsel||Callisto Expansion; #NotMe||Bullying, discrimination, and sexual misconduct|
|Complaint intermediary||Spot; Vault Platform||Bullying, discrimination, and sexual misconduct|
|Type .||Example .||Features for end users .||Remit .|
|Complaint carrier||Alphabet helpline; iSight; Lighthouse Services||All workplace concerns (e.g., ethics compliance, fraud detection, health and safety violation, sexual misconduct, etc.)|
|Complaint counsel||Callisto Expansion; #NotMe||Bullying, discrimination, and sexual misconduct|
|Complaint intermediary||Spot; Vault Platform||Bullying, discrimination, and sexual misconduct|
The first model, complaint carrier, is a general case management system. As one product, iSight, describes itself, it is a “software solution” to “respond and resolve quickly.” Sexual harassment complaints compose just one of many workplace ethics and compliance violations. As such, complaint carriers simply deliver reports from users to relevant case managers. They generally provide an open form for users to fill in liberally, or provide basic questions of who, what, when, and where to collect relevant information. They do not tend to elaborate on who will receive reports and what happens to reports after submission.
The second and third models are similar, but crucially different in how they position themselves in relation to the employer. Both models see complaint counsels take it upon themselves to direct employee users to internal (via HR) or external (legal) procedures. In doing so, complaint counsels prefer legal expertise and process to employers’ internal practices. Callisto, for example, directs users to legal option counselors, attorneys who are “selected and vetted by Callisto.” In contrast, complaint intermediaries have greater faith in HR and position themselves as active agents mediating the interaction between employers and employees. Complaint intermediaries also permit multiple incidents to be included in a report and, in the case of Vault Platform, offer some degree of choice in case manager.
As these three types demonstrate, complaint management softwares offer two reporting options. They present legal recourses as preferable and authoritative: procedural changes to legal and political institutions are favored in place of changes at the local level. Other routes of reporting or help-seeking, such as seeking confidential mental health practitioners or, where applicable, speaking to union representatives, do not appear at all in the systems considered in this article. This overreliance on legal redress is stabilized through what media and communication scholar Gillespie (2018) describes as the “politics of visibility” of communicative technologies—the choices communicative systems make to render certain possibilities visible, while erasing others, and their normative impact in shaping users’ expectations. The self-evident logic of data-driven systems thus forecloses the possibility of nonlegal remedies for complainants.
Exceptionalizing sexual harm
Complaint management systems assume a gendered use scenario in which the user is female and the accused is male. I argue that this gendering serves to emphasize sexual harm over other forms of misconduct, and that this is done through the computational logic of categorization. Categorization is the prevailing logic of communicative technologies: softwares must collect and communicate appropriate information. As information scientists Bowker and Star explain, “To communicate…we must first classify” (Bowker and Star 2000, 68). The ways information is collected to create discrete categories “are not only powerful in the way they leave out things that do not fit; they also embody the structural logic of a system of classification” (Crawford and Gillespie 2016, 418). This classificatory drive of reporting systems, when intertwined with a singular eye toward sexual harassment, exceptionalizes sexual harm by diluting other forms of harm and positioning the user (almost always assumed to be female) as a victim.
A key component of dominance feminism theorizes sexual harassment as “male dominance and female subordination” (Halley et al. 2018, 36). As Halley explains, “dominance feminist thinking places sexual wrongs front and center, and assimilates other seemingly nonsexual wrongs to sexual ones” (38). This is most clearly demonstrated by how complaint management softwares take sexual harassment complaint as the default. While many softwares’ remit extends beyond sexual harassment, the information they take and presume reveals their gendered assumptions. Alphabet helpline is a telling example (see figure 2). Alphabet helpline offers one standard reporting interface regardless of the nature of the complaint. After collecting basic demographic information, the interface begins by asking the user to identify “persons engaged in this behavior.” The highlighted portion in figure 2 prefigures the “person engaged in this behavior” as a man (“John Doe”) in a senior position (“Night Supervisor”). No other demographic information about the offenders is requested.
The user is similarly predicted to be a woman. The most direct representation of this assumption is on Vault Platform’s dashboard page, where a faceless figure of a woman is locking up a vault (see figure 3). The dashboard is where users’ reports are stored, and the visual representation of the user as a woman indicates the company’s expectations of use. Other softwares reveal their gendered expectations through their use of color and symbols. #NotMe and Spot, for example, both employ a primarily pink and purple color palette against a white, minimal backdrop, an “aesthetics of white femininity” that signals the developers’ expectation about ideal user base, according to feminist media scholars Murray and Sapnar Ankerson (2016). How the reporting interface intuits the user’s sex and nature of the complaint through visual and technological arrangements exemplifies its focus on sexual wrongs over others.
I have thus far investigated how data-driven complaint management softwares embody and legitimize the dominance feminist paradigm of sexual harassment. Examining which feminist ideas gain purchase through particular technological arrangements illuminates a need for a critical feminist inquiry that does not prefigure feminism as universal and shared. While I have focused on reporting tools in the United States, similar tools are deployed with parallel rhetoric elsewhere (Lakshané and Chinmayi 2017). As new technologies emerge, there is a greater need for a critical feminist inquiry that can capture, hold, and assess the complexities of how and why particular feminist ideas and technologies intertwine when it comes to advancing gender justice.
Kate Sim is a doctoral candidate at the Oxford Internet Institute, University of Oxford, where her dissertation studies how data-driven reporting technologies reconfigure the ecosystem of campus safety in the United States. She is currently a PhD research intern at Microsoft Research’s Social Media Collective, where she examines how platform companies figure ideas about harm and care in their reporting mechanisms. Her academic and public writing spans algorithmic accountability in public life, research ethics for computational researchers, and data practices for survivors of gender-based violence.
The author would like to thank the anonymous reviewers and special collection editors Payal Arora and Rumman Chowdhury for taking the time to engage with this article so carefully, especially during the exceptional circumstances of the ongoing pandemic. She extends her special gratitude to Anna Gibson, Amelia Hassoun, Amanda Haziz-Ginsberg, Rebecca Lewis, and Nahema Marchal for their insightful and generous comments on earlier drafts and thoughtful conversations about this work.
The author declares no potential conflicts of interest with respect to the research, authorship, or publication of this article.
The vocabulary of combat is used widely to describe various technologies deployed in relation to sexual harassment in the workplace, education, and others in popular discourse. See Troy Wolverton, “How an Academic Specialist in Human Memory Created a Chat App That’s Helping Companies Fight Harassment and Discrimination,” Business Insider, November 11, 2018, https://www.businessinsider.in/how-an-academic-specialist-in-human-memory-created-a-chat-app-thats-helping-companies-fight-harassment-and-discrimination/articleshow/66579914.cms; Lauren Kassien, “5 Apps Fighting Back against Sexual Assault,” Paste Magazine, February 6, 2015, https://www.pastemagazine.com/college/five-apps-fighting-back-against-sexual-assault/; and Juana Summers, “Smartphone Apps Help to Battle Campus Sexual Assaults,” NPR, August 13, 2014, https://www.npr.org/2014/08/13/339888170/smartphone-apps-help-to-battle-campus-sexual-assaults. Similar language is used to describe apps deployed in a non-US context. See Abby Young-Powell, “Reclaiming the Streets: The Apps Helping Women Tackle Sexual Violence,” Guardian, March 8, 2018, https://www.theguardian.com/careers/2018/mar/08/reclaiming-the-streets-the-apps-helping-women-tackle-sexual-violence; Victoria Dannemann, “In Latin America, Women Turn to Apps to Combat Sexual Violence,” DW, May 10, 2019, https://www.dw.com/en/in-latin-america-women-turn-to-apps-to-combat-sexual-violence/a-50664628.
See Callisto, https://mycallisto.org/. In Callisto, when users submit a record of their incident of sexual harassment, they are encouraged to input perpetrator(s)’ information. This information, such as an email address or a social media URL, is then stored by the reporting system. When records share the offender, the perpetrator’s information is used as a unique identifier to match those records. Users are then alerted that they have a shared offender, which draws on the repeat offender theory by Lisak and Miller (2002). The technological operationalization of this theory draws on Ayres and Unkovic (2012).
Callisto’s theory of sexual assault prevention and response is largely informed by the repeat offender theory (Lisak and Miller 2002). Repeat offender theory of campus sexual assault explains a subset of campus sexual assaults, but as academic researchers in sociology, psychology, and public health have extensively discussed, it is not a comprehensive or instructive explanation of how and why students experience sexual assault. See Jennifer S. Hirsch and Shamus Khan, Sexual Citizens: A Landmark Study of Sex, Power, and Assault on Campus (New York: W. W. Norton, 2020); and Ana Gantman and Elizabeth Levy Paluck, “What Is the Psychological Appeal of the Serial Rapist Model? Worldviews Predicting Endorsement,” SSRN Scholarly Paper (Rochester, NY: Social Science Research Network, June 1, 2018), https://papers.ssrn.com/abstract=3190670.
The URL redirects to Leda Health, which appears to be the kit’s renamed and rebranded vendor.
I am aware that my use of the term “inclusion” may be misleading to some readers who are wont to equate it with the politics of equality, diversity, and inclusion (EDI) in the tech industry. My use of the term has a clear scope in technology-making, but the word choice is in dialogue with feminist critiques of EDI initiatives. As many scholars have extensively written, EDI initiatives’ impacts are limited and, when effective, prioritize the interests of white, entrepreneurial women. For critiques of EDI initiatives, see Sara Ahmed, On Being Included: Racism and Diversity in Institutional Life (Durham, NC: Duke University Press, 2012) and Sirma Bilge, “Whitening Intersectionality,” in Racism and Sociology: Racism Analysis Yearbook 5, ed. Wulf D. Hund and Alana Lenton (Berlin: Lit Verlag/Routledge, 2014), 175–205. For analyses specific to the tech industry, see Christina Dunbar-Hester, Hacking Diversity: The Politics of Inclusion in Open Technology Cultures (Princeton, NJ: Princeton University Press, 2019); and Sarah Myers West, Meredith Whittaker, and Kate Crawford, “Discriminating Systems: Gender, Race, and Power in AI,” AI Now Institute, https://ainowinstitute.org/ discriminatingsystems.html.
For Callisto, see TED, “The Reporting System That Sexual Assault Survivors Want,” https://www.ted.com/talks/jessica_ladd_the_reporting_system_that_sexual_assault_survivors_want?utm_campaign=tedspread&utm_medium=referral&utm_source=tedcomshare; for #MeToo Kit, see Alex Williamson, “DIY Rape Kit Startup’s 23-Year-Old Founder Says She Is a Sexual Assault Survivor,” Brooklyn Daily Eagle, September 5, 2019; for Botler.ai, see Zosia Bielski, “Stepping Up: Woman Who Was Stalked Uses Her Own Experience to Develop AI-Powered App to Provide Information for Other Victims,” Globe and Mail, October 20, 2020, https://www.theglobeandmail.com/canada/article-stepping-up-woman-who-was-stalked-uses-her-own-experience-to-develop/.
Much of public outrage focused on the power imbalance between the makers and the users of the app, but overlooked yet crucial actors here are app providers. It was app providers’ decision to remove LegalFling that ultimately led to its public demise. Apple and Google cited “mature content” as a reason, but it is safe to assume that this decision was largely in response to public hostility, given that the app itself did not feature sexually provocative content. App providers, as Tarleton Gillespie (2018) argues, are thus a “venue for public contestation” as corporate entities like Apple and Google determine which apps become publicly visible. In doing so, app providers’ decision to make available or remove certain apps is “unavoidably political.” Tarleton Gillespie, “Exodus International: Banned Apps, App Store, and the Politics of Visibility,” in Appified: Culture in the Age of Apps, ed. Jeremy Wade Morris and Sarah Murray (Ann Arbor, MI: University of Michigan Press, 2018), 51–60.
Scholarly writings widely consider Antioch College’s Sexual Offense Prevention Policy (SOPP) in 1991 as the first implementation of affirmative consent policy. It was widely mocked, including in one Saturday Night Live episode (see https://snltranscripts.jt.org/93/93bdaterape.phtml), but the administration maintained the policy. See SOPP here: https://antiochcollege.edu/campus-life/sexual-offense-prevention-policy-title-ix/. Today some advocates, including the sex positivity community, look upon it as a progressive alternative, though others, both from conservative and progressive standpoints, critique its simplicity and broad remit. For more detailed history of the sexual politics surrounding the affirmative consent standard, see Michelle J. Anderson, “Negotiating Sex,” S. Cal. L. Rev. 78 (2004): 101–38; Joseph Fischel, Screw Consent: A Better Politics of Sexual Justice (Oakland, CA: University of California Press, 2019).
In 2014, for example, the state of California signed into law the affirmative consent standard for higher education institutions. See California Senate Bill 967 here: https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=201320140SB967. For other US states with similar policies, see http://affirmativeconsent.com/affirmative-consent-laws-state-by-state/.
This is, of course, an incomplete representation of the dynamic between the system vendor, client (employer), and user (employee) as none of these entities are monolithic. The end user may be a senior member of a company, and back-end users from the HR department are also employees. Because I am primarily interested in how these dynamics are encoded in the design of misconduct management systems, I use “client” and “user” distinctly to emphasize the product vendors’ understanding of their role and responsibility in relation to the two groups.
In this regard, how anonymity works (or doesn’t) is similar to what Nissenbaum terms the “contextual integrity” of privacy. See Helen Nissenbaum, “Privacy as Contextual Integrity,” Washington Law Review 79 (2004): 119–58.