Harassment & SocialVR (Relationship / Types / History / Solutions)
What follows are selected pages from my completed thesis, A Framework for Understanding and Detecting Harassment in SocialVR. They compose the majority of its background section. If this is a topic of intense interest, you may read fully. However, if you are overly familiar with some of the topics, scrolling through for topics of interest may be more valuable. Each new section can be found in bold. The concluding section has solutions for the problem which are elaborated upon in further sections of the thesis. A concise description of the NLP section appeared in an earlier post. Enjoy!
Workplace Harassment
Discussions and legal actions regarding harassment may have originated with workplace harassment since it predates the internet and it is an environment which brings together a diverse group of adults wherein there is generally a need to communicate and form relationships with others. Where a group of people who lack mutual understanding come together for a shared purpose, there lies the potential to create hostile environments either out of ignorance, apathy, or malicious intent. A hostile work environment is one in which employee behaviors lead to changes of emotional discomfort to the working environment or abuse (Rotundo, 2001). In this case, harassment would be classified as the behavior that leads to this environment and it can be severe enough to prompt emotional distress and the victim’s resignation from their position resulting in lost wages.
Landmark Harassment Cases
As an example, one legal case finalized in 1993 comes from the US Supreme Court that tied harassment to the psychological injury of the victim and discriminatory workplace practices based on the employee’s sex, race, national origin, or religion (Case no. 92–1168, Teresa Harris v. Forklift Systems, Inc.). In that case representing discriminatory practices against an employee’s sex, the defendant had referred to the plaintiff as a ‘dumb ass woman’ and asked her to pick up items from the ground so he might look inappropriately at her body alongside other male colleagues (Epstein, 1995). A 1998 US Supreme Court case (Case no. 523 U.S. 75, Onacle v. Sundown Offshore Services, Inc.) clarified sex-based workplace harassment as being possible against transgendered persons, between litigants of both the same and different genders, and other instances of gender non-conformity such as homosexuality. Cases of racially motivated harassment in the workplace have also been taken up and prosecuted by the US Supreme Court, including Vance v. Ball State (Case no. 570 U.S. (2013)). The behavior that led to this law suit involved a fellow employee intentionally blocking Vance’s path, weirdly smiling at her as if to ridicule her, intentionally banging cookware around her, and making Klu Klux Klan references to her, but the judgement of this case clarified the employer responsibility for harassment cases where there is a power differential, such as the instigator is a supervisor of or has a supervisory role over the complainant (Woska, 2014).
Harassment Definitions
The Equal Employment Opportunity Commission (EEOC) goes into greater detail on what constitutes sexual harassment at work by giving a list of potential scenarios along with the classification: verbal, non-verbal, or physical (EEOC, 1992). Examples of verbal harassment might be sexually explicit jokes, questions, or even sounds while non-verbal harassment may include the exposure to pornographic material or inappropriate gestures. Physical harassment involves the touching of another person or oneself in a threatening and potentially sexual manner. According to the EEOC, what links these behaviors as examples of harassment may not be their sexual nature, but the fact that they are ‘unwelcome’ to the recipient. The victim of harassment need not outwardly protest the unwanted treatment. Whether the conduct is unwelcome, and thereby defined as harassment, depends on how the individual considers it.
Harassment in Schools
Harassment concerns extend also to educational institutions, which are similar to the employment environment insofar as they include diverse sets of people who will be required to interact with one another. Students, with a varying knowledge of standards for behavior and typically lessened consequences for violating those standards, may find a greater likelihood of experiencing harassment as either the culprit or the victim. In university, where the parents may be uninvolved in such cases, there is typically a code of conduct that lays out a standard process for reporting breeches to this code. As an example, the University of California Berkley in their Code of Conduct defines harassment as actions that prevent a student’s participation in programs or activities in the university and extends the previously listed bases for harassment to include age, marital status, veteran status, and disabilities (Section V, Article 102.09). The Administrative Guide to Stanford University repeats this sentiment, describing harassment as ‘unreasonable interference’ or the creation of a ‘hostile environment’, a term also used in the context of the workplace. Furthermore, harassment may occur on repeated occasions or on a single occasion if the infraction is particularly extreme (Stanford University, 2016).
Street Harassment
It is also possible to find instances of harassment beyond the confines of formal institutions. Depending the location, gender-based harassment can be a common occurrence on the streets or in public settings and it has been given the name street harassment. Unlike the previously mentioned forms of institutional harassment, street harassment is a sexually motivated type of harassment done to women by people who are strangers to them (Bowman, 1993). As others have defined it before, this may include verbal, gestural, or physical assault intended to objectify or humiliate women (Peoples, 2008). It traditionally may receive less attention from lawmakers and academics than other forms of harassment since the perpetrators will typically be unknown and its potential harms, such as the feeling of being threatened or negative body image, are less quantifiable than the loss of career or academic opportunities. Likewise, countries such as the United States grant freedom of speech as a constitutional right and legislating street harassment could be interpreted as a violation of those rights (Nielsen, 2000). However, the freedom of movement has long been argued as a civil right and the inducement of fear through sexually aggressive speech or actions would present a limitation to women’s access to their basic human rights. Based on the US Supreme Court rulings in the aforementioned workplace harassment cases, I will include all sex-based harassment, not exclusively heterogenous, as a candidate for street harassment.
Hate Speech
The provided definition of street harassment does not adequately cover all forms of harassment occurring in a public place since it fails to include instances which are motivated by violence or the spread of hostility towards a specific group. For these instances, we will use the term hate speech which, according to Anne Weber of the Council of Europe’s Committee of Ministers is ‘…understood as covering all forms of expression which spread, incite, promote or justify racial hatred, xenophobia, anti-Semitism or other forms of hatred based on intolerance, including: intolerance expressed by aggressive nationalism and ethnocentrism, discrimination and hostility against minorities, migrants and people of immigrant origin’ (Weber, 2009).
Unlike street harassment, hate speech does not require that a member of the group under discussion be present. One person expressing ill feelings toward a group may be committing harassment by spreading their hostility among those people listening. The laws pertaining to hate speech vary from nation to nation, but United States law has been explicit in its consideration of hate speech as being protected according to the First Amendment of the Constitution (Bleich, 2014). If a crime committed against another person is motivated by hatred towards a group, it may be considered a hate crime and the perpetrator will receive an unfavorably modified sentence base on those overt declarations, but having and expressing those sentiments are not punishable by law.
Harassment Summary
Having covered different types of harassment and the importance of the location in which they occur, it might be helpful to illustrate the differences here. Sex-based harassment between people of different races is still street harassment so long as race is not the focal point for that harassment and it happens outside any formal institution, such as a workplace or school. A person who hypothetically discusses raping another person in a public setting is still performing verbally based street harassment since, even though rape is a form of physical assault, it is still spoken, sexually themed, and creates a threatening environment. On the other hand, if a White person in the presence of a Black person raises a fist to the sky, cocks their head, and sticks out their tongue, it is likely that this is mimicry of someone being hanged by a noose. Using United States history of lynching Black populations as the context, this is a strong candidate for both non-verbal harassment and hate speech. If one person corners another or a group of people crowds around an individual, this could be an instance of physical harassment, which could be sex-related or hate-related depending on the motivations of the crowd. These distinctions between different types of harassment will become important when discussing it in the context of virtual worlds where examples not unlike these were found.
Multi-User Dungeons
Multi-User Dungeons (MUDs) are online recreation centers and an early version of virtual reality in which visitors may act out a fantasy with multiple other users, where the imagined environment is built upon its text-only representation and the players assume an identity of their choosing. Though this venue of interaction has been long absent from the public eye, the number of multi-user games numbered in the hundreds back in 1993 and the type of verbal engagement among users ran the spectrum from mannerly to graphically sexual (Rheingold, 2000). Design decisions regarding the appearance of the collectively imagined environment might be based on an individual host of a dungeon or the environment could be designed by the whole group and in real time. Likewise, pseudonyms were generally used and, should they so choose, a user’s physical appearance could be determined through description by the users themselves. The dungeons could be centered on a theme, such as the Medieval or Science Fiction, and it could be directed towards a common goal or game that users play together.
Virtual Realities
As consumer versions of virtual reality headsets that immersed people in digital 360-degree environments were released throughout 2016, including: Vive, Oculus Rift, Samsung Gear, and PSVR, the association between MUDs and the term virtual reality has lessened considerably. VR has come to be an immersive, and perhaps interactive, visual environment mediated by HMDs and controllers. Users might be able to move within that space and there may also be audio components suited to the environment. There is 360-video content in which the user is a passive observer, but VR may also contain objects with which users interact. These objects cannot be spoken into existence as with MUDs. They must be designed and scripted, either by the developer or individual content creators, so they may be handled by the user and the objects may physically respond to one another. In short, a photo-realistically rendered, fully immersive, and fully interactive VR experience would be indistinguishable from reality as far as the eyes and ears are concerned (Steinicke, 2016).
Social Spaces in VR (SocialVR)
An experience in VR is defined here as SocialVR when multiple users simultaneously inhabit a virtual space where they are capable of interacting verbally and through the movement of their avatars. Variations of these virtual spaces in SocialVR include the simultaneous visitation of users to a single-user VR experience by means of an external program, but these are not in discussion here.
Communication in SocialVR
Communicative interactions in SocialVR are generally spoken. Communication through text is sometimes available, but it is not often used because the act of typing with VR controllers is slow and attempting to type with a real keyboard while in VR may, at the moment, be problematic. Most SocialVR platforms attempt a faithful simulation of in-person speech, where its volume is loudest directly beside the speaking avatar and decreases the further from them one travels. If spacious enough or there is an obstruction, speech in the environment may be inaudible to other users who are an adequate distance from the source. There are cases in which speakers can use available objects to project their voice at an equal volume throughout the environment. These objects may be microphones or megaphones (Rec Room), which correlate to objects in material reality, or more fanciful objects like magical cookies (Anyland), which do not have a widely known correlation. Therefore, users of environments with a large enough capacity can reasonably expect to encounter multiple conversations happening simultaneously throughout the environment. Users will often join, leave, and rejoin conversations, so the social norms of discourse, such as leave-taking, may not apply (Schourup, 2016). The breaking down of social norms is exacerbated when the platform or its users suffer technical failures which force them out of the conversation.
Movement in SocialVR
The movement of avatars can vary from user-to-user and platform-to-platform depending on the technology being utilized by users and the availability of its support within the SocialVR platform. Minimally, the movement of avatars will include their travel on the X-axis, Z-axis, and sometimes Y-axis. SocialVR platforms will often indicate when a user is speaking, either through manipulation of the avatar’s mouth (Rec Room, Facebook Spaces) or a flashing light emitted by the avatar syllabically corresponding to the users’ speech (AltspaceVR). Most SocialVR platforms use hand-held controllers whose movements are captured by sensors and represented as the users’ hands within the virtual environment. Some SocialVR platforms have support for motion capture, which will interpret the body, head, and limb movements of users (High Fidelity, AltspaceVR). In virtual environments, users may sometimes move from point A to point B through continuous, linear travel, but teleportation between points is more common because it is faster and generally more comfortable, which is important for the avoidance of motion sickness (Bozgeyikli, 2016). In teleportation, users direct a cursor to a distant spot on the terrain and they are immediately transported to that point when the necessary input has been received.
Privacy in SocialVR
SocialVR experiences can be divided into two types, private and public. Private experiences, or events, within SocialVR are between the hosts and their invited guests. These private experiences are intended to bring together people who are already somehow connected through the platform, people who have begun a conversation and want to continue it uninterrupted, or people who have met elsewhere and connected in the SocialVR platform. People meeting privately may consider SocialVR a venue for being with preexisting friends and they exclude unknown users out of convenience, but it is also a sure way of preventing contact with harassing or otherwise unwanted users.
In most SocialVR platforms, experiences in public environments are the default because they do not require an additional setup process. These experiences take place in common areas which all users of the SocialVR platform may visit. In some cases, these users need not even be registered and will be designated as guests. Some SocialVR platforms put upper limits on the number of visitors allowed within a given environment, also referred to as a room or, more literally, server. A user may or may not know the other users in their room and, while within capacity, users may freely travel between rooms.
Avatars in SocialVR
Users will inhabit avatars during their time in SocialVR and each platform differs in its customizability options for avatars. There are SocialVR platforms offering only generic avatars, which may be customized with regard only to clothes, gender, skin color, hair color, and eye color (Rec Room), while other platforms have a collection of avatars to choose from, which may be humanoid or robotic (AltspaceVR). A few SocialVR platforms offer greater degrees of customizability or even the ability for users to design and upload avatars to their SocialVR account directly (VR Chat). To be permitted, avatars may need to conform to the platform’s aesthetic, but other platforms will put no restrictions on their avatars’ appearance beyond the prevention of nudity. This means that the avatars of some platforms will include human-like physical curvature, which may even be highly sexualized or fetishized, and the avatar forms of other platforms will consist mainly of straight lines. There have been a few attempts at avatars that are photo-realistically representative of the user inhabiting them (EmbodyMe), but otherwise there would be little reason to assume a user looks at all like their avatar. This means that a user and their avatar might not share the same physical properties, including gender and skin color. Changing the appearance of one’s avatar is done either by accessing a menu or traveling to a specific location, like a dressing room, in the social VR platform. This means that the avatar of an individual user may change multiple times within a session, one moment inhabiting an icon of popular culture and the next moment an anime-style character in a school uniform. Despite these changes, the users will still be identifiable by a username, even if it is a generic name of a guest account, and that name will either be continually visible or accessed by clicking on their avatars.
Environment in SocialVR
The environments that users inhabit may also change throughout a session. Some SocialVR platforms allow for party membership, meaning when one party member leaves one environment for another, the other party members are invited to join (Rec Room). Users may be motivated to move to different environments for the sake of exploration or to pursue a desired activity. The environments may be developed either by the SocialVR platform or the users themselves. They may be constructed within the platform or imported from a game engine. The environments may also have interactive components, such as a bed to lie down in, a torch to give light, or a writing instrument for use on a flat surface. Some environments also allow for the inclusion of external media, such as photos, web content, and streaming videos, which may be exhibited by the SocialVR platform or the user directly. This shared content is the main feature of some environments and, arguably, it is the main part of some SocialVR platforms (Basement VR).
Harassment in MUDs
MUDs represent an early form of computer-mediated virtual reality because they enabled symmetric dialogue between users who could interact within environments. To create elements within an environment or make objects, users needed only to write about them within the dungeon, so others may acknowledge their existence and interact with them. As seen below, this can be a powerful tool in the hands of a harassing user even though the abuse happens through text alone and people do not share a literal space. As it is in-person, discriminatory or threatening behavior is emotionally detrimental to the victims and it is harmful to the virtual world itself as wronged users abandon the platforms or assume male identities to avoid being harassed (Fox, 2016).
Even if dialogue within a dungeon is directed towards specific users, the chat may be visible to everyone present in the space and, therefore, each of them may be subjected to harassment through humiliation or expressions of hatred. Take the following case, which is a real example of how two MUD users (ViCe and Aatank) performed sexual harassment against other users (sm, st, and rani) with only textual cues (Herring, 1999):
<ViCe>Aatank man i got women here u’ll fall in love with!!
<Aatank>vice like who
<ViCe>Aatank a quick babe inventory for u: st / sm and rani :)
<Aatank>sm hi u can call me studboy. what color are your undies
<ViCe>haha ¤ ¤ ¤ Action: Aatank rushes up to st and yanks her panties off. BOO!
Also in 1993, a case of simulated rape was reported in a MUD named LamdaMOO, where an abusive user assumed the identity of two female users, thereby forcing demeaning acts upon them which prominently featured sodomy and pubic hair (Huff, 2003). The incident initiated a broader conversation on virtual identities, censorship, and means of preventing harassment in virtual spaces. These cases do not represent forms of institutional harassment since any registered user has access to the MUD. Rather, this type of harassment is nearer to street harassment because it may happen between strangers and its resulting humiliation impedes the victims’ ability to move and act freely within the environment. One can easily imagine a case of hate speech also taking place in the same venue. Harassed users may feel frustrated and anxious due to their inability to participate meaningfully in the MUD. They may feel greatly embarrassed by the experience of being publicly targeted for the violation of social norms and the text-based sexual aggression could easily be triggering for users. This disruption to their experience could cause them to leave the platform, or perform gender masking in which users choose male or gender-neutral names to avoid unwanted, sexual attention (Fox, 2016)
Some MUDs had regulations that users were expected to adhere to while interacting in dungeons, but others lacked such precautions, especially in the pre-commercial days of the Internet. Creators of the MUDs seemed not to know what behavior to expect from their visitors. Users and administrators of MUDs may have had the ability to kick out offending players, but they had little, if any, power to prevent them from returning to the dungeon soon after. It is within this set of circumstances that a new type of harassment was conceived of and it has found renewed relevance in modern iterations of SocialVR.
Harassment in SocialVR and Counteractive Measures
At a minimum, harassment in SocialVR may be spoken to a user or acted out physically by standing too closely, participating in unwelcome touching, or acting out sexually suggestive pantomime. In this medium, gender masking generally ceases to be an option since users may suppose one another’s gender by the sound of their voice. SocialVR has already experienced a high-profile case of sex-based harassment when a woman was groped while playing the game QuiVR (Belamire, 2016). As of writing, how the body is captured in virtual reality is limited and, therefore, the sophistication of a user’s gestures is low, but this aspect of virtual reality is seeing continual advancement (Han, 2017). Some solutions to harassment have been implemented in response to verbal and physical harassment, and more may be done still, but SocialVR platforms are designed as a source of entertainment and a means of connecting to others. Building an environment in which users are continuously on guard, carrying a full arsenal of defensive measures, could easily undermine the purpose of the experience (Shriram, 2017). Therefore, the ideal measures for stopping harassment should be automated or easily accessible to the user, but they would not be obtrusive to the positive experiences of SocialVR. To elaborate further, there may be language and behavior that is appropriate between friends, or even users with mutual romantic interest, that would not be appropriate among other users, so we would ideally not want to stymie a good experience in SocialVR for the sake of defending against something negative. Furthermore, if users are given a reactive feature, making the victim responsible for initialization, then the means of accessing the anti-harassment tool should be clear. It should not require too many steps and it should definitively end the harassment from the offending user for at least the length of the session. Finally, the solution to harassment should not be subject to abuse, giving the offenders another way to disrupt the experiences of others as can happen with online tools (Ehrenkranz, 2017). The solutions discussed in the following sections are user-initiated features as automated tools seem to not be active.
Muting
The mute feature allows one user to silence another within the virtual space. The effect may or may not be reversed by the user who is muted. Additionally, the muted person may be silent only to the muting party or to everyone using the SocialVR application. Some platforms allow a user to mute any other user, or themselves, by simply clicking on a menu button beside the user’s nametag and the reason may be for the sake of harassment (AltspaceVR), but it’s more often to silence someone experiencing disruptive feedback from their microphone or someone speaking too loudly at a public event and preventing others from hearing the main presenter. In harassment cases, it could be effective in signaling one’s annoyance to the offending user, but the muting is quickly reversed, doing nothing to prevent the harassment and perhaps goading the offender into further attacks. At the same time, not allowing the muted person to unmute themselves would unfairly penalize the innocent. It would also fully remove real offenders from participation, but still enable them to enact forms of physical and non-verbal harassment against other users.
Blocking
Blocking, or Ghost Mode, is another option in some platforms, where the harassed user may click a menu button near their harasser or perform a specific gesture in the harasser’s direction (Rec Room). The only added action may be a request for confirmation. When given, the offending user will neither be visible nor audible to the blocker and vice versa. The two users will be unaware of the other’s presence within that space even though everyone else will be. This effectively ends instances of non-persistent harassment or annoyance by other users, but it does not prevent blocked user from returning under an alternate account, perhaps using another email address, to do further harm. Also, if the user proves to be a general nuisance, they must still be blocked by every individual user, which requires a lot of menu access cumulatively.
Kicking
Kicking is a feature of a SocialVR platform in which a group of users, or the host of a private room, may vote to remove a player from a room. When one player initiates the “kicking” of another, bystanders receive a notification in which they may also vote to remove that player. To illustrate, Rec Room is a SocialVR platform consisting of a common area that connects to multiple games and each game area may contain multiple rooms, the number depending on the games’ capacity. Once kicked, the player is ejected from the game and there is a short delay before they may re-enter, but the kicked player will be unable to join the same specific room again. This puts the decision to censor behavior into the hands of groups who may be frustrated by harassing behavior or the kicked players attempts to sabotage a game. However, the same tool can be abused by players who might wrongfully remove good players from a game in order to increase their chances of victory. In a justified incident of kicking, harassing players still have access to all other areas and it will not protect the victims of harassment if they leave the game where the harassment occurred.
Bubbles
Rather than filter out other users through muting or blocking, some platforms have a version of the protective bubble feature through which, from the users’ perspective, the physical form of another user’s avatar may not pass (High Fidelity, Rec Room, QuiVR).
Figure 3 AltspaceVR Buttons (Left) for Self, Bubble on Bottom (Right) for Other Users, Block in the Center, Mute on the Right
If another player nears or penetrates the invisible barrier, the body of the intruder will initially fade and then disappear. This is a solution working against sexually themed attack or intentionally intrusive users, but not verbal harassment or offensive mimicry from a distance. The bubble may be turned on at all times, but the user may choose to disable it should they want to come closer to other avatars. The radius of the bubble may be customized in some platforms, giving each user as much personal space as they require. Here, the user experience may suffer since they must turn off the bubble when making voluntary physical contact, but it might be a worthwhile tradeoff where the other users are unknown, or harassment seems likely. Bubbles may be called upon by accessing the menu, but there are some platforms that utilize gestures, such as raising one’s arms outwards, to access the protective bubble (D’Anastasio, 2016). Since it is a simple gesture, this solution presents a fast way to escape an uncomfortable or threatening situation, but the harasser is still present within the space at a short distance.
Reporting
SocialVR platforms generally have a method of reporting, or flagging, harassers, which can either be done from menus within VR or through a standard form available on the website. The terms of service for these platforms vary in the amount of detail in their descriptions of harassment and the penalties enacted for each type of harassment may not be stated explicitly. Potential outcomes for harassment claims are sometimes given and they may include the suspension of an account, the closure of an account and further blocking of a user’s Steam account, or complete blockage of access by someone using a specific IP address. The enforcement and penalties for harassment are at the discretion of the administrators of the SocialVR platform, whose interests in preventing access to their platform may conflict. First of all, barring someone from accessing the platform directly lowers the number of users on the platform and removing an individual user could potentially lead to the loss of the social network connected to that user. Therefore, the risk of losing the harassing users might be weighed against the likelihood of retaining the harassed users and the potential for further harassment from the offending users in the future. In addition, heavy handed enforcement of harassment policies carries the potential for a backfire effect in which a network of users engages in systematic trolling behavior for the explicit sake of disrupting the SocialVR platform entirely (Binns, 2012).
Admins
SocialVR platforms with a large enough usership are known to keep admins stationed in continually populated common areas. The admins are humans employed by the platform to monitor users’ behavior by remaining in the environment with them, engaging in conversation with them and warning them away from excessively harassing behavior. This solution is the surest method of classifying harassment, but it is likely to be untenable when SocialVR usership grows and it will be a superfluous position when unsupervised and automated methods of harassment detection become available.
Trolling Behavior
Since SocialVR integrates aspects of online and in-person communication, an understanding of behaviors relevant to both arenas will give a broader picture of the players involved in an instance of harassment. Online trolling is defined as malignant actions intending to compromise a social environment and, as studies have shown, this behavior is often correlated with the sadistic tendencies of trolls generally. Since they share the same potential for harm and havoc alongside the cloak of virtual anonymity, one can assume that the troll’s motive of deriving pleasure from another’s pain crosses over from the old domain of online social networks to the new domain of SocialVR. It is also consistent with trolling behavior to abuse or skirt systems of preventing their harassment, so their unwelcome behavior may continue unabated. Trolls have also been known to coordinate their attacks against entire platforms if, for example, they disagree with the introduction of a new policy, and this may be disruptive to every other user in a highly publicized manner (Higgin, 2013).
Concurrently, SocialVR may simulate the experience of being physically present with a person insofar as users may see one another by proxy and speak to each other in real time. For this sense of presence, SocialVR lends itself to sexual advances by users who may feel a heightened sense of gratification from their behavior. That is not to say every case of sexually themed conversation or physical movement is unwanted or improper as some users log in specifically to meet with a romantic partner or flirt congenially with other users. However, sexually motivated users with harmful intentions may easily address other users or initiate physical contact inappropriately while there is a lessened chance of repercussions for this behavior. It can be argued that the unique sense of presence and altered mobility that comes with modern SocialVR, to a degree, increase both the likelihood of harassment taking place and the stakes for the social VR platforms to keep that harassment from happening.
Proposed Methods for Detecting Harassment in SocialVR
The current methods listed above for preventing or responding to harassment all share the property of being user initiated. Making users responsible for responding to harassment against them requires educating them on the available anti-harassment tools and encouraging them to use it. However, if tutorials on preventing harassment become mandatory for registration on the platform, this has the potential side effect of discouraging new users from a lengthy registration process while making them wary of the SocialVR experience since they may now expect to be harassed (Burns, 2008). There may be an additional reluctance to use these tools on the users’ part because they foresee it as a source of conflict if the harasser learns of their action. Finally, the harassed users, on principle, may not wish to disrupt the harassing user’s experience.
One answer to identifying and preventing harassment may come from an automated response initiated by tools in the SocialVR platform rather than the user. The tools needed to detect harassment would begin with user profiling that considers data on the users, such as length of membership and history of abuse, and users would receive a score based on this profile. Features of the virtual environment may also count towards the score, such as the time of day and current number of users. Transcriptions from the users’ sessions may be taken and NLP may be used to find patterns in harassing speech, including the repetition and syntactic context of taboo words and hate speech. Analysis of the transcribed discourse would also result in a score to be added to the total and, if the score rises above a preestablished threshold, an action is triggered to deal with the potential harassment.
Since two-dimensional and three-dimensional image creation is also available to users of most SocialVR platforms, there should also be automated tools for preventing its abuse. The image could either be sexually explicit, related to hate symbols, or written words. There should be a method of capturing user-generated images and classifying them as harassing. Resulting actions taken by the platform should alleviate the potential hostility sparked by the drawing and prevent it from reoccurring in the future.
REFERENCES
[The Awesome Patman]. (2013, January 8). /b/ Habbo Hotel Raid — 2013. Retrieved from https://www.youtube.com/watch?v=6TDgAfQJRYY&t=116s
D’Anastasio, C. (2016, October). VR Developers Add ‘Superpower’ To Their Game To Fight Harassment. Retrieved from http://kotaku.com/vr-developers-add-personal-bubble-to-their-game-to-fi1788237241
Belamire, J. (2016, October). My First Virtual Reality Groping — Athena Talks — Medium. Retrieved from https://medium.com/athena-talks/my-first-virtual-reality-sexual-assault-2330410b62ee
Binns, A. (2012). DON’T FEED THE TROLLS! Managing troublemakers in magazines’ online communities. Journalism Practice, 6(4), 547–562. Retrieved from http://www.tandfonline.com/doi/abs/10.1080/17512786.2011.648988
Bleich, E. (2014). Freedom of expression versus racist hate speech: Explaining differences between high court regulations in the USA and Europe. Journal of Ethnic and Migration Studies, 40(2), 283–300. Retrieved from http://www.tandfonline.com/doi/abs/10.1080/1369183X.2013.851476
Bowman, Cynthia Grant. “Street harassment and the informal ghettoization of women.” Harvard Law Review (1993): 517–580. http://www.jstor.org/ stable/1341656
Bozgeyikli, E., Raij, A., Katkoori, S., & Dubey, R. (2016). Point & teleport locomotion technique for virtual reality. In Proceedings of the 2016 Annual Symposium on Computer-Human Interaction in Play (pp. 205–216). ACM. Retrieved from https:// dl.acm.org/citation.cfm?id=2968105
Burns, M., Bodine, K., Melnikova, O., & Geller, S. (2008). Required Registration Lowers Online Conversion Rates. Forrester. (as cited by Gualtieri, M. (2009). Best Practices In User Experience (UX) Design. Design Compelling User Experiences to Wow your Customers)
EEOC (1992) Preventing Sexual Harassment (BNA Communications, Inc.) SDC IP .73 1992 manual. (as cited in KAMBERI, F., & GOLLOPENI, B. The Phenomenon of Sexual Harassment at the Workplace in Republic of Kosovo.)
Ehrenkranz, M. (2017). Trolls keep outsmarting anti-abuse tools. Will Twitter’s new system actually work? Retrieved from https://mic.com/articles/168041/trolls-keep-outsmarting-anti-harassmenttools-will-twitters-new-system-actually-work
Epstein, D. (1995). Can a dumb ass woman achieve equality in the workplace-Running the gauntlet of hostile environment harassing speech. Geo. LJ, 84, 399. Retrieved from http://heinonline.org/HOL/Page?handle=hein.journals/
Fox, J., & Tang, W. Y. (2016). Women’s experiences with general and sexual harassment in online video games: Rumination, organizational responsiveness, withdrawal, and coping strategies. new media & society, 1461444816635778. Retrieved from http://journals.sagepub.com/doi/abs/10.1177/1461444816635778
Han, D. T., Sargunam, S. P., & Ragan, E. D. (2017, March). Simulating anthropomorphic upper body actions in virtual reality using head and hand motion data. In Virtual Reality (VR), 2017 IEEE (pp. 387–388). IEEE. Retrieved from http://ieeexplore.ieee.org/abstract/document/7892339/
Herring, S. C. (1999). The rhetorical dynamics of gender harassment online. The Information Society, 15(3), 151–167. Retrieved from http:// www.tandfonline.com/doi/ abs/10.1080/019722499128466
Higgin, T. (2013). FCJ-159/b/lack up: What Trolls Can Teach Us About Race. The Fibreculture Journal, (22 2013: Trolls and The Negative Space of the Internet). Retrieved from http://twentytwo.fibreculturejournal.org/fcj-159-black-up-what-trolls-can-teach-us-about-race/
Huff, C., Johnson, D. G., & Miller, K. (2003). Virtual harms and real responsibility. IEEE technology and society magazine, 22(2), 12–19. Retrieved from http://ieeexplore.ieee.org/abstract/document/1216238/
Nielsen, L. B. (2000). Situating legal consciousness: Experiences and attitudes of ordinary citizens about law and street harassment. Law and Society Review, 1055–1090. Retrieved from http://www.jstor.org/stable/3115131
Peoples, Fatima Mareah. “Street harassment in Cairo: a symptom of disintegrating social structures.” African Anthropologist 15.1&2 (2008): 1–20. Retrieved from https://www.ajol.info/index.php/aa/article/viewFile/77244/67691
Rotundo, M., Nguyen, D. H., & Sackett, P. R. (2001). A meta-analytic review of gender differences in perceptions of sexual harassment. Retrieved from http:// psycnet.apa.org/record/ 2001–18662–009
Schourup, L. C. (2016). Common discourse particles in English conversation. Routledge.
Shriram, K., & Schwartz, R. (2017). All are welcome: Using VR ethnography to explore harassment behavior in immersive social virtual reality. In Virtual Reality (VR), 2017 IEEE. 225–226. IEEE. Retrieved from http://ieeexplore.ieee.org/ abstract/document/7892258/
Stanford University. (n.d.). Article 1.7.1 Sexual Harassment. Administrative Guide. Retrieved July 17, 2017 from https://adminguide.stanford.edu/chapter-1/subchapter-7/policy-1-7-1
Steinicke, F. (2016). Being Really Virtual: immersive natives and the future of virtual reality. Springer. Retrieved from https://books.google.com.tr
University of California Berkley. (n.d.) Code of Conduct. Retrieved July 17, 2017 from http://sa.berkeley.edu/student-code-of-conduct
Weber, A. (2009). Manual on hate speech. Council of Europe (as cited in Koncavar, A. Hate Speech in New Media. Academic Journal of Interdisciplinary Studies)
Woska, W. J. (2014). THE CONTINUING DEVELOPMENT OF THE LAW ON SEXUAL HARASSMENT. Legal and Regulatory Issues in Human Resources Management. 207–228. Retrieved from https://books.google.com.tr/
