Cyber Violence:

Herring, Susan C. (2002). Cyber Violence: Recognizing and Resisting Abuse in Online Environments. Asian Women 14 (Summer): 187-212

Cyber Violence:
Recognizing and Resisting Abuse in Online Environments

Susan C. Herring

Indiana University, Bloomington USA

herring@indiana.edu

Introduction

Hundreds of millions of people currently use the Internet to enhance their lives and those of others. Yet a growing segment of the online population abuses the Internet for antisocial purposes, to stalk, harass and prey on other users, often with distressing effects. Internet-mediated aggression is a global phenomenon, and, disturbingly, it is on the rise,¹ lending prima facie credence to the dystopian view that computer-mediated communication exacerbates bad behavior.² To make matters worse, the tide of online violence is rising at a time when the Internet has moved from being a luxury to a necessity of daily life for educated people throughout the industrialized world. 'Cyber violence' thus stands to have negative impacts on a very large scale.

This paper is concerned with abusive online behaviors, which I group under the label 'cyber violence'. As a first step towards understanding this phenomenon, I define and identify major cyber violence types, illustrating each with case studies gathered from the Internet, and summarizing recommendations for responding to each. The advantages of defining and classifying cyber violence include 1) making the behaviors easier to recognize and name when they occur, 2) allowing strategies of resistance and redressive action to be articulated that are tailored to the demands of each, 3) distinguishing cyber violence from other, less serious forms of annoying online behavior, and 4) revealing underlying relationships between cyber violence and other phenomena, such as pornography, that might otherwise go undetected, but that help us to situate cyber violence within a broader perspective.

Defining Cyber Violence

I define cyber violence as online behavior that constitutes or leads to assault against the well-being (physical, psychological, emotional) of an individual or group. What distinguishes cyber violence from traditional off-line forms of violence is that in the former case, some significant portion of the behavior takes place online, although it might then carry over into offline contexts. Cyber violence thus may, but need not, have a physical component, and much of the harm caused by cyber violence—as indeed by offline violence—is psychological and/or emotional (which is not to say less real or destructive). Finally, cyber violence may be targeted at individuals or groups, the latter being more characteristic targets of cyber violence than of offline, physical violence, due to the ease with which a single perpetrator can gather information about and make contact with large numbers of people on the Internet. This is another aspect of online violence that can cause it to have widespread effects.

Violence and Gender

Violence is related to gender. Research has shown that men are disproportionately the perpetrators, and women disproportionately the victims, of violence in the physical world (Cyber-stalking.net, 2002). Cyber violence shows a similar pattern. Women were the victims in 84% of online harassment cases, and men the perpetrators in 64% of cases³ reported to the organization Working to Halt Online Abuse in 2000-2001 (WHO@, 2002). For many female Internet users, online harassment is a fact of life. One out of five adult female Internet users reported having been harassed online as of 1994 (Brail, 1994), and as many as one out of three female children reported having been harassed online in 2001 alone (Thomas, 2002). Among children, girls are targeted at almost twice the rate of boys (Finkelhor et al., 2000).

Males are also victims of violence (particularly of violence perpetrated by other males), and females also commit acts of violence, both online and offline. However, to ignore the larger gender pattern associated with violence is to miss a basic insight into the social reality of violence as a means of control and intimidation. That is, it tends to be perpetrated downward along a power hierarchy, thereby reinforcing societal gender asymmetries.

Classifying Cyber Violence

One obstacle to taking effective action against cyber violence is that it tends to be viewed as less serious, less "real" than violence in the off-line world. This is due in part to the relative novelty of the phenomenon (and of cyberspace as a whole); cyber violence does not conform to our familiar prototype of violence in a number of respects. As shown in Figure 1, violence can be situated along a continuum from more to less prototypical.

More prototypical violence <----------------------------------------> Less prototypical violence
Off-line	Online
Physical	Virtual
Action	Symbols
Intentional harm	Harm not intended
Targeted against an individual	Untargeted, diffuse
Perpetrator is socially marginal	Perpetrator is an average person

Figure 1. Dimensions of violence

A prototype is a mental representation of a complex concept in terms of its default or "most typical" realization. When we think of violence, we typically think of off-line behavior before we think of the Internet; we think of physical aggression before we think of deception or mental cruelty; and we think of action before we think of symbolic behavior via words or images. As the old saying goes, "Sticks and stones may break my bones, but names can never hurt me." Our prototype probably also involves an intentional perpetrator—non-intentional harm is usually characterized by other terms, such as 'accident'—with a specific target or targets, although untargeted violence (known by terms such as 'rampage' and 'mayhem' when it occurs offline) is also possible. Last but not least, we expect perpetrators of violence to be socially marginal types—possibly with a history of violent or criminal behavior—rather than average, well-adjusted individuals (cf. 'white collar crime', which often goes undetected because of this assumption).⁴ Cyber violence is less prototypical than physical violence in where and how it takes place, in allowing perpetrators to deny their intent to harm more easily (see below), and in enabling "normal" people to perpetrate widely-targeted harm, without requiring that the perpetrator be in an extreme emotional state (or risk his or her life) to carry it out.

Because cyber violence differs from our prototypical associations of violence, it may be difficult at first to recognize it for what it is, and accordingly, harder to resist and punish it. Thus a necessary first step in fighting cyber violence is to identify and name its manifestations.

Types of Cyber Violence

Four basic types of cyber violence are distinguished here:

1. Online contact leading to off-line abuse

2. Cyber stalking

3. Online harassment

4. Degrading representations

The numbering of the types is intended to suggest their distance from "prototypical" violence, with (1) being closest, and (4) most distant, from the real-world, physical prototype. Each type is defined and discussed below.

1. Online contact leading to off-line abuse

The first type of cyber violence involves online misrepresentation (usually someone represents themselves as being nicer or more socially desirable than they actually are) leading to abusive off-line contact, including, but not limited to, financial fraud, theft, unwanted sexual contact, and/or beating. The misrepresentation aspect invokes issues of deception and betrayal of trust; accordingly, this phenomenon is sometimes discussed in terms of those issues (and in relation to issues of 'cyber trust' more generally).

An example is the Katie Tarbox case, which received considerable publicity in the United States several years ago:

Example 1:

Katherine Tarbox was thirteen when she met twenty-three-year-old "Mark" in an online chat room and became close to him through a series of evolving email messages. When she finally met him in person, however, he turned out to be a 41-year-old named Frank Kufrovich with a history of pedophilia. He molested the girl in a hotel room. (Tarbox, 2000)

This case has several features that are characteristic of prototypical forms of violence: in addition to deception, it involves physical abuse (sexual molestation), a (male) perpetrator with a history of abusive behavior, and a highly vulnerable victim, a 13-year old girl. It is the sort of story that gives parents nightmares, and that gets widespread play in the popular media, in part because its familiar components trigger archetypal fears of predation.⁵

The second case is also an instance of online misrepresentation leading to offline abuse, although the resultant harm is less serious:

Example 2:

“I once met a lady-friend on the Internet. We went out a few times and I fell in love with her. Unfortunately I did not realize that she was lying to me the whole time. The more money I lent to her the more she lied. Now she is in jail and I have less money than I had before. I wish that I had checked out her past before I fell in love with her and subsequently lent her money.” (http://www.whoisshe.com/)

The victim in this case is a male, and the perpetrator a female, reversing the typical gendered power dynamic associated with violence. Moreover, the offline harm is financial (fraud) and emotional (deception), rather than physical. Nonetheless, the case fits the definition of cyber violence in that it involves online behavior that leads to assault against the well-being of an individual. It also features a perpetrator with criminal tendencies.

In contrast, most people would probably not consider the third case, an early classic of Internet deception, an example of cyber violence:

Example 3:

A male psychiatrist in his fifties carried out a long-term impersonation of Joan, a young, disabled woman, on a computer network. Communicating only via computer or mail, the impostor developed intimate friendships with several women. Upon learning the truth, however, the women felt betrayed; some mourned the loss of their relationship with Joan. (Van Gelder, 1985)

Although the women who had become close with 'Joan' (including, in some cases, having cyber sex⁶ with 'her'—'Joan' was bisexual) were intentionally deceived, the offline harm they experienced was restricted to emotional reactions to events that had taken place online. Moreover, although the perpetrator was an older male, he was socially respectable (a psychiatrist), and his alleged motive was not to harm, but rather to explore the potential of the new medium to experience for himself what it was like to be a woman. For many Internet users and scholars, the psychiatrist's behavior is to be expected in a text-only environment that invites play with gender identity (e.g., Bruckman, 1993; Danet, 1998; McRae, 1996). The implicit moral position underlying this view is that behavior that can not effectively be prevented must therefore be tolerated.

Such ambiguous cases notwithstanding, there are clear limits on when deception should be considered actionable abuse. It seems self-evident that disappointing online romances are not inherently violent. Although disillusionment can feel like deception, and rejection can feel like abuse, both are within the realm of normal risks associated with romantic intimacy, online and offline. This does not preclude the possibility that a romance turned sour can develop violent aspects, of course.

Recommended responses

How can Internet users avoid online misrepresentation leading to offline abuse? How should victims of such behavior respond? Most of the advice currently available places the onus on users to be wary: they should not trust that people they meet on the Internet are who and what they say they are; they should take precautions if they meet Internet acquaintances face-to-face, and report them to the authorities if they are the victims of abuse.

Because violation of trust is often at issue in this form of abuse, some sources recommend doing (or hiring someone to do) a background check to verify the identity, history, and reputation of online acquaintances before getting involved with them emotionally, sexually, financially, or otherwise (http://www.whoisshe.com/). Some discussion forums operationalize this principle in the form of “trust metrics" according to which people new to the forum can not join in unless some number of old and trusted members are willing to vouch for them. Some evidence suggests that groups that employ trust metrics experience less disruptive and abusive behaviors as a result (Levien, 2000).

People who choose to meet offline with someone they have previously only encountered online are advised to take precautions to protect their personal (physical) safety. Suggestions include to meet for the first time in a public, well-lit place, in daylight, in the presence of a friend, or if alone, to inform a close friend of the meeting place and check in with him or her by phone while there (Egan, 2000).

If, despite these precautions, abuse occurs, the abusers should be reported. This is especially important if there is evidence that they are repeat abusers; failure to report them makes it more likely that they will continue their pattern of abuse with others. Online anti-cyber violence organizations take in reports and sometimes intervene to stop abusers, including reporting events to law enforcement authorities in cases where laws have been broken (WHO@, 2002). Katie Tarbox reported her abuser to the police; he was prosecuted and sentenced to 18 months in jail for pedophilia (Tarbox, 2000). In cases where the abusive behaviors do not meet the criteria for legal intervention, as in the case of 'Joan' in example 3 and other "cyber lotharios", Bell and de la Rue (n.d.) recommend that victims make information about the abusers and their behavior publicly available, as a means of warning others away from them.

2. Cyberstalking

The second type of cyber violence is cyberstalking, defined as online monitoring or tracking of someone’s actions with illegitimate intent. Because it involves using the Internet to gather personal information about its target, cyberstalking is a violation of privacy, and is sometimes discussed in the context of Internet surveillance. To the extent that the target is aware that s/he is being stalked, cyberstalking functions as a form of intimidation: "I know who you are; I know where you (and your children) live, and … (implied: I will use that information to harm you)". An implicit threat of harm can be sufficient to intimidate; in other cases, cyberstalking leads to explicit threats of physical violence (including death threats) and/or actual off-line contact.

Example 4 is a case involving online stalking accompanied by explicit threats. The case in example 5 combines online with offline stalking.

Example 4:

Duwayne I. Comfort, a University of San Diego graduate, stalked female students on the Internet. Comfort used a professor's credit card to buy information about the women via the Internet. Comfort sent about 100 messages that included death threats, graphic sexual descriptions and references to the women's daily activities. (http://www.unc.edu/courses/law357c/cyberprojects/spring00/cyberstalking/)

Example 5:

Assistant Professor Pamela Gilbert from the University of Wisconsin was stalked by a university lecturer, Tim—a guy she had a couple of dates with. Tim posted her picture on sex sites, told mutual colleagues that Pamela was involved in Satanism, hounded her online, and got one of his students to follow her to "gather research" for a supposed book he was writing about her life. The student became suspicious when Tim started talking about a gun. (Gilbert, 1997)

In both of the above cases, the victims—strangers in the first case, and an acquaintance (a former lover) in the second case—were contacted directly by the stalkers. In both cases as well, the stalkers behaved in a threatening manner, leading the victims to fear that their lives were in danger.

A less clear-cut example, albeit one that raises similar issues, is the much-publicized Jake Baker case summarized in example 6.

Example 6:

Jake Baker sent email messages to an Arthur Gonda in Ontario, Canada from November 1994 to January 1995. In these messages, he discussed his interest in violent sexual acts. One of these emails contained a graphic fictional account of the rape, torture, and murder of a female classmate of his at the University of Michigan. (http://www.unc.edu/courses/law357c/cyberprojects/spring00/cyberstalking/)

Jake Baker was tried in court and found innocent of criminal wrongdoing, in part because his stories were not addressed to their ostensible target (the young woman did not know of their existence prior to being informed about them by the university). Debate also centered around the amount of risk posed by Baker to the young women—how likely was he to act out his violent fantasies? The lawyers for the defense argued that Baker's writings were creative acts of imagination that bore no necessary relation to his future behavior. In short, no intent to harm could be established.⁷ Jake Baker's violent and degrading depictions of women were thus construed as victimless, a problematic position that is considered further in reference to pornography in section 4.⁸

Many cyberstalking cases involve frustrated romantic and/or sexual interest. Cyberstalkers (both male and female) often target former lovers, or individuals about whom they entertain sexual fantasies. However, unwanted sexual solicitations in and of themselves do not constitute cyberstalking, particularly when they occur in isolation and do not persist after the solicited individual has said "no". Sexual come-ons are generally considered to be a fact of Internet life, especially in chat environments, and especially for users with female-sounding login names (Bruckman, 1993; Herring, 2001). Although such behavior is not unproblematic, it falls outside the definition of cyber violence presented here.

Recommended responses

How can Internet users avoid cyberstalking? Recommendations on this topic tend to focus on restricting access to one's self, and on reporting stalkers to the authorities as a means of obtaining additional protection and/or to force the stalker to desist.

A key issue in cyberstalking is privacy. To avoid being stalked in the first place, users are cautioned not to give out personal information such as phone numbers or addresses to strangers on the Internet (although it is increasingly easy for others to locate such information on the basis of a name or email address alone). WHO@ also recommends "ego-surfing"—searching under one's own name on the Web to find out what information is publicly accessible, and requesting to have it removed when it is too personally revealing. Targets of cyberstalking are sometimes advised to “get off the Internet”—this may prevent further abuse if the stalker does not know much about the target's real identity, but it is not a very satisfactory solution, in that it restricts the target's access to necessary resources. Many cyberstalking victims change their email address and login identity, some change their home phone number, and others move their residence to an unlisted address (Cyber-stalking.net, 2002).

As in other types of online abuse, victims are advised to save all evidence of stalking, and if the stalker does not desist, report him or her to the appropriate authorities. It may be possible to obtain a restraining order against the stalker, and some stalkers (such as Duwayne Comfort in example 4) have been arrested and convicted.

3. Online harassment

Online harassment is computer-mediated words, gestures, and/or actions that tend to annoy, alarm and abuse another person (cf. Black’s Law Dictionary, 1990). A crucial component of harassment is that the behavior is repeated—a single instance of abuse, such as an insulting email message, does not generally constitute harassment—and persistent, even after the harasser has been told to desist. The nature of the harm caused by online harassment is diverse, and can include disruption, insult/offense, and defamation of character. In the first, the target's resources (such as time and energy) are wasted; in the latter two, the target's self-image and reputation are attacked.

Example 7 is a case of online harassment that involves both types of harm.

Example 7:

Author Jayne Hitchcock exposed an Internet scam by a group of people calling themselves the Woodside Literary Agency. In retaliation, the agency launched a series of email bombs to her, her husband, and her lawyer. Then, the harassers forged posts in her name to hundreds of newsgroups. The posts indicated that Jayne was interested in having people call or stop by her house to share their sexual fantasies with her. Her home address and phone number were included. (WHO@, 2002)

Jayne Hitchcock's case involves elements of stalking (the perpetrators knew where she lived, and she changed residences to escape them) in addition to harassing behaviors (email bombing her email account [disruption] and sending forged posts that defamed her character). This is an example of retaliatory harassment—she was harassed because she did something online (in this case, exposed a scam) that angered someone else.

In contrast, the harassment in examples 8 and 9 was unprovoked. Both incidents took place in online chat environments.

Example 8:

A participant on LambdaMOO in the guise of an evil clown named Mr. Bungle used a "voodoo doll"—a computer program that creates an effigy of another user—to force legba and a nondescript female character named Starsinger to perform violent sex acts on themselves in public. (Dibbell, 1993)

Example 9:

Two female friends chatting together on IRC were repeatedly sexually propositioned and verbally abused by two male chat channel administrators. When the women protested, the administrators kicked them off the channel. (Herring, 1999)

These two cases are clear examples of cyber violence (and non-prototypical examples of violence in the traditional sense), in that they involve non-physical (verbal) abuse that takes place entirely online. Although specific individuals were abused, the targets appear to have been selected opportunistically (women who just happened to be in the environment at the time). The perpetrators, a young male college student in New York City ("Mr. Bungle") and two young male IRC administrators, were 'average' (or in the case of the IRC administrators, more statusful than average) Internet users, and the former claimed he was merely experimenting with the new medium, effectively denying—like the psychiatrist who pretended to be 'Joan'—any intent to harm.

Examples 8 and 9 involve harassment in that the targets were repeatedly insulted and abused, even after they protested the offending behavior. In addition, actions symbolic of physical violence (virtual rape; 'kicking off' the chat channel) were performed against them. Although these actions did not have the physical consequences that the analogous real-world actions would have had, they were nonetheless distressing to the victims.⁹ It is notable that in all three harassment episodes, the female victims were abused in sexual terms, although their prior behavior did not involve any sexual component.¹⁰ However, harassment need not be sexual in nature in order to fit the definition of 'online harassment'.

Some Internet users maintain that harassment that takes place in chat rooms and discussion forums is not a problem that should be legislated, but rather is "only words", a manifestation of free expression. Users who do not like such behavior should "just press the delete key", or avoid the online environments in which it occurs. Such views are often advanced by people with libertarian politics, who feel that "unruly behavior" online is a small price to pay for individual freedom (Pfaffenberger, 1997). In contrast, the position taken in this paper is that online harassment constitutes abuse, and has harmful effects both on individual victims and on groups of users. As such, it should not simply be tolerated.

A number of common annoying online behaviors are not considered harassment (and hence not cyber violence) according to the definition given here, although in many cases the issues they raise are closely related. Hate speech targets groups but can also be defended as expression of views, rather than as simple abuse. Trolling indiscriminately targets naïve users, albeit usually not repeatedly so. Flaming tends to be issue-specific rather than a means of generalized harassment. Spam is not personally targeted; moreover, it is ostensibly not intended to offend or intimidate, but rather to persuade users to take certain actions, such as visiting recommended Web sites. However, any of these behaviors directed towards specific individuals and repeated, as an annoyance, after the recipient has made it clear that the behaviors are unwanted, would be considered online harassment.

Recommended responses

Since online harassment is often essentially unprovoked, short of choosing one's words carefully or avoiding communicating with strangers on the Internet, there may be little one can do to avoid it. However, targets of online harassment can choose to respond so as to minimize the harm to themselves and others.

Spertus (1996) identifies two categories of response, which she terms "technical" and "social". Technical means for fighting online harassment include blocking unwanted messages, for example all messages from a known harasser, using protocols such as email filters and 'kill files' (on Usenet) and 'ignore' commands (on MOOs and IRC). The advantage of filters is that they give users control over what they will see, while preserving others' free speech. However, filters are less than ideal, in that they are reactive rather than proactive (some harassing content must come through before the filter can be set), and in that they do not prevent others in a forum from seeing the harassing content; only the individual user is excluded, which can actually make matters worse.

Social means for fighting online harassment include sharing information about the harasser publicly, in such a way as to damage the harasser's credibility and reputation, making it less likely that he or she will be in a position to harass others in the future (Spertus, 1996). This may also be achieved through reporting the harassment to the perpetrator's system administrator or employer, or in cases where laws are broken, to the police (but cf. Gilboa, 1996, who could not get the police to take her complaints seriously).

The target's first goal, however, should be to get the harassment to stop. Online anti-abuse organizations such as WHO@ recommend responding once to the harasser with a polite request that the contact be discontinued, and ignoring him or her thereafter. Such organizations will sometimes intervene to stop harassment, including referring cases to law enforcement authorities. Anti-defamation cases have been brought to trial and judgments obtained against the perpetrators. However, with the exception of "defamation lawsuits against message board posters by companies wanting to silence their online critics" (Benner, 2002), legal action has thus far rarely (if at all) been taken against harassers who operate entirely within online chat rooms and discussion forums.

4. Degrading representations

The last category of cyber violence is degrading representations, defined as online representation of women in words or images that invites disrespect and/or harmful behavior towards women in general. Issues raised by such representations include degradation and objectification—in this case of women, although other groups could conceivably be represented in degrading ways as well. Such behavior is not a prototypical form of violence,¹¹ in that the representations are not created to harm a specific target; moreover, the harm that arises from them may be indirect and diffuse. Such cases are often discussed in terms of freedom of expression, offensiveness, and standards of decency. They are included as a form of cyber violence in that they constitute and can lead to assault against the well-being of women, both individually and as a group.

The first case of degrading representations involves textual representation in the form of a list.

Example 10:

An e-mail message was distributed across the Internet in November of 1995 by four male undergraduates at Cornell University. The message contained a list entitled, "Top 75 reasons why women (bitches) should not have freedom of speech." The misogynistic and often violent reasons included:

· Stupid says as stupid does (and is).

· When men whistle at them in the street, they should just shut up and obey anyway.

· If she can't speak, she can't cry rape. (http://joc.mit.edu/cornell/75.reasons.txt)

This list is degrading in that it represents women as stupid, subordinate to men (like dogs), and deserving of (sexualized) violence. Although its authors claimed it was a "joke", it taps in to, and arguably reinforces, sexist and misogynistic cultural attitudes. Many women were offended and angered by the message.

The second case involves both text and images. The "Babes on the Web" case was discussed extensively on the Internet in the mid-1990s.

Example 11:

In 1995, an American named Rob Toups put up a Web site called "Babes on the Web". The site consisted of unauthorized links to photographs of professional and academic women on the Web, who were rated by Toups on the basis of their sexual attractiveness. Some women whose pictures had been linked to the site without their knowledge were surprised to receive crude propositions from men who had seen the pictures. (Spertus, 1996)

Toups' site was degrading in that it sexualized and objectified women, rating them in term of their physical appearance, without their permission (and in some cases, despite their requests that he remove the links to their photographs). Toups compounded the offensiveness by adding text to his site to the effect that he was "exercising his First Amendment rights", and if women didn't like it, they could "go cry to NOW [the National Organization for Women]". The site also included an intimidating photograph of Toups pointing a large shotgun at the viewer. A number of women protested the site (Spertus, 1996).

In retrospect, "Babes on the Web" seems mild compared to the fare served up on the World Wide Web today. In particular, hard-core pornographic representations are more readily available now than in 1995. Pornography is defined as "material that combines sex and/or genital exposure with abuse or degradation in a manner that appears to endorse, condone, or encourage such behavior" (Russell, 1993). The overwhelming majority of pornographic representations (with the exception of those intended for a male homosexual audience) depict women.

Example 12:

25% of Internet users aged 10-17 were exposed to unwanted pornographic images in the past year. 8% of the images involved violence, in addition to sex and nudity. For example, a fifteen-year-old boy searching for information on wolves for a class project came across a bestiality site showing a women having sexual intercourse with a wolf. (Finkelhor et al., 2000)

This image is degrading in that it portrays women as sexually depraved (having sex with an animal). Finkelhor et al. (2000) found that a higher percentage of young Internet users reported being "very or extremely upset" at receiving unwanted pornographic images than at any other category of abusive online behavior. The children were also less likely to report such abuse to authorities, presumably out of a sense of embarrassment or guilt.

In principle, Web pornography that is sought out and enjoyed is not cyber violence, even (arguably) when it portrays violence against women. In the short term, such behavior is victimless—unless the consumer of porn then goes out and treats women badly as a result of having viewed pornographic images. In practice, however, it is difficult to draw a clear line between unwanted (potentially harmful) exposure to pornographic representations and the availability of pornography on the Web in general. Pornographic materials, including graphically-explicit hardcore images, have become ubiquitous to the point that it is difficult to prevent exposure to them. An innocent typographic error in a URL can lead to a porn site (some sites count on this), and searches for many common terms (such as 'girl' or 'wolf') pull up porn sites in large numbers (Finkelhor et al., 2000). Moreover, even non-violent pornography objectifies women sexually, thereby contributing to wider societal attitudes that make women more likely targets of violence (especially sexual violence) than men (Birsch, 1996). In these respects, pornography can be seen as related to cyber violence, even when it is consumed voluntarily.

Not all sexually explicit materials online constitute cyber violence. Non-degrading words or images that arouse people sexually (known as 'erotica') are not harmful as long as they are produced and consumed by consenting adults. Included in this category are logs of cyber sex interactions, live video images broadcast via CUseeMe to multiple participants (Kibby & Costello, 2001), and erotic images exchanged between lovers. The problem remains of how to limit access to such materials by children.

Recommended responses

It is currently fashionable to advocate the use of technological filters to limit exposure to degrading representations, especially to prevent access to porn sites by children.¹² Such a position takes the view that the offending materials should be allowed to remain available to those who wish to access them. However, filters work imperfectly, sometimes blocking legitimate materials and failing to block objectionable ones.

In cases where the materials are considered to have little or no redeeming value, users, individually and in groups, can protest the offending materials, complaining directly to the individual(s) responsible. In the case of the "Babes on the Web" site, several women complained to Toups, put up anti-Toups sites, and linked to one another's sites in a spontaneous, grassroots form of protest. Under this pressure, Toups quietly removed the site later the same year.

Finally, one can complain about offending materials to the perpetrator's Internet Service Provider or the organization hosting the Web site, in an attempt to get them to remove the materials, and perhaps to punish the perpetrators. In the "75 reasons" case, complaints to Cornell University administrators caused them to take action against the young men who had produced and distributed the list. However, the increasing pervasiveness of pornography on the Web may make it less likely that an individual site would be removed today as the result of complaints than might have been the case several years ago.

Further Issues

This paper has attempted to define cyber violence and classify four of its major types. The argument has been advanced that cyber violence is more difficult to recognize and resist than offline forms of violence, because it diverges from the violence prototype in several important respects. Identifying these points of divergence is necessary in order to understand what cyber violence is and is not, and ultimately, to determine what an appropriate societal response to it should be.

Of course, lack of familiarity with cyber violence is not the only obstacle to recognizing and resisting it. The Internet itself fosters and abets abusive behavior by rendering perpetrators more anonymous and less fearful of retribution than they would be in physical space. At the same time, computer-mediated abuse typically leaves a trace (an email message, a routing path pointing back to an IP number, etc.), such that most perpetrators who are reported are eventually identified (WHO@, 2002). Nonetheless, the perception of anonymity appears to be a disinhibiting factor that leads otherwise normal individuals to give expression to their aggressive impulses in situations where they might not otherwise do so.

Ideologies of Internet communication play a role as well, notably in defining what counts as socially (un)acceptable behavior online. Libertarian views promoting individual freedom of expression can be used to justify harassment (Herring, 1999), and probably contribute to people's willingness to put up with pervasive behaviors such as flaming, spamming, hate speech, and sexual come-ons (to say nothing of Web pornography, which is actively defended by free speech advocates¹³). In contrast, discourses that construct cyber violence as a problem often invoke ideologies of personal safety (in the case of stalking and harassment; e.g., Magid, 2000) and community standards of decency (in the case of degrading representations; e.g., Biegel, 1996). In an important sense, cyber violence must be legitimized discursively and ideologically before it can effectively be fought.

Finally, societal norms and expectations related to gender and violence can be an obstacle in the fight against cyber violence, just as they are in the fight against physical violence. Violence, especially when it has a sexual component, tends to be underreported due to feelings of guilt or embarrassment (Finkelhor et al., 2000; Thomas, 2002). Females, in particular, are taught to believe that they are somehow to blame if they are sexually aggressed; this social conditioning makes them less likely to report acts of sexual aggression. A deeper problem is that violence against females is so widespread, and manifested in such diverse forms, that it is considered "normal" by many females and males. Thus most teenage girls, if asked if they have ever been sexually harassed, are likely to respond "no", but when asked specific questions, are able to report numerous harassment incidents, which they take to be simply "the ways things are". Finkelhor et al. (2000) observed a similar response pattern from the 10-17 year old children they interviewed for their study of online sexual abuse. The naturalization of violence—especially, of violence against females—must be challenged before cyber violence can be identified and resisted.

A Comprehensive Response

The contextual factors identified in the previous section contribute to the pervasiveness of cyber violence, and inhibit attempts to resist it. While these factors are not inherently immutable, changing them will require time and concerted, collective effort. In the meantime, steps can be taken to reduce the incidence of cyber violence and its harmful effects.

Finkelhor et al. (2000) propose a comprehensive response to online abuse that involves active intervention at all stages of the abuse process:

Reduce the quantity of offensive behavior (e.g., mobilize people in a campaign to “clean up” standards of Internet behavior)
Avoid/shield people from its likely occurrence
Increase the level of reporting (e.g., encourage Internet “neighborhood crime watch” posters)
Provide help to protect those threatened or harassed from any consequences

Although it comes first in the list, reducing the quantity of offensive online behavior is the ultimate goal, and one that requires effort on multiple, simultaneous levels—societal, cybersocietal, and individual—to achieve. Mobilizing people to demand a "cleaner" online environment requires identifying and justifying certain behaviors as unacceptable. The present paper is one proposal towards meeting that goal.

Currently available recommendations focus mostly on the second and third stages of the process, avoidance of/protection from abusive behaviors, and reporting of such behaviors as can not be avoided. The present paper has suggested that in order to do this effectively, it is necessary to distinguish among different cyber violence types, as each raises unique issues and challenges—trust in the case of online misrepresentation, privacy in the case of cyberstalking, damage to resources and reputation in the case of online harassment, and degradation in the case of pornography—calling for different avoidance and response tactics.

Last, as the incidence of cyber violence increases, online organizations such as WHO@, SafetyEd International, WiredPatrol (formerly CyberAngels) and CyberTipline are stepping up their efforts to assist targets of online abuse. Working together with search engines such as Yahoo! and Internet Service Providers such as AmericaOnline, these organizations provide in situ intervention to stop abuse, referrals to legal counsel and law enforcement agencies, information and advice, and statistics about online abuse. They, too, must make judgment calls about what is and is not abuse, in order to allocate their resources effectively.

Interestingly, WHO@ also excludes hate speech, flaming, spamming and trolling from their definition of online abuse, as was proposed here. This raises a larger question of how to deal with behaviors that fall outside the definition of cyber violence proper, but that are nonetheless widespread and problematic. Viewing related but "less serious" behaviors together with cyber violence sheds light on the larger forces—technological, ideological , and societal—that shape the online environment as a social space in which "bad behavior" occurs. At the same time, there is a need to set the bounds of what constitutes actionable abuse. Defining cyber violence is a first step towards meeting this goal.

Missing from the response strategy proposed by Finkelhor et al. (2000) is mention of a need for public policy, including legislation on online abuse, a point which other anti-online-abuse advocates raise forcefully (WHO@, 2002). Ten years ago, almost no such policies or legislation existed, constituting a lack of deterrence, and making it difficult for victims of cyber violence to seek redress. Classification of behaviors plays an important role in the legal realm, e.g., in lawmaking and law enforcement.

The present paper is not intended to determine what the specific legal consequences of any type of cyber violence should be, nor indeed to suggest that all forms of cyber violence constitute punishable crimes. Some of the examples discussed here have the status of criminal activity, others of currently legal but socially unacceptable behavior, and still others of behavior that is unacceptable to some, but tolerable (or enjoyable) to others. Still others are ambiguous cases that do not clearly fall into one category or the other. Ultimately, how a society responds to acts of cyber violence should include a consideration of the extent and the seriousness of the harm produced, and of community standards of (in)acceptability regarding classes of abusive behavior. I leave this as a problem for further discussion.

Notes

¹ For example, a participant at the UNESCO Chair Symposium on 'Women's Rights, Cyber Rights' held at Sookmyung Women's University in Seoul, Korea on May 31, 2002 noted that a Korean Internet Service Provider which recently implemented a system for users to report abusive behavior online is receiving one million reports of such behavior per month (Shim, 2002). In the United States, WHO@ (2002) reports that it receives 100 reports of online harassment per week, 95% of which are legitimate cases.

2 See, e.g., Kiesler et al. (1984).

3 Women were the perpetrators in 29% of the cases, and in 7% of cases, the gender of the perpetrator is unknown (WHO@, 2002).

4 Of course, all of these stereotypes about violence can be (and regularly are) broken in the real world.

5 Similarly horror-inducing in this archetypical sense is the case of a young woman, Amy Boyer, who was murdered in 1999 by a former classmate who had never spoken to her, but who followed her every move through information obtained on the Internet (http://www.unc.edu/courses/law357c/cyberprojects/spring00/cyberstalking/cyberstalk/cases.html). This is a case of cyberstalking in the classification system presented here.

6 Cyber sex is online sexual activity involving the exchange of erotic words or images for purposes of mutual arousal. At issue in the case of 'Joan' (and even more ambiguously in the cases reported in McRae, 1996) is whether the deception that can arise when someone discovers that their cyber sex partner is a different gender (or sexual orientation, or race, or age, etc.) than they were led to believe can be considered a form of abuse, and the individuals who misrepresented themselves held morally accountable.

7 This outcome underscores the interpretive challenges posed by definitions of violence based in whole or in part on the notion of perpetrator 'intent'.

8 The Jake Baker case becomes additionally problematic when compared with the cyberstalking case that resulted in the murder of Amy Boyer, described in note 5. In both, college-age males posted materials on the Internet describing their violent fantasies (including murder) about a (former) female classmate. On the basis of the Internet evidence, Jake Baker appeared no less likely to act than Amy Boyer's murderer.

9 See discussions of each case in Dibbell (1993) and Herring (1999).

10 This is consistent with a larger societal pattern, reflected in the vocabulary of the English language, according to which females are demeaned through their sexuality (Schultz, 1990).

11 But cf. Catharine MacKinnon and Andrea Dworkin's critique of pornography as violence against women (1988).

12 Cf. the recently-overturned Children's Internet Protection Act (CIPA), which would have withheld US government funding from libraries that did not use filtering software to restrict access to pornographic materials on the Web (American Library Association, 2002).

13 See, e.g., Strossen (1995), and for a counter position, Russell (1995).