Introduction

MonitorA is an observatory of online political and electoral violence against candidates, a project carried out by AzMina Magazine and InternetLab that started in 2020. On this edition, we worked in partnership with Núcleo Jornalismo, and received funding from Reset, Luminate and IDRC. Here we present:

i) data collected by MonitorA from September to November, 2022;

ii) analyses on the applicability and effectiveness of the Law on Gender-based Political Violence during the 2022 elections carried out by Núcleo de Inteligência Eleitoral (Electoral Intelligence Center), under the coordination of Lawyer Samara Castro and, finally,

iii) dialogues with data about TikTok, part of a netnography elaborated in consultancy with Lux Ferreira and InternetLab.

The analyzes of the collected data were carried out by the three organizations: the data analysis tool was developed by Núcleo Jornalismo, and the lexicon of researched terms, by InternetLab and AzMina Magazine. MonitorA collected comments related to 198 women candidates and 24 man candidates to state and federal executive and legislative offices, on Twitter, YouTube, Facebook and Instagram platforms. We hope that this report will shed light on the need for an intersectoral and intersectional look at gender-based political violence, thereby contributing to the formulation of measures to prevent and to seek to eradicate political violence before, during and after electoral periods.

InternetLab

is a research center in law and technology. It develops research on human rights and technology oriented to public policies, and publishes books, reports, articles and guides. Since its foundation, it has been working on the topic of online gendered violence, as well as addressing issues targeting women and privacy, and online feminist activism. Learn more: internetlab.org.br

Revista AzMina

is a non-profit institute that combats various types of violence targeting Brazilian women. It produces a digital magazine, maintains an application to combat domestic violence, and uses a tool to monitor women's rights in the National Congress, besides conducting campaigns to combat violence against women in Brazil. Learn more: azmina.com.br/instituto-azmina/

Núcleo Jornalismo

is an initiative that covers the impact of social media on people's lives, exploring both the traditional journalistic production and new content and technology forms. The site was created with the mission of producing data-based investigations and bringing more transparency to the government, to political debates and to public policies in different spheres and locals. Learn more: https://nucleo.jor.br/

Núcleo de Inteligência Eleitoral

is a project coordinated by the Lawyer Samara Castro and it is composed by lawyers Nathália Criscito, Roberta Battisti, Camila Tsuzuki and Matheus Valente. It operates in the articulation, consulting and provision of strategic litigation services to civil society organizations. It aims to develop and elaborate coordinated actions to combat misinformation, extremist groups, the misuse of social media and messaging applications for electoral campaigns, and other electoral irregularities in a strategic and effective way.

Executive Summary

MonitorA was created with the goal of monitoring the 2020 municipal elections, analyzing how the internet is used by candidates and voters as a space for engagement. In the 2020 edition, we identified that political violence on the internet is directed in a very specific way to certain social groups marked by gender, race, sexuality, generation, etc., impacting especially the exercise of political life of women, black people, the elderly, and LGBTQIAP+. According to the results of the 2020 survey, women candidates are not criticized for their political actions, but for being who they are or for not acting as they are supposed to act. Commonly, the attacks reduced the intellectual capacity of female candidates, questioned the character of their private lives, and made fat-phobic, racist, and misogynistic comments about their bodies. On the other hand, men, with the exception of the elderly, transgender, and gay men, were not targets; the comments focused on their political actions and the questioning of their ideologies.

Since 2020, the political scenario has changed: the Law of Gender Political Violence (Law No. 14.192/2021) has been approved, and the crime of political violence is now typified by the Law of Crimes against the Democratic State (Law No. 14.197/2021), the discussions about political violence have gained strength in the public debate, with the participation of candidates and other important social actors, such as the third sector, platforms, and the political parties themselves.

Given this new context, the purpose of the monitoring changed: it was no longer necessary to demonstrate that the problem of gender political violence existed, but it was necessary to observe it and to complexify what we had to say about it. To this end, the 2022 edition of MonitorA was guided by the following questions:

1. How can we defend the possibility that politicians continue to be charged without implying that it is legitimate for female candidates to suffer harassment, psychological and misogynistic violence, or other types of violence?
2. Is all hostile language toward candidates political violence? Or can we sometimes be faced with an offense that is moral, which can be considered an attack, but is not necessarily political violence?
3. Should all hostile and potentially disrespectful comments be removed from the platforms?
4. How can one differentiate an attack from an insult, even though the insult may be harsh or sound disrespectful?
5. How can we ensure that male and female voters can show discontent with the candidacies, and at the same time ensure the safety and integrity of the candidates and the democratic process?

Offensive Content Classification

To answer these questions and to complexify the analyses about gender political violence on the internet, MonitorA 2022 brought a methodological change regarding what we consider offensive content. This year, we began to differentiate what we consider attacks and insults. The distinction between the two categories that use hostile language allowed us to delimit which are the contents that exceed the limits of freedom of expression, transforming the political environment into an inhospitable scenario for the entry and permanence of candidates, especially, female candidates in institutional politics.

We consider insults as content that has hostile and disrespectful language, but that are not attacks per se, even though they may be considered harsher than a simple criticism. Among the contents classified as insults, we could not perceive criteria directly associated with gender, race, region, religion or sexuality inequalities. Which means that they are used similarly for men and women, for white, indigenous and black people, for trans and cis people. We believe that, in principle, insults should not necessarily be removed by platforms.

On the other hand, attacks are characterized by attempts to inferiorize candidates. Often, historically aggressive terms are used against people belonging to historically marginalized groups, such as women, black people, indigenous people, LGBTQIA+ people, and people with disabilities. We believe that attacks should be removed from platforms.

The definition of insults and attacks should not lead to the understanding that insults should be naturalized in the political game. Massively repeated insults can have even more forceful impacts on the decision of whether or not candidates continue in institutional politics, especially for women. An isolated attack, on the other hand, may have less effect than massively published insults.

It should also be noted that this differentiation applies to political figures, that is, people who run for or act in institutional politics and who therefore occupy a central place in the very functioning of democracy.

Data Collection

MonitorA 2022 took a multi-method approach, with different research fronts. In addition to monitoring candidacies on Twitter, Instagram, Facebook, and YouTube based on a lexicon of offensive terms, a netnography was also conducted on TikTok and a monitoring of cases of gender political violence filed during the election campaign.

To monitor gender-based political violence on social networks, we selected 198 profiles of candidates, among women and men, from all regions of the country, for the positions of State and Federal House of Representatives, Senate, Presidency and Vice-Presidency. We chose candidates who were distributed among different ethnic-racial backgrounds, ages, gender identity, sexual orientation, regionalities, and diverse ideological positions. The monitoring was done by collecting posts, comments and other interactions made by users on Twitter, Instagram, Facebook and YouTube. The collection was carried out in an automated way, through programming with Python language. The data collected was filtered from a lexicon of offensive terms, built and updated from previous research experiences, and refined from the data collected in this edition of the project.

In total, 1,697,774 tweets; 892,500 comments on Instagram; 167,095 comments on YouTube; and 65,761 comments on Facebook were collected. This data was filtered based on the lexicon of offensive terms and then manually analyzed.

In relation to netnography, the consultancy of Lux Ferreira was hired, which focused on the narratives, patterns of content production, mobilization of tools and reactions of other users, in dynamics of interaction by comments on the platform for sharing short videos.

Finally, a monitoring of the applicability and effectiveness of the Gender Political Violence Law during the 2022 elections was carried out by the Electoral Intelligence Center.

The data found

Below are the main findings of this report:

• Based on TikTok's ethnography, we noticed that if in other social networks such as Twitter, offenses made to candidates and political personalities in their posts and profiles are recurrent, the functioning of TikTok's algorithm tends the opposite - presenting content to people who agree with it in its For You area, not being as affected by the conduction of themes and people to the trending topic as on Twitter. With this, disagreements and offenses are scarcer on the short video platform, and are limited to statements of vote.
• Michelle Bolsonaro and Janja Lula da Silva, although not candidates, were extremely relevant figures in the elections for the Presidency of the Republic. To understand how political violence affected the then first lady candidates in the week before and the week after the first round of elections (between September 26 and October 9), we monitored Janja's Twitter and Instagram profiles and Michelle's Instagram profile. When analyzing the offensive content directed at the candidates' wives, we noticed that religiosity and notions of morality provided the keynote of the offensive narratives against Janja and Michelle on Twitter and Instagram. In the comments with hostile content directed at Michelle, religion appears mainly linked to morality, to a religious duty that they accuse the former first lady of not fulfilling. Michelle is understood, according to these hostile narratives, as a person who uses God's name in vain, as a woman who does not act as an evangelical woman should act. The most offensive comments with a religious content are those that make reference to Bolsonaro's presence in a masonic store. When analyzing the hostile narratives directed at Janja, a much higher degree of aggressiveness is noticeable. The attacks and offenses published against the first lady focus mainly on religious intolerance against religions of African origin. This intolerance gains misogynistic contours in cases that refer to the pombagira in a pejorative way and to Janja's sex life. These cases show that political violence is not restricted to those running for office, but may also affect people directly connected to the candidates.
• Male candidates belonging to historically marginalized groups can also be targets of political violence, and it is precisely from social markers of difference - such as age, being GBT+ or non-white - that violence manifests. This dynamic demonstrates a relevant difference between male and female candidates, since, unlike women, the men analyzed here were not attacked for being men, but for belonging to other minority groups. Candidates Kim Kataguiri and Fernando Holiday were the main victims of this kind of attack. In Kataguiri's case, the offensive comments revolved around his Japanese descent, which was repeatedly brought up as a negative aspect, leading to stereotyped and derogatory comments regarding cultural aspects of this community. Fernando Holiday, on the other hand, was frequently attacked for the political positions he defended as a black man. Most of the offenses tried to point, through a racist and violent vocabulary, supposed inconsistencies between Holiday's speech and his ethno-racial belonging.
• The expression "you are a disgrace" had a high incidence in the collections, appearing in different discourses, such as "you are a disgrace to black people"; "you are a disgrace to women", etc. It is worth remembering that this type of narrative did not occur without context. During the presidential debate on August 28, 2022, then-re-election candidate Jair Bolsonaro (PL) attacked journalist Vera Magalhaes, claiming that she was "a disgrace to Brazilian journalism." The use of shame [shaming] mechanisms is a common practice in misogynistic attacks, and can also be associated with discredit, since comments of this kind are often associated with claims such as "you are worthless," reiterating the intention to discredit.
• The 2022 elections had the highest participation of trans candidates ever recorded.  MonitorA followed the social networks of 11 of the 78 registered candidates in this election, analyzing both offensive content and the main narratives that were built about these candidates. We observed that much of the political debate that trans and transvestite candidates tried to engage in on social networks during the electoral campaign ended up being hijacked by ideological-party conflicts, political violence, debates about public safety and disputes over the representativeness of historically minority groups. The agenda related to issues important to the candidates, therefore, failed to make it into the political dialogue when they were related to issues understood as "identity agenda".  Users' interactions with female candidates reproduced the scenario of national political polarization, with the defense of progressive parties and candidates being the main key to attacks on trans women who were trying for positions in the Legislative branch. This fact made transphobic comments become an additional layer of violence, opposing the messages of support and welcoming to the candidates, which represented an important part of the interactions.
• In relation to the monitoring of the applicability and the level of effectiveness of the Law of Political Violence against Women during the electoral period, we identified obstacles, both in accessing data and in the effectiveness of the law to protect women candidates. From searches in the PJe and TSE's unified consultation, only twelve cases were identified, filed during the electoral period: four of them were related to irregularities committed by the parties, and eight were cases of harassment or threats directed at female candidates or female representatives. The number of identified cases was lower than we supposed we would find, since data from the National Council of Justice (CNJ) informed that in the last 15 months 7 cases of gender political violence were registered every 30 days. To confirm the number of cases filed, we sent letters to the TSE and to the MPF, with requests about the number of complaints involving gender-based political violence. TSE denied access due to the risk of LGPD infringement. The MPF, on the other hand, forwarded a report with 62 cases that are being processed by the body. However, these cases could not be analyzed because, currently, there is no system that allows to follow the entire process with access to more detailed information.
• Among the cases identified, most of them did not make specific reference to gender political violence on the internet. Of the 12 cases found, 4 of them referred to irregularities in the distribution of party funds. These actions deal with three themes: competence for the judgment of these cases, investigation of fictitious candidacy to fulfill the minimum percentage of women candidates required by law, irregularity in the distribution of positions to a federal deputy, and lack of transfer of resources from the Special Fund for Campaign Financing to female candidates. Eight cases were identified that involved offensive attacks on women candidates based on art. 326-B of the Electoral Code. Of these, four cases involved gender political violence in face-to-face settings during the conduct of the electoral campaign. Only one of the cases identified occurred on digital platforms. The other three cases are still in the investigation phase, with the establishment of the inquiry and the crime report. Because they are at an early stage, until the publication of this report it was not possible to have detailed access to the information of the processes.
• Despite the unquestionable achievement from the enactment of a law destined to look specifically at and take care of women's candidacies, in order to ensure the space of this portion of the population in elections, it has become evident that the current scenario does not corroborate with the intended effectiveness. The plurality of channels able to receive reports of violations of the legislation disperses news of crimes by the victims. The consequence is the delay in the initiation and progress of investigations during the short electoral period. Furthermore, we did not identify any cases between August 16, 2022 and October 30, 2022 (the 2022 election period) whose investigation was concluded and which resulted in an electoral criminal action before the courts. In other words, even if the victims reported the eventual violence they suffered during the 2022 election period, none of them will effectively have the legal protection brought by Law 14.192/21, because the current procedure is not fast enough to ensure it.

Recommendations

Platforms and Internet Application Providers

1. Commitment to the development policies and guidelines for protective usage in face of political violence
2. Improving practices of data access and transparency for researchers
3. Commitment to protective enforcement and to transparency of usage policies and guidelines
4. Improving policies aimed at public figures and political office holders
5. Improving platforms' internal reporting channels
   

Legislative Branch

1. Improving legislation on gender political violence, with the inclusion, on the Law of Gender Political Violence, of protective measures and of liability of political parties

Judiciary Branch and the Public Prosecutor's Office

1. Improving access channels and transparency mechanisms
2. Collection and compilation of data on gender-based political violence
3. Public Prosecutor's Office: improving reporting channels
4. Building strategies, within the scope of the Electoral Justice, to confront gender-based political violence and misinformation

Executive Branch

1. Inclusion of combat against political violence in the governmental agenda
   

Political Parties

1. Improving supporting tools for women candidates in cases of political violence, creating mechanisms for reporting and for internal liability
   

Civil Society

1. Creating and strengthening networks about elections and violence (political and electoral, gender, race, etc.)

Women, black, indigenous, LGBTQIAPN+ and elderly candidates are disproportionately targeted by online political violence. This was the main finding of the first edition of MonitorA. According to the results of the 2020’s survey, women candidates are not criticized for their political actions, but for being who they are or for not acting as they supposedly should. Commonly, the attacks questioned the intellectual capacity of the candidates, the morality of their private lives, and made fatphobic, racist, and misogynistic comments about their bodies. Men, on the other hand, except for the elderly, trans and gay candidates, were not seen through the same lenses: the comments focused on their political actions and on questioning their ideologies.

In 2022, we found a different scenario: Firstly, in 2020, there were municipal elections and, in 2022, general elections. The challenges to choose which candidates would be monitored, considering an intersectional perspective, were faced, this time, from previous gaps and experiences. In addition to this methodological maturation, we also needed to consider the limitations regarding the amount of engagement candidates for the general election received. In municipal elections, although the number of candidates is larger, profiles tend to receive fewer comments, and there tend to be fewer mentions to candidates on social media. When it comes to general elections, we find a scenario of fewer candidacies, but with stronger individual engagement. In this sense, we must adapt to each of these realities.

As an example, in 2022, we did not count on technical and logistical structure to monitor candidates such as Lula and Bolsonaro, who received many comments and mentions on social media, which did not allow us to compare them to the candidates Soraya Thronicke and Simone Tebet, as we had done with candidates for mayors and their opponents in 2020. Collecting data from these two candidates would take up much of our processing and storage capacity, and would demand more time than we had to develop the project as quickly as necessary.

Thus, on this edition, the list of men candidates to monitor was reduced, and we gave priority to diversifying the profiles of women candidates. Another important point is that, along the data collection, the access to Twitter’s API had no significant changes. Data from Meta platforms, however, were more difficult to collect, when compared to the collection performed in 2020. For this reason, data about Instagram and Facebook could only be analyzed halfway through the survey, since only then could we reach a statistically significant amount of data. The comparisons between platforms that we could have made along the election period were, thus, compromised.

Although, in practice, we have faced the methodological issues mentioned above, we must emphasize that the phenomenon of gender-based political violence - articulated to violent discourses directed at other historically marginalized groups - has been observed by society from new perspectives. We have before us a social scenario in which we see a greater engagement of society to face this phenomenon. Civil society organizations and the Legislative power, for example, stood out as sectors that understand violence directed at candidates and elected politicians as an obstacle to electoral integrity, and as a barrier to women's political careers. Such understanding was one of the factors leading to the creation and revision of specific legislation on the subject.

The 2022 election was the first to be held after the enactment of the Law on Gender-based Political Violence (Law No. 14,192/2021) and the categorization of the crime of political violence by Law on Crimes Against the Democratic Rule of Law (Law No. 14,197/2021). With a specific legal framework and with the action of public authorities, the debate around gender-based political violence has become more complex. Political violence is now a concern connected to other social phenomena that cross the debate about electoral integrity, such as: misinformation, platform ad rules and transparency of platform decisions regarding content moderation and, more generally, the application of their terms of use.

The connection with other topics and the consequent expansion of the attention to political violence brings us not only the responsibility of demonstrating the concreteness of the political violence phenomenon, but also the responsibility of refining our view as to defend the combat of political violence without falling into the trap of silencing voters: the demands from users to their candidates must have room to happen.

This idea began to concern us when we understood that hostility, to some extent, may be part of the political game, and that not all hostile language can be considered an attack: at times, we are facing merely an insult.

In this sense, if, as stated above, it was no longer necessary to demonstrate that the problem existed, it was still necessary to observe it and complexify what we had to say about it. The following questions have guided us along this edition of MonitorA:

1. How can we defend the possibility that politicians continue to be questioned without finding it legitimate for candidates to suffer harassment and psychological, misogynistic or other kinds of violence?

2. Can all hostile language directed at candidates be considered political violence? Or, at times, are we facing a moral offense that can be considered an attack, but not necessarily political violence?

3. Should all hostile and potentially disrespectful comments be removed from the platforms?

4. How can we differentiate an attack from an insult, even if the insult is harsh or sounds disrespectful?

5. How can we ensure that voters are able to show discontentment with the candidacies and, at the same time, ensure the safety and integrity of candidates and of the democratic process?

To begin the process of answering these questions, or, at least, to deepen the debate and raise new questions, we started by differentiating what we consider an attack and an insult. The distinction between the two categories that use hostile language allowed us to distinguish which contents go beyond the limits of freedom of speech, transforming the political environment into an inhospitable scenario for the entry and permanence of candidates, especially women candidates, in institutional politics.

Insults versus attacks: When should contents be removed from platforms?

We consider an insult every content characterized by hostile and disrespectful language, but which is not properly an attack, even when it is harsher than mere criticism. Among the contents classified as insults, it is not possible to notice criteria that are directly associated to inequalities of gender, race, region, religion or sexuality. That is, they are used in a similar way for men and women, for white, indigenous and black people, for trans and cis people.

These are, therefore, common terms to different profiles of candidates. Within insults, we find words that intensify the criticism and discontentment that users feel in relation to candidates. We believe that, at first, insults should not necessarily be removed by platforms. As much as they can create hostile environments, they, at the same time, allow voters to make demands to candidates. Insults such as "thief", for example, may be related to a complaint or demand based on a suspicion of corruption. Candidates and politicians can and must be questioned by users.

Attacks, on the other hand, are characterized by attempts to inferiorize candidates. Aggressive terms are often used against people belonging to historically marginalized groups, such as women, blacks, indigenous, LGBTQIA+ and people with disabilities. There is also the possibility of attacks using moral offenses to refer to candidates. These offenses, depending on the context, can also be considered attacks.

In the case of attacks, the discourses produced create a hostile environment and often discourage some people from working in institutional politics, or influence the way third parties read women candidates particularly. Attacks tend to be done in public. The humiliation of a candidate and the attempt to tarnish her image are ways of trying to push her away from the political scene. Here, it is common to use resources such as dehumanization, sexual offenses, harassment, shaming for alleged moral defects, attacks on political or religious ideology, intellectual discrediting, incitement to physical violence and threats, in addition to fatphobia, transphobia, lesbophobia, misogyny, homophobia, biphobia, ableism and racism. We, therefore, believe that attacks should be removed from platforms.

It is also necessary to consider that different kinds of attacks can overlap in the same discourse, victimizing the candidates in different ways. An example is when someone is called an "evanjegue" (expression resulting from the combination of evangelical and donkey), a term that has appeared a few times in this issue. In this case, we have a mix of an attack to religious ideology and of dehumanization. In "communist bitch", we find an overlap of misogyny and an attack to political ideology. The word “communist” would not be considered an attack or an insult if it appeared alone, but when "communist" and "whore" are articulated, we understand that there is an attack, since there is an association of an ideological position and misogyny. The same would be true for "conservative slut". "Slut" in this case, would enter the category "sexual offense", but this category alone would not be sufficient to understand what is being said, after all, "slut" articulates with "conservative", thus being both an offense related to sexuality and political ideology. It is necessary, therefore, to consider the articulations of different kinds of attacks that may arise.

The categories of insults and attacks are only a methodological framework that has allowed us to outline possible ways to distinguish the different levels of hostility arising on social media or, in other words, what we believe should be removed or should remain on platforms. In practice, however, attacks and insults are not necessarily self-excluding. The same tweet or comment can contain both an insult and an attack. For classifying hostile content, we have developed a guide in which we define potentially offensive terms as an "attack" or an "insult". When a term was considered an attack, we also registered what kind of attack that was. It must be said that the kinds of attacks can also vary with context.

This guide, however, was questioned and the classes were modified whenever we understood that a comment or tweet was not clear enough for those analyzing it. If there were doubts whether it was possible to fit them in the previously established classes, they were debated by the research team. Some variations needed to be thought individually, which made us consult each other throughout the research process. The many consultations and variations, as well as debates around the meanings, warned us about how subjective and culture-based the classifications between what was in fact an attack or insult were. Here, the old caveat about the importance of context remains valid.

As ideal classes, however, the differentiation between insult and attack was important for us to think about what, at first, should or should not be removed from the platforms. Here are some examples of terms and their classes:

What defines someone as a public figure?

Some digital platforms as Twitter and TikTok consider that public figures should have their pages observed through other lenses.  Profiles of public figures, from this point of view, need to be more tolerant to hostile content.

What defines, however, a person as a public figure? Is it the number of followers? The engagement the pages achieve? Having a verified badge? In addition to the questions surrounding the classification of who can be considered a public figure, we can also ask:

i) Should politicians be inserted in the same rationale as other public figures?

ii) Should people who are part of politicians’ families be understood as public figures?

iii) What weight should social markers of difference, such as gender, race, ethnicity, religion and sexuality, have in defining the treatment of public and/or political figures in social media?

Apparently, people who take on political life can be treated as public figures by platforms, just as soccer players are. There is, however, a question that must be raised: by taking on a public office, the social influence of a personality who acts in politics is the same as that of a person who professionally competes in sports? We know this is untrue. The employment of platforms’ terms of use, in the case of presidents, for example, tends to be more flexible for some heads of state than for others. Take, for example, the cases of false information about COVID-19 and how long platforms took to act when misinformation came from a political leader from the Global South, compared to information coming from leaders from the Global North. The problem was similar, but the search for a solution took longer in one case than in the other.

In addition, unlike other public figures, people who run for office or who work in institutional politics occupy a central place in the very functioning of democracy. Politicians and candidates have duties, powers and responsibilities that derive from their participation in the structure of democracy. These obligations and responsibilities related to public offices do not resemble the democratic obligations of other public figures. In this sense, it does not seem that both categories should be subject to the same rules, since each of these categories have different impacts on the democratic sphere. Thus, a hostile political space can represent, in this context, an even greater barrier to the democratic participation of historically marginalized groups, which may be more vulnerable targets to hostility in social media.

The criterion for treating people who hold offices as public figures seems rather problematic. There is another point that needs to be raised: when an individual from a historically marginalized group is also a public figure, on the terms of the platforms, the permission for their profile and/or page to be more hostilized would not imply the acceptance that it is legitimate to treat people belonging to historically marginalized populations in a hostile manner? If a trans candidate, who is a public figure in the terms of platforms, is more offended than a trans woman who is not a public figure, does it not mean that it is valid to treat trans people in general in such a way? The simple classification of who is in fact a public figure, and how these profiles are more open to insults and/or attacks, is itself quite controversial.

Moreover, in the case of political violence targeting politicians, it is necessary to consider that people victimized by this social phenomenon are not restricted to those who directly run for political offices. There are also indirect victims of political violence: children, husbands, family members in general, friends and wives of political personalities can also be indirectly attacked. Being a spouse of a candidate, should one also be treated based on the policy that platforms direct to those who are considered public figures?

Here, each case seems to be unique.

Let us consider two different contexts: in the first, there is a municipal election. Who is running is a very popular candidate in the city, but he is married to a woman who is quite distant from political life. She only accompanies him in some situations, but never gives speeches; her social media are private and there is seldom any political content in them. In the second case, there is a candidate running for general elections. The first lady postulant acts actively in the campaign spaces, and uses her social media to help her wife.

Thinking about both cases sheds light into a basic criterion: not always a relative or spouse of a political figure is socially placed as a political personality. The criteria for defining attacks and insults that we postulate on this edition of MonitorA, therefore, would not apply to the person who is not a political figure, only to the one who actually is.

When we looked at the profiles of Janja Lula da Silva, the current first lady, on Twitter and Instagram, and the profile of Michelle Bolsonaro, the spouse of former President Jair Bolsonaro, on Instagram, we doubted whether we should think of them as we think of candidates. The first questions were: should offenses such as "thief" and "corrupt" be treated as insults – as we were doing in relation to the candidates - or as attacks? At first, we considered that Janja and Michelle should not be measured by the same standards we used for the list of monitored profiles.

A few months after we made this decision, we realized that considering Janja and Michelle as public figures did not imply in saying that all first lady postulants should be treated the same way. These two, specifically, are public figures who have actively engaged in campaigns and conducted rallies on their own. If there is a way of thinking about their spouses and other candidates, this way should also be applied to them.

Stating this, however, does not cancel out the following facts:

i) the notion of public figure that treats equally famous singers and people who act in institutional politics does not seem to make sense;

ii) the formulations of insults and attacks we have made here are directed at candidates and politicians who are actively engaged in the public sphere;

iii) policies designed for public and/or political figures should consider that these people may belong to historically marginalized groups, which means that an adequate look at the consequences of political violence on these individuals should also be considered.

Gray zones: not all fit within insults and attacks

The definition of insults and attacks should not lead to the understanding that insults should be naturalized in the political game. Massively repeated insults can have even more severe impacts on the decision of candidates to continue or not in institutional politics, especially for women. An isolated attack, on the other hand, can have a lower impact than massively published insults.

The exclusion of content through moderation is not, therefore, the only solution to be adopted by platforms or to be required by researchers. The greater autonomy of users and proactive actions of platforms should occur jointly whenever possible. A good example of that practice happens on Instagram and Twitter, which allow users to put filters for words they would not like to read on their profiles.

Moreover, it is necessary to consider what seems to go beyond both the notions of attack and insult. One of the common ways to offend candidates, on 2022’s edition, was to make moral criticism to candidates. Some recurring terms we found were "junkie", "pothead" and "immoral". As we see, they are not related to terms commonly directed at people who belong to historically marginalized groups, and, for this reason, they seem to be categories that could be used against any candidate, regardless of gender, race and/or ethnicity, sexuality, etc.

At first, we tended to think that moral offenses, as well as insults, should not be removed from platforms. Then, we thought about the kind of impact offenses of this order can have on candidates’ lives. Considering that, in recent years, we have experienced in Brazil a process of constant moralization and judgment of ways of life – typical of moments when far-right governments are in power –, is it valid to disregard the impact of moral offenses, especially massive ones, on the lives of political figures?

The answer is not simple. We believe that some of the moral offenses could be removed because they fit in the concept of bullying of some platforms. At the same time, our understanding of attacks, that is, contents that should be removed from social media, was more in line with understandings of what would be hate speech according to the platforms’ policies. Does the phenomenon of political violence combine the definitions of hate speech and bullying?

Both hate speech and bullying are umbrella-terms that seek to cover different kinds of violent speech that may be present on platforms. Would political violence, then, theoretically, be covered by already existing policies on the terms of use of several platforms? It is not that simple. Policies like these are designed for global contexts, but local contexts impose the need for other kinds of observations of what is socially occurring.

Moral offenses are a great example to consider. The current political scene in Brazil demonstrates that offenses articulated to moral views on certain people, especially those who are part of historically subalternized populations, can result in irreversible damage to their lives, leading to the silencing of candidates or to their withdrawal from political spaces. Moral offenses are often accompanied by other social meanings, and here lies our concern about the existing policies not being able to take into account the complexity of political violence. Will a candidate from the countryside who has been called a junkie suffer the same impact as a candidate from a progressive niche of a metropolis who has been called a pothead? The meanings of moral offenses vary depending on who receives them. In this sense, choosing new concepts and meanings of violence that are linked to specific contexts seems to be fundamental.

To summarize, we question the risks of moral offenses entering the criteria of contents to be removed by platforms. In order to be able to include moral offenses in the category of attacks, should we have a specific policy for political violence? How can we observe, in case this policy is created, if it is being applied? Can existing bullying and hate speech policies mitigate the effects of political violence? These are difficult answers if we do not have access to data that allows us to make these analyses, and/or if there is no transparency about how these issues have been handled by the platforms in which candidates and voters are present.

In order to conduct the analyses of MonitorA 2022, we selected 198 profiles of candidates, among women and men, from all regions of the country. We chose candidates belonging to different ethnic-racial backgrounds, ages, gender identities, sexual orientations, regions and varied ideological positions.

The monitoring was carried out through the collection of posts, comments and other interactions made by users on Twitter, Instagram, Facebook and YouTube. The collection was performed in an automated way, through programming with Python language.

In relation to TikTok, a netnography was conducted on the platform throughout the electoral campaign period. Netnography, a term coined by Robert V. Kozinets (2015), is a set of practices and research methods focused on social phenomena that occur in the online space. In the case of TikTok, the research conducted by Lux Ferreira focused on narratives, patterns of content production, use of proper tools and reactions of other users, in dynamics of interaction through comments on the short video sharing platform.

The lexicon of offensive terms, that is, words and terms often used in sexist hate speech on social media, has been updated from previous research experiences, and refined based on the data collected in this edition of the project. This lexicon was used to filter the results obtained, for an in-depth study of violent narratives directed to candidates.

Developed in different stages, the methodology was carried out as follows:

Methodology steps

Selection of monitored candidates

Update of the lexicon of offensive terms

Collection of data on Twitter, Instagram, Facebook and YouTube

Creation of a viewing platform

Data analysis

Publishing of journalistic content

Responsible

We collected:

1,697,774 tweets

11.96% had offensive terms

892,500 comments from 758 posts

0.79% had offensive terms

167,095 comments from 1,071 videos

8.92% had offensive terms

65,761 comments from 6,833 posts

11.31% had offensive terms

The data were collected between September 4th and November 4th, 2022.

Which candidates were monitored?

In this edition, we analyzed candidates for offices in State and Federal Chambers, Senate, Presidency and Vice-Presidency, from different parties and positions in the political spectrum, in all Brazilian states and in the Federal District. In addition, we sought to compose a diverse sample in the aspects of race/ethnicity, gender identity, sexual orientation, disability, age group, maternity and religion.

In order to ensure a representative sample of the diversity of women candidates in the 2022 elections, we committed to the following selection criteria:

50% of black women, when possible, within each office;

at least 10% of candidates belonging to transgender, non-heterosexual, with disabilities, people over 60, and mothers.

Based on these parameters, we defined the monitoring of:

• 5 presidential and vice-presidential candidates;
• 24 pre-candidates for state governments;
• 10 candidates for the Senate (two per region, from the largest and the smallest states);
• 81 candidates for the Federal Chamber (3 per state);
• 54 candidates for Legislative Assemblies (2 per state);
• 24 male candidates for the control group (following the same diversity criteria).

Monitoring: methods, monitored social media and collected data

The 2022’s edition of MonitorA took place between September 4 and November 4, beginning at the start of the electoral campaign and ending a few days after the second round of voting. Along the period mentioned above, we monitored the profiles of 198 candidates on Twitter, YouTube, Instagram and Facebook. The criterion for collecting a post was the presence of terms listed on MonitorA’s Lexicon. The tool, created in 2020 and updated in 2022, brings together words and expressions often used in acts of gender political violence.

To collect data from profiles on Instagram, as the platform does not offer access to its API, we developed a code that simulates a user collecting comments. For this, we used Python language and Selenium library. We accessed the accounts of each of the selected profiles for analysis and collected the latest photos and comments.

YouTube provides a more user-friendly API/interface to collect comments. The only challenge we faced was the daily limit of requests, and we overcame it by using multi-account API keys. The collection was carried out through a code in Python.

On Facebook, we developed a script that mimics the behavior of multiple users. Each was responsible for collecting the most recent posts from each monitored candidate. Then, comments from these posts were processed, cleaned, and pre-categorized as offensive or not, with the help of regular expressions.

On Twitter, we searched for posts that mentioned any of the monitored profiles. We used the Rtweet package in R, which makes requests in the Twitter API. We used regular expressions to identify, in an automated way, what was potentially offensive, and then proceeded with manual analysis.

After all the collections, we structured, cleaned and sent the data to a consolidated database. Finally, the data was made available for viewing in the Metabase dashboard. In this "wall", you can customize the categories, dates and filters, and then generate analysis samples.

Each of these databases was analyzed separately, by using pivot tables and individual verification of each post. In these analyses, we considered whether the content was an insult and/or an attack. If so, we detailed which terms characterized this condition, and in what category it fitted, among the possibilities of biphobia, ableism/psychophobia, intellectual discrediting, dehumanization, ageism, fatphobia, homophobia, political ideology, religious ideology, incitement to violence/threat, inferiorization, insult, lesbophobia, misogyny, moral offense, sexual offense/harassment, racism, disgust or transphobia.

The human checking aimed to identify the most frequent discourses in the practice of online political violence against women candidates, the profiles that most attack them, candidates who are more often victimized, the strategies used, among other possible interpretations.

Each of these analyses, at national, regional or local level, has been transformed into news articles, digital contents for social media and reports aimed at combating online gender violence, which serve as resources to demand changes to the platforms.

Some limitations

In 2022, the selection of candidates sought to create a scenario that represented the diversity of Brazilian women. However, in an attempt not to concentrate the analyses on certain groups, we understood that we did not make an exhaustive analysis that would contain, for example, candidates who are notorious victims of online political violence and who were in the pledge. This was a precaution for us not to include only the candidates we knew to be victims of political violence, making room for us to find other women who were also victimized, but who did not have this situation sufficiently acknowledged.

In this sense, we can cite as an example the choice of the same number of candidates for legislative positions by state, which excluded names such as Sâmia Bonfim and Marina Silva. In order to try to get around the absence of certain personalities, considering that the state of São Paulo has many candidates of great national projection - repeated targets of gender political violence –, we carried out a separate collection with state candidates for separate analysis.

It is also relevant to highlight a limitation on the individual analysis of publications, which, in this edition, involved nine professionals. In attempting to homogenize the classifications, we gathered all the professionals involved to evaluate collectively the first 100 posts, and to fine-tune the understanding of the characteristics of each one. From this meeting came the bases for the creation of our coding guide, shared with the whole team, and permanently adjusted according to each new discussion.

Still, throughout the entire execution of MonitorA, different terms and expressions were reclassified, varying with subjective and also contextual understandings. Thus, we understood that human analysis cannot be absolutely objective and regular, and the research results may vary with the multiplicity of coders.

Femininities in dispute on TikTok

In an electoral dispute marked by political violence and misogynistic attacks on women journalists and in which women's votes were decisive, it is not surprising that women around presidential candidates – especially their spouses – have gained centrality; on TikTok, this was not any different.

It is not news that Bolsonaro mobilized Michelle's image to reduce women’s rejection and ensure the votes of evangelics. Videos showing Michelle's speeches at religious events and photo collages with background music intended to provoke emotion (such as Unstoppable, by Sia) were recurrent, showing her using sign language, participating in charity activities, etc. Michelle's image, presented in accordance to conventional feminine aspects, articulated to events such as Bolsonaro's speech on September 7 referring to her as "princess", and the couple's visit to Queen Elizabeth II's funeral, generated countless videos with photo collages of Michelle praising her elegance. In the comments, people compared her to Lady Diana, named her "the princess of Brazil" and "a real first LADY". All this signals the reference to certain traditional gender and class roles that are suitable for the president's spouse, which involves modesty, gestures and clothing that would belong to a social hierarchy and mark her distinction.

The engagement of Janja Lula da Silva, spouse of the PT candidate, was also intense, but followed slightly different strategies. Janja has her own profile in TikTok, unlike Michelle, so she manages her own image more directly on the platform. On the few available videos, the focus was on rallies and marches, showing her with a microphone in hand, frequently wearing clothes that are socially read as more common, such as jeans and T-shirts. Janja was also often seen addressing women to talk about gender violence, debt of families headed by women, and feminist activism.

In a video of Poder360, in response to Bolsonaro's statement that "single men should look for a woman, a princess to marry", Janja, probably in a rally, said: "I want to ask you to turn on the lights of your cell phones, because I want to see if there is any princess around here. Or here. There isn’t, you know why? Because here there are only women! These are the women who will win this election in the first round for Luis Inácio Lula da Silva!"

The reactions to their different forms of participations came in different forms of gender-based aggression - in very specific ways, referring to the dynamics of each platform, but connected to broader misogynistic manifestations. An emblematic case was that of the video with Janja's statement: following common practice on TikTok, it was heavily edited and shared in order to make her statement ridiculous. The Bolsopatriota profile, for example, took her question "I want to see if there is any princess around here", and inserted photos of Benedita da Silva and Gleisi Hoffman along with screams of fright. Then, the video returns to the excerpt of Janja's speech, in which she says "there isn’t!", followed by a brief video of Bolsonaro laughing. The message is clear: women supporting Lula not only are not princesses, but are ugly, scary. Among the over 6,250 comments, women repeat "we, princesses, are with Bolsonaro", "the First Lady Michelle is a princess” and “princesses are from the right-wing”.

Attacks to left-wing women, feminists and lesbians through comments about their appearance is nothing new; however, such an offense is updated, quickly reproduced, and appropriated in particular ways on TikTok, by assigning sounds, by distorted image collages, and video editing. Similarly, platform usage dynamics are readjusted in the reproduction of violence, favoring the creation of "bubble filters", a term coined by Eli Pariser to deal with the intellectual isolation that can result from over-customization of algorithms. If on other social medias as Twitter offenses made to candidates and political figures on their posts and profiles are recurrent, the operation of TikTok's algorithm tends to the contrary - presenting content to people who agree with it on its For You area, not being so affected by initiatives of driving themes and people to trending topics, as on Twitter. This is noted on the video comment section – both for the left and right of the political spectrum, the vast majority of the comments are statements of support and praise of the content and the individuals represented. Disagreements and offenses are scarcer, and are limited to declarations of vote (in left-wing posts, people commenting "together with Bolsonaro" and "Myth 2022"; in right-wing posts, people commenting "Lula 13"). One of the exceptions to it involved Janja - on her videos, regardless of the content -, whether inviting people to sing Lula's jingle with her, or in photo collages with her husband: among the comments, there are messages from Bolsonaro’s supporters declaring "I prefer Michelle", "who is Janja, she’ll never be a princess", "poor thing" and "she wants to be Michelle and never will".

Janja, however, is not the only one targeted by attacks loaded with gender assumptions. Videos criticizing Michelle Bolsonaro have multiplied in social media during the elections, and although some intended to demonstrate contradictions between her statements of moral superiority/Christian belonging and practices of religious intolerance and accusations of corruption, others had contents aimed at ridiculing her. On one of the profiles, for example, a comparison is made between Janja and Michelle (having as background image the counterposition of Janja smiling and Michelle with an expression of suspicion), with the statement "one is a religious fanatic, semi-literate, with high-school degree, a very poor high-school, a political illiterate. They say she dated politicians before marrying the current president of the republic, they say that she has even been arrested, that she has a grandmother and an uncle involved in drug dealing, and worst of all is what I said in the first place: she is a blind religious fanatic, probably a schizophrenic who believes she is speaking to the holy spirit while she spins and rolls her tongue. The other is a sociologist, (...) has a postgraduate degree in History at UFRJ and holds an MBA in public policy." Offenses to religious belongings, suspicions around her sexuality, and psychophobia are articulated to an elitism associated with schooling and intellectual capital.

On the platform's patterns of use, we did not observe openly violent attacks to candidates or other women who achieved prominence throughout the election period, such as journalists and political commentators. However, dynamics of humiliation crossed by gender assumptions spread with irony were constant. Examples of these are records of the electoral debate for the government of Piauí, which took place on August 16, highlighting the participation of Lourdes Melo (PCO). Initially, the platform has viralized an excerpt in which the debate host interrupts Melo's speech and asks her to address a question to some candidate. She indignantly replies, "Do you intend to shut me up? Do you want to protect the candidates?" The video uses the common humor attribution tools, such as focusing and pausing on expressions of confusion and using emojis in captions. Then, several excerpts of speeches of the PCO candidate were also shared, with even more intense use of video editing tools. In a publication of the "best moments" of her participation in the debate, instrumental background music similar to that of comedy films was inserted. In addition, sound effects were added to mock her speech. By saying that "as a public servant, she lived under the whip of the governor", one hears the sound of a whip, and when she talks about the head of Executive’s authoritarianism and disrespect towards public power employees, an acceleration of the video is employed, producing in spectators the sensation of digression and reverie in Melo's speech. While she talks about topics such as police brutality targeting socioeconomically vulnerable population, the camera focuses on expressions of laughter among the other candidates participating in the debate.

By presenting her as someone confused, who behaves combatively for no reason and is displaced from reality, such videos not only make the candidate seem ridiculous, but also make what she says sound laughable. Thus, statements worthy of occupying a political debate, such as the treatment of public employees by heads of the executive, the alignment of candidates to elite interests and the support for Bolsonaro’s agenda are disregarded, because their image made object of derision contaminates the content of her speech. This is evident in the comments, all mocking her, in reference to her mental stability, her age and her literacy: "inviting people of this level leads to this", "how does this lady's family let her run for office?" and "if she takes a 5th grade test, she won’t reach 50% of the score."

This coat of irony to express and conceal humiliation and violence is also present on videos resulting from a search for PT’s candidate to Federal Deputy in Rio Grande do Sul, Maria do Rosário. Content produced about her was mainly based on linking her to mental imbalance and unjustified behaviors. On one of the first results of the search for her name, the screen splits into a montage that inserts her face into a body wrapped in a straitjacket, while the other half broadcasts a video in which she supposedly passes quickly through a space in the Chamber of Deputies, bumping into men who were around. In the homemade, cut-out recording of unknown context, one of the men is heard saying "she's crazy, she's attacking us", and the scene of her bumping into someone is repeated in slow motion. The comments reiterate the questions about her sanity: "this one is nuts", "this woman is completely imbalanced", and "it's easier for Pablo Vittar [a transsexual woman] to get pregnant of Thammy Gretchen [a transsexual man] than for Maria do Rosário to pass a psycho technical exam".

Another result broadcasts a video in which former Deputy Jair Bolsonaro threatens to slap her and says that "he would never rape [her] because [she] doesn’t deserve it”. Bolsonaro pushes her and Maria do Rosário says "but what is this?". She repeats the sentence a few times, but what is noticeable is that the video edition inserted this audio throughout the duration of the publication, in order to produce the perception that Rosario entered a loop of uncontrolled repetition. In the comments, many users mock her: "they say she's been saying 'what is this' until today LOL", "she had no argument and was like BUT WAT IS DIS", "the vinyl is scratched ", "the slap was so hard that bugged her lol". There are also troubling sayings such as "that's when he won me", "I didn't even know him before, and this was when I became a fan", "that's when he won my first vote", "if you’re looking for trouble…", etc. The suggestion of mental imbalance and unreasonable conduct, such as unmotivated attacks to men that would justify their "defense reaction," is evident on these videos. As with Lourdes Melo, the mocking around her extends to what she has to say; here, the attribution of an aggressive madness to Maria do Rosário trivializes the violence committed against her and delegitimizes her indignation.

It is surprising that the comment tab does not present the same hostility noted on platforms like Instagram and Twitter. In videos with women protagonists and electoral content, the tone of the comments is massively complimentary. In general, manifestations from users are supportive and congratulatory to the video’s participants. We often read, from the left and right wings, "you said it all!", "I make your words mine", "I only saw the truth", etc. Thus, it is suggested what was pointed out by Letícia Cesarino (2022), in reference to Chun (2016), about the confirmation dynamics provided by the operation of algorithms on platforms such as TikTok: segmentations are homophilic, connecting equals with equals, in which

algorithms deliver to users custom worlds that confirm their individual frameworks – in cyber terms, which contain an excess of positive feedback. As a result, users feel fully legitimized in their opinions and views (...). (2022:105)

Moreover, the pattern of the engagement dynamics in this tab is not very complex – the comments tend to have the same form and to be repeated by users: "Bolsonaro 22", "Lula president", "Bolsonaro in the first round", "Myth", "Lulathief", with full use of emojis, are the majority of the comments. Elaboration of ideas, even when they are a little longer, do not usually reach further than one sentence. There are also few interactions between users on the comments section, suggesting that there is no appropriation of it as a dialogical field.

Some patterns of comments, however, stand out. One of them, following the aforementioned patterns of platform use, refers to the way humor is mobilized both as mockery and as attack. Their tones vary considerably; on videos of young girls dancing to the sound of Bolsonaro’s jingle, we can see comments such as "I loved it, scored 13", and on videos of girls showing that they vote for Lula, we can see comments such as "I liked it 22 times". Other patterns, however, replicate the violent dynamics of video productions.

One of these patterns refers to comments such as "your fan", particularly frequent in videos that somehow criticize the candidate for Federal Deputy Nikolas Ferreira. Both in posts by Duda Salabert (a trans woman who was also a candidate for Federal Deputy in Minas Gerais) and Felipe Neto (an influencer), which directly or indirectly pointed out questionable actions of Ferreira (such as involvement with people accused of corruption and dissemination of misinformation), comments that mention Ferreira and present texts such as "your fan, Nikolas”, “loves you so much lol", "and this closeted love lol check this out Nikolas" are predominant. The use of male pronouns to refer to Duda Salabert is frequent, disrespecting her female gender identity – a factor that contributes to the general suggestion brought by the comment: that criticism is made because of an obsession based on a repressed, homosexual love. It is a form of invalidating criticism and of homophobic mockery towards the content’s authors.

Religious wars: How religion determined the agenda on Janja’s and Michelle’s social media

In the presidential race, the campaigns for Lula and Bolsonaro made efforts to win the vote of women, which, according to Datafolha, were the largest group among the undecided electorate. As part of the strategy for winning the female electorate – and, in the case of Bolsonaro’s campaign, female and evangelical voters - Janja and Michelle Bolsonaro played an important role in the their husbands’ campaigns. The political performance of both, however, is not new and did not begin in 2022. Janja is a militant affiliated to the Workers Party (PT) since 1983, and was a central part on the organization of the permanent demonstration placed in front of the superintendence office of the Federal Police in Curitiba while Lula was imprisoned. Michelle worked directly in the coordination of projects of the Ministry of Citizenship and in programs of the Ministry of Women, Family and Human Rights during her husband's term (2019-2022).

Throughout the election period, Janja participated in the organization of Lula's agenda and in media strategies. In Bolsonaro’s campaign, Michelle acted to gain support from the evangelical electorate. Bolsonaro's wife, also evangelical, markedly explored her religious identity, with aesthetic changes and the use of typical discourses of the Pentecostal repertoire, focusing on a supposed battle of good against evil that would be fought during the elections.

The attacks and insults directed at Janja and Michelle on social media were marked by misogyny, religion, narratives that disputed notions of morality and by religious intolerance.

In the weeks after and before the first round of elections (between September 26 and October 09), we monitored Janja's Twitter and Instagram profiles and Michelle's Instagram profile. We wanted to understand what kinds of attacks and insults were directed at the spouses of the presidential candidates and what narratives were constructed around these two figures. Just as the religious debate was central in the election campaigns for both candidacies, when we analyzed the hostile narratives against the two women in the media, we also observed attacks and offenses on religious grounds.

Within the discourses of "fight of good against evil" and "spiritual warfare" that set the tone of the 2022 political and electoral debate, the insults and religious attacks directed at Janja and Michelle operated quite differently. On the one hand, on comments directed at Michelle, the mobilized narratives revolved around a "false Christian", someone who "uses the name of God in vain" and a "bad evangelical"; in Janja's case, the narratives revolved mainly around religious intolerance against religions of African origin.

On Michelle's Instagram, in the two weeks analyzed, we identified at least 273 comments with offensive content and/or insults. Most of the comments with offensive terms mentioned alleged corruption cases, referring to the former first lady with derogatory terms like "Micheck" or "Michelão". In some cases, the accusations of corruption directed at Michelle were accompanied by hostile comments that claimed the former first lady was not a true Christian and would be using God’s name in vain:

You’re not ashamed of using God’s name in vain, first learn what it is to be a Christian, then you can fast!

@__marcia_cabral_ good hatred, an invention of Christians without literacy. Let them remove Micheck’s confidentiality two months from now

@michellebolsonaro using the holy name of the Lord in vain, right micheck?

On other comments, the narratives around Michelle’s false Christianity takes on more aggressive outlines, in which they claim she will "burn in the fire of hell", that God would be ashamed and that He "would have killed" people like Michelle:

God should be ashamed of all of you who are using His name in vain if God were here on earth you guys would have killed him because you guys are cruel because the God I believe in he is peace, he is love

Brazilians have no food, they have been fasting for 4 years you are a joker using this, it’s crazy, you’ll be the first to burn in the marble of hell

The form of expression of Michelle’s religion is also targeted by offenses on her Instagram. Expressions like "crazy fanatic" are used to attack the former first lady:

If it were, dear @michellebolsonaro would go to hell? You're a crazy fanatic... you think only your faith is divine ... spoil votes (if it wasn't for you he'd have + votes

The actions attributed to her husband, Jair Bolsonaro, directly impacted the offenses and the way Michelle was offended on Instagram. The videos that circulated during the electoral campaign, in the second round, in which Bolsonaro was seen in a freemason lodge, also resulted in offenses directed at Michelle. Although the videos were not directly related to the former first lady, on Instagram comments, Bolsonaro's presence in the freemason store led users to call her a "cursed heretic" and "pure demon".

Go dance samba in the freemason store. You damn heretic.

Now we know why @michellebolsonaro spinned like a pombagira on the terreiro she is pure demon #bolsonaria

In Janja's case, along the two weeks analyzed, we identified at least 799 comments with attacks and/or insults on the First Lady’s Instagram and Twitter profiles. The number of offenses directed at Janja was significantly higher than the number identified on Michelle Bolsonaro's accounts. This difference occurred for two reasons: (i) Michelle does not have a profile on Twitter, platform which concentrates more offensive posts in comparison to other social media; and (ii) on Instagram, a social media in which both have a profile, Janja receives proportionally more offenses than Michelle.

On Janja's social media, misogyny and religious intolerance - often combined - were the main categories of attacks directed at the first lady. Misogynistic narratives against Janja are primarily mobilized on Instagram, and comments of religious intolerance predominate on Twitter. After the end of the first round, we noticed that the number of religious offenses directed at the current first lady increased almost eight times. Among these offensive comments and tweets, the most frequently used term to offend Janja was "macumbeira".

The religious offenses directed at Janja revolved around two main axes: on the one hand, comments of religious racism against religions of African origin; on the other, accusations that Janja would be a "false Christian" and that she would be resorting to Christian symbols for electoral purposes.

In response to Janja's images and posts at the Círio de Nazaré in Belém, hostile comments on Twitter claimed that the first lady was "pretending to be a Catholic" and using Christian images and saints to garner votes:

@JanjaLula @kbene But when you have yours, you know, and you turn against him, just to ask for a vote, pretending to be Catholic, won’t He demand from you??? Go spend in cachaça and live chicken for despacho, right mrs janta (‘dinner’)

@JanjaLula @pretaderodinhas everyone is becoming a believer since they were little children

@JanjaLula The tramp seeking vote by using the name of the Saints

@JanjaLula Now she appeals even to the saint. A candle for the d.... and another for God. That's disgusting.

In many cases, in an offensive way, Twitter and Instagram users claimed that Janja was wrong to participate in the celebrations of the Círio de Nazaré, because she would be a "macumbeira". These comments directed at Janja are based on an idea that there is an intrinsic incompatibility between the celebration of Catholic and Christian rites with the celebration of rites of religions of African origin, reducing and denying any religious syncretism.

@JanjaLula You're going to the wrong place...

Your place is the TERREIRAO, DANCING AND SPINING IN YOUR MACUMBA!!,

#FORAPT

#ForaLulaLadrao , criminal, dissimulated, false, dictator, communist.

@JanjaLula One day wants to spin like a pombagira. On the other, next to St. Francis of Assisi. One thing or the other.

Most of the religious offenses directed at Janja focus on addressing her as "macumbeira" and as "pombagira", even accusing her of being a "Satanist":

@JanjaLula There are no drums for you macumbeira

@JanjaLula Satanist macumbeira

@JanjaLula Go there pombagira

"Macumbeira @JanjaLula her face is of an evil pombagira.....Heart of hatred, revenge, she got the right person @LulaOficial they’re a match because they breathe hate towards the people. ALL THEY WANT IS TO STEAL OUR MONEY!

Crossing religious intolerance and misogyny, religious offenses with sexual tone often appear on Twitter targeting the first lady, calling her a "slutty macumbeira" and using the figure of pombagira as a gender offense:

@JanjaLula @kbene Pombagira may not like it. She is jealous https://t.co/UhHAzSbaEY

@JanjaLula Macumbeira slut

@falom3rmo @marinasvianna @44cryptom @cafecomferri Lula went to the hospital to buy viagra to fuck macumbeira @JanjaLula . Next subject

Pombagira is an entity of the religions of African origin, which, according to Bantu mythology, is the orixá of paths, crossroads, road forks and communication. Pombagira, thus, represents the feminine spirit of Exu (SILVA, 2007). For Candomblé and Umbanda, Pombagira is a guardian and protector that promotes characteristics such as virility, balance, sensuality (FREIRE et al., 2019). For some Christian communities, on the other hand, Pombagira is a representation of evil (FREIRE et al., 2009), linked to moral unrulyness, hysteria, behavioral disorder and vulgarity (SIMAS; RUFINO, 2018).

The use of the Pombagira figure to offend and attack women is part of a conception that the offended woman would have an unbridled, unmeasured sexuality, that she would be an immoral woman. This association with immorality and vulgarity can be understood as "a trait of epistemic racism imbued with the sexist ideologies that condition us" (SIMAS; RUFINO, 2018, p. 90).

It is worth noting that Twitter and Instagram policies expressly prohibit hate speech based on religion. Twitter's hate-spreading conduct policy, for example, states that it is not permissible to promote violence, directly attack or threaten others based, among other things, on religion. Meta, controller of Facebook and Instagram, also includes "religion" as a protected category in its policy on Hate Speech, defining attacks as "violent or dehumanizing speeches, harmful stereotypes, statements of inferiority, expressions of contempt, repulse or rejection, slurs and incitement to exclusion or segregation," and prohibiting "the use of harmful stereotypes, which we define as dehumanizing comparisons historically used to attack, intimidate, or exclude specific groups."

Race and masculinities: the profiles of Kim Kataguiri and Fernando Holiday

Political violence is not restricted to women in politics. Since 2020, we have observed that men from historically marginalized groups - such as the elderly, GBT+ and non-whites -, their bodies, moralities and personal aspects of their lives have also become targets of political violence.

Aiming at understanding online violent and offensive narratives about men who work in politics, we monitored YouTube, Facebook, Instagram and Twitter profiles of eight candidates for federal deputy of the State of São Paulo. The analysis, in addition to reinforcing the findings of MonitorA 2020 regarding the specificities of the approaches and discourses within attacks directed at men, demonstrates the relevance of an intersectional analysis, since, unlike women, the men analyzed here were not attacked for being men, but because they belong to other socially marginalized groups. It is through the articulation between social markers of difference beyond masculinity that violence is manifested.

In this sense, on this section we analyze the narratives around Kim Kataguiri and Fernando Holiday, who were targeted by racial offensive narratives in their profiles. Although we understand that the act of being a man cannot be limited to a sole experience, because masculinity itself is operationalized in different ways, the characteristics that drive candidates away from hegemonic masculinity, that is, a masculinity that manifests itself as a rule (CONNELL; MESSERSCHMIDT, 2013), seem to guide the attacks.

In the case of Kim Kataguiri, his Japanese background, which makes him a member of the yellow community, was repeatedly mentioned in a pejorative manner. The use of the term "Jap", by itself or followed by a second adjective, as in "ugly Jap", "stupid Jap" and "bad character Jap", stands out among the attacks against the candidate. The use of other derogatory terms, such as "China" and "cata koquinho" (‘coconut gatherer’) was also observed.

"@KimKataguiri embarrassing yourself, China? Just study to understand it" (TWITTER)

@augustodeAB @KimKataguiri the only thing this ugly Jap has done in life is to embarrass himself! Ahahahahaha https://t.co/G6BsbZLlg1" (TWITTER)

"You’ll be embarassed cata koquinho, you’ll get roasted in the discussion" (YOUTUBE)

As seen in the publications above, the racial belonging of Kataguiri is an important marker in the offenses. The use of terms such as "Jap" and "China", a country that has no relation to the candidate, demonstrate how Asian descendants face a process of generalization of their origins.

The use of the term "Jap" seems so commonplace in the national vocabulary that it is not even restricted to offensive posts, appearing even in comments praising the candidate:

"I said [he] WILL NEVER debate with the little Jap.

And like every Bolsonaro supporter, he's a coward.

NEITHER PETISTAS NOR BOLSONARISTAS CAN FACE KIM" (YOUTUBE)

"The Jap is a genius lol" (YOUTUBE)

Thus, being a person of Asian origin, in the Brazilian context, implies a series of stereotypes, expectations and prejudices in highly reductionist terms, which not only disregard the individuality of these persons, but also erase their history.

This phenomenon becomes even more evident when we compare it to the experience of being read as a white person. In Brazil, being classified as white means that this person is not easily framed in an ethnic or racial group. The movement of framing individuals in racial groups occurs when we are facing historically minorized groups, which is the case of Asian, indigenous and black descendants. White people are, thus, read from a neutral place, which allows them to be seen simply as human beings (MOREIRA, 2019, p. 40). Being read as a human being for historically minorized groups requires disputes and the social recognition of their demands, which, in Brazil, is still an ongoing process.

In this sense, whiteness is constructed based on invisibility, while the subaltern position of other groups, that is, the groups read as racialized, is reaffirmed exactly in the differences. The use of Kataguiri's Asian descent as a form of attack brings visibility to a discursive tool: the white point of view analyzes the yellow presence to keep it segregated from the hegemonic standard identity, that is, the white identity.

A second point that cannot be ignored is how cultural and racial markers are operationalized in some of the offenses against Kim Kataguiri. The term "cata koquinho" (meaning ‘coconut gatherer’), for example, represents a racial offense that uses a typical sonority of Asian names to attack the candidate. What is observed, therefore, is a derogatory cultural representation of symbols from this minority group. This kind of discourse can be understood as a manifestation of superiority on the part of the aggressor, who believes that their cultural traditions or experiences have a higher importance (MOREIRA, 2019).

In Fernando Holiday’s case, most of the offenses were also based on the candidate’s race. Unlike Kataguiri, the offenses against Holiday seem to majorly connect to people's expectations towards him because of his race, as some of the following comments show:

"@FernandoHoliday I don't know if there is a far right or far left, but there are certainly people lacking character like you, a racist white wearing, who knows how, a black skin. Disgusting." (TWITTER)

"@FernandoHoliday (...) You THINK LIKE THE PLANTATION MANOR [i.e. the plantation master]. AND THAT’S ALL. They sleep in peace because of you." (TWITTER)

@FernandoHoliday You are a disgrace to black Brazilians! A GUY LIKE YOU DISGUST ME... (TWITTER)

"But now that's time to analyze the blacks for them to benefit you don't want to, Holiday? You're a disgrace as a black!" (INSTAGRAM)

"it’s crazy. shameful to see a black defending the side of the oppressor just to get views, I think he’s never heard the term historical reparation right? (...)" (INSTAGRAM)

Thus, as it can be seen from the examples set out above, most of the offenses against Holiday concerned his political position. If these comments had been put in such a way as to only point out inconsistencies on the candidate's positioning, or to criticize his political competence, the content of the publications would be in accordance with the majority of the contents addressed to other candidates (MonitorA, 2020). However, because he is a black man, most of the comments are not judging his individual stances, but his behavior as a black man. By stating that he would be "a white racist wearing a black skin"; "a disgrace as a black" or that he would think "like the plantation manor", the offenses employ a racist and violent vocabulary, even making allusions to the slavery period to attack the candidate.

Holiday's ideological stance is, therefore, operationalized in articulation to his ethnic-racial belonging. There is a visible difficulty on the part of the aggressors to deal with the candidate's political opinions and positions, and the relevance of this statement lies in the fact that they do not attack Holiday for thinking as he does, but rather for being a black man who thinks as he does. The social markers that cross the candidate's masculinity, especially his race, nourish the content of the offenses, and point to an attempt to limit him to a single version of what a representative of his ethnic-racial group should be like.

In these dynamics, racialized candidates, as opposed to white men - treated simply as men - are forced to deal with stereotypes and expectations that do not concern them as individuals. This is manifested as violence not only by the racist tone of the majority of the comments, but also by their imprisoning trait.

d. "You are a shame": distinct narratives in different contexts

The use of the expression "you are a shame" was present in several collections, being frequent to the point of becoming a category for encoding offenses. It appears in different contexts, among which "you are a shame to black people"; "you are a shame to women", etc.

When the user said that the candidate was "a shame" or "I'm ashamed of you", we considered the content as an insult. In other words, we did not think it should necessarily be removed from the platforms. On the other hand, when making allegations such as "you're a shame to women", we understood that shame was associated with gender expectations, so it had a similar weight to "you're a shame to black people" mentioned above, which made us categorize it as an attack.

In the case of Simone Tebet (PMDB), the main expressions were "you are a shame to this country", "National shame", "You are a shame", "shame for women". Alongside Soraya Thronicke (União Brasil), Tebet was quoted in 1,050 tweets with the phrase "you are a shame".

On Twitter, variations appeared such as "what a shame you are for MT", "you are a shame in the Senate" and "you shame women". Part of the attacks that questioned the representation of the senators used their gender to authorize criticism: "I'm a woman and you don't represent me."

Here, we again have the opposition between being a woman and not recognizing the other as representative. To say that someone does not represent us is not an offense, that is, it is not an attack or insult. However, the narrative itself seeks to delegitimize that a woman candidate may place herself as "representative of women". It is interesting to see how, in different contexts, gender crosses the political debates that take place on social media.

It is worth remembering that the narrative of shame did not occur without a context. During the presidential debate of August 28, 2022, on TV Band - in pool with TV Cultura, UOL and Folha de São Paulo –, then candidate for re-election, Jair Bolsonaro (PL) attacked journalist Vera Magalhães, after her mentioning the misinformation propaganda practiced by Bolsonaro during the Covid-19 pandemic. Offended, the former president said,

"Vera, I wouldn't expect anything else from you. I think you sleep thinking about me. You must have some kind of passion for me. You can't take sides in a debate. A shame for Brazilian journalism."

Facing Bolsonaro's attack, we sought to analyze the contents and to understand if there was a direct relation between the way Bolsonaro addressed women and, afterwards, the way attacks happened on social media. Accessing the mechanism of shame is a common practice in misogynistic attacks, according to researcher Sarah Sobieraj, President of the Department of Sociology at Tufts University (USA) and Associate Professor at Harvard University. In an article that brings together 38 in-depth interviews with women with a strong digital presence, she identified “shaming” among the top three practices of online gender violence, in addition to discrediting and intimidation, especially of a sexual nature.

In a way, we can associate shaming with discrediting, since allegations such as "you're a shame" are often associated with arguments of the kind "you're worthless", which is also a demerit. When we observed the tweets that marked the @ of Simone Tebet and Soraya Thronicke, on August 28 and 29, just after the debate mentioned above, we found more than 500 tweets referring to the candidates as incompetent, unprepared, incapable, insignificant, opportunistic, exploiter, stupid or idiots. In addition, 31% of the offensive publications mentioned Vera Magalhães, and 215 of them directly attacked the journalist by using derogatory adjectives and wordplay.

"General" politics and "identity" politics: mobilized discourses around trans candidates  

According to the National Association of Transvestites and Transsexuals (Antra), Brazil had 78 trans candidates in the 2022’s elections. MonitorA followed the social media of 11 of them, among which three were elected: Erika Hilton (PSOL-SP) and Duda Salabert (PDT-MG), as federal deputies, and Linda Brasil (PSOL-SE), as state deputy. Candidates Alexya Salvador (PT-SP), Athena Roveda (PDT-RS), Benny Briolly (PSOL-RJ), Biana Nunes (MDB-AL), Rafaela Esteffans (MDB-AP) and Robeyoncé Lima (PSOL-PE) will be alternates, and Paula Benett (PSB-DF) and Thabatta Pimenta (PSB-RN) were not elected.

In the first election in which transgender candidates were elected to offices in the Federal Legislative power, we looked at how the narratives about these candidacies took shape during the electoral campaign. In the investigation of transgender candidates, unlike other groups we monitored, we did not limit the analysis to offensive content, but expanded it to the main narratives built about these candidates. To this end, we performed an automated lexical analysis with the Iramuteq software, which allowed us to generate charts of Factor Correspondence Analysis, composed of clouds of statistically relevant words. The analysis of these debates was made on Twitter and Instagram profiles of the mentioned candidates.

We observed that great part of the political debate that transgender and transvestite candidates tried to propose on social media during the electoral campaign ended up hijacked by ideological-party conflicts, political violence, debates about public safety and disputes over the representativity of historically minorized groups. Agendas related to important issues for the candidates, therefore, could not reach the political dialogue when they related to issues read as "identity". User interactions with candidates reproduce the scenario of national political polarization, in which the defense of progressive parties and candidates was the main key to attacks to trans women running for office in the Legislative. Such fact made transphobic comments turn into an additional layer of violence, opposing messages of support and acceptance to the candidates, which represented an important part of the interactions.

These dynamics had significant variations depending on which social media was under analysis. On Instagram, we identified five groups of words linked to the candidates, especially Hilton and Salabert, and to elections in general, which focused on expressions of solidarity in face of the violence they suffered. Expressions such as "strength", "take care", "courage" and "resistance" appeared frequently for Duda Salabert (PDT-MG), in reference to death threats that victimized her in August. Around Erika Hilton, there were expressions of encouragement such as "community", "unite", "congress", "political group", "popular" and "let’s go". A more general group, distributed among all analyzed candidates, also brought terms such as "hope", "congratulations", "incredible", "wonderful", "exciting", "joy" and "pride".

The presence of less hostile and more welcoming interactions throughout the analyzed period may be related to Instagram's architecture, which apparently privileges the user's contact with personalities they admire or look up to in a way. Also on Instagram, unlike other platforms, the administrators of the candidates’ pages count on tools to delete comments and/or not show mentions of violent nature, and are able to fix comments they consider positive on posts.

On Twitter, we initially collected 62,000 tweets, but the analysis focused on the 6,000 that included potentially offensive terms. In this sample, hostile interactions with transgender and transvestite candidates were concentrated in three groups: (i) political and ideological-party polarization; (ii) public security and political violence; and (iii) representativity, identity and minorized groups.

Hate speech

Tweets with offensive words are divided in three discursive categories:

Ideological-party dispute and political polarization

Debate on violence and firearms

Identity agenda and minorized groups

Groups of words on offensive posts on Twitter

Data show little relation among the three kinds of discourses

The first gathers common narratives in the presidential campaign: challenges to the next federal government on economy, education, health and corruption, as well as mentions to Jair Bolsonaro (PL), Lula da Silva (PT) and Ciro Gomes (PDT). The second group of terms refers to public safety and political violence, discussing episodes that occurred outside the internet and the firearm agenda. In the third and final group, the narrative revolves around the representativity of socially minorized groups, gathering attacks and demonstrations of support to candidates. This separation indicates the existence of groups limited to certain debates.

Filthy and unresolved creature, always seeing things through the wrong side of history. Guns save lives when used by goodwill citizens (different from your kind). Tell your bodyguards to go unarmed from Monday on, how about that? Trash!!!

I respect who choose to assume what they are, she feels she’s a woman and that’s how she should be called. And if you’re looking at weenie, many macho men love one under the table

I got it: it’s like Lula, who is a thief and feels like an honest guy! Right

The research also showed that Twitter users bet on the discourse of transphobia to attack the ideological positioning of transgender and transvestite candidates. In order to understand the content of the offensive speeches, MonitorA analyzed a sample of 1,000 tweets, randomly selected from the 6,000 that contained potentially offensive terms. Among them, 8.2% (82 tweets) were considered, in fact, offensive. Often, aggressions used language belonging to the trans universe disrespectfully, but without name-calling or direct attacks. In some cases, transphobia appeared by the use of "he" to refer to transvestites or "she" to refer transgender men. It also appeared, more aggressively, in terms such as "unresolved", "undefined being", "aberration", "macho" and "faggot".

Be ashamed of yourself. You’re not trans, lying creature

You’re a shame to the lgbt movement. FAKE! YOU HAVE NEVER BEEN TRANS

After transphobic comments, insults such as "hypocrite" and "ridiculous" were the second largest category of hostility (18 tweets), followed by attacks with terms such as "imbecile", "stupid" and "donkey", which promote intellectual discrediting of the candidates (15). Those related to political ideology (12) employed terms such as "militonta" (dumb militant), "psolenta" and "disgusting ptista" (terms mocking members of the parties PSOL and PT). Misogynistic offenses (8) accused candidates of "mimimi" (an onomatopoeic reference to ‘whining’), among other offensive comments.

The analysis of attacks shows some challenges for content moderation, such as the need to consider the specificities of this group, which includes as an important point the matter of addressing and identifying these persons. The use of a male pronoun, for example, will not necessarily be detected as a problem, but, in the trans context, it can represent important violence, as posting images of trans people before their transition.

Therefore, the context becomes essential in the identification of transphobic attacks, a fact that points to the need for greater sensitivity and training on the part of the tools and individuals responsible for content moderation, as well as the presence of the LGBTQIAPN+ population as an active part in the construction of consistent guidelines for combating this kind of violence.

The 2022’s elections were the first to be held after the enactment of the Law on Gender Political Violence (Law No. 14,192/2022), which establishes rules to prevent, suppress and combat political violence against women during elections and in exercise of political rights and public functions. According to data from the National Council of Justice (NCJ), in the last 15 months, 7 cases have been recorded every 30 days, demonstrating that challenges regularly faced by women’s participation in politics aggravate with the approach of new elections.

The rapporteur of the Bill on Gender Political Violence in the Deputy Chamber, Deputy Ângela Amin (PP/SC), mentioned MonitorA in her opinion for the approval of the text. She detailed data from the 2020’s observatory, and argued that "in this revolting scenario, adopting effective measures to combat political violence against women is a pressing measure."

Given the importance of the legal acknowledgement that political violence needs to be addressed and of the role of MonitorA in her argument, we were interested in understanding how the legislation would actually be applied, and how online violence would be understood by the judiciary power.

We worked, therefore, in partnership with the Núcleo de Inteligência Eleitoral (Electoral Intelligence Center), in a research aiming to understand the applicability and level of effectiveness of the Law on Gender Political Violence during the election period.

How was the mapping carried out?

For the analysis to be possible, the research was carried out in two stages. In the first stage, we mapped the cases of gender political violence filed between August 16, 2022 (beginning of the electoral campaign) and October 31, 2022 (end date of the 2022 elections). For the search of legal actions, the cumulative combination of the following words was defined as search terms: (i) "political violence" and "gender"; (ii) "Art. 326-B" and "electoral code" and "gender political violence"; and (iii) "political violence" and "woman". In relation to the victims, we adopted as criteria the selection of cases of candidates for elective offices or people who identify with the female gender in who hold elective office. The monitoring was not extended to victims of political violence committed against those who were not candidates or elected members along the defined period. Based on this initial search, twelve cases of political violence against women were identified.

Due to the low number of cases identified, the second stage of the research was conducted, through an interview with the Work Group to Combat Gender Political Violence, with the objective of understanding the mechanisms for complaints, and the reasons for the low number of judicialized cases about gender political violence identified in stage one.

Additionally, we forwarded a letter to the Superior Electoral Court (SEC) requesting information on judicial actions that have been proposed within the temporal scope of the research, and sent a letter to the Federal Public Prosecutor's Office (FPO) requesting information on the number of complaints involving political violence on the grounds of gender.

The Superior Electoral Court denied access to the data because it is impossible to provide it in an anonymized way, and the sharing of this kind of information would violate the General Data Protection Law. The Prosecutor’s Office forwarded a report with 62 cases that are being processed in the Office, without considering cases that may be under process in the Electoral Prosecutors’ Office of each of the 27 electoral zones and that have not yet come to their attention. These 62 cases, however, could not be analyzed, because, currently, there lacks a system that allows the process to be fully followed with access to detailed information.

Thus, based on the two stages of the research - and despite the methodological challenges resulting from the absence of a unified channel for the search for judicial actions - twelve cases were analyzed. Of these, four cases were related to irregularities committed by the parties, and eight cases were related to harassment or threats directed at candidates or holders of offices. The results found are described below.

b. Cases in progress at the Electoral Court

b.1. Irregularities related to parties

Among the cases identified, the majority did not make specific reference to online gender political violence. Of the 12 cases found, four referred to irregularities of distribution of party funds. These actions deal with three themes: competence for the trial of these cases, investigation of fictitious candidacy to comply with the minimum percentage of women candidates required by law, irregularity on the distribution of offices to a federal deputy, and absence of budget transfer from the Special Campaign Financing Fund to women candidates.

1. Launching of fictitious female candidacy by the party
   

The first case (process no. 0601408-34.2022.6.06.0000) is being processed by the Regional Electoral Court of Ceará (REC-CE), as an electoral judicial investigation, whose complaint is based on Art. 10, §3, of Law 9,504/97, which states that "all parties or coalitions must fill a minimum of 30% and a maximum of 70% of candidacies of each sex".

On this case, the plaintiff of the action states that her name was included as a candidate to state deputy by Partido Liberal, PL (Liberal Party). She claims to have been a victim of political violence because she was the target of a fictitious candidacy, carried out only to meet the gender reserve provided for in the electoral law. On her action, the author requires (i) the suspension of the analysis of the requirements for candidacy registration, (ii) the suspension of the access of PL-Ceará to the electoral campaign’s public funds (Special Campaign Financial Fund and/or Party Fund) until an investigation is made about the alleged fictitious candidacy, (iii) that an investigation is carried out on the alleged fraud in the composition of the candidate list, and (iv) the extinction of the candidacy records of all candidates for State Deputy from PL-Ceará, as well as the establishment of disciplinary procedure or a criminal action against the leaders of the party.

Until November 30, in a monocratic decision, the requests to suspend the analysis of requirements for candidacy registration and to suspend the access of PL-Ceará to campaign public resources were denied. The Regional Electoral Court of Ceará determined the summoning of the legal representative of the party association and of all the candidates chosen by the party to run for State Deputy offices to be part of the process and to present a defense. The action is still in progress.  

1. Lack of transfer from the Special Campaign Financial Fund to women candidates
   

The second group of identified cases refers to disputes over the transfer of funds to women's candidacies. Both cases are under the Electoral Supreme Court (cases no. 0601155-14.2022.6.00.0000 and no. 0601152-59.2022.6.00.0000). The actions are grounded, respectively, on Resolution No. 23,605, Art. 6, § 1, I and I and on Resolution 23,607, Art. 17, § 4, I, II and Law 14,192/2021, Art. 2 and 3 and single paragraph.

In both cases, the self-declared black candidates allege the absence of transfer of funds from the Special Campaign Financial Fund to women candidates. The authors of the complaints require compliance with the rules of the apportionment of the election campaign financing fund, with proof of compliance with the obligation in the file of the process, under penalty of a daily fine of R$ 100,000.00 and under penalty of blocking the campaign accounts. They also require (i) that abusive conduct be recognized by the leaders of PL-Amazonas, (ii) that the rights to women's and black’s quota be recognized, and (iii) that the conduct be framed as an offense of political violence against women (Law 14.192/2021, Art. 2, 3 and single paragraph).

Neither case has been received by the Superior Electoral Court. The Court considered that the case does not comply with the legal requirements to authorize the proceeding of the complaint, on the view that the Court does not have jurisdiction to adjudicate the matter. According to the Electoral Court, the Electoral Court of Amazonas, the cases’ original court, would have jurisdiction over it. None of the victims appealed the decision.

1. Irregularity in the distribution of Federal Deputy offices

Finally, the last case refers to the alleged irregularity in the distribution of offices (case no. 0600444-56.2022.6.27.0000). The case proceeds under the Regional Electoral Court of Tocantins, and the complaint is grounded on Art. 7, caput, of Law 9.504/97.

In this process, the victim sought to challenge the registration of candidacy before the Superior Electoral Court of Goiás for alleged irregularity in the distribution of Federal Deputy Offices. The victim claims that she was a candidate for a Federal Deputy office by Partido Progressista-TO (Progressive Party), and that she would have presented to the party and carried out pre-campaign. At a party convention, she would have pleaded for a seat to run for federal deputy. However, PP would have mistakenly withdrawn her name and informed the Electoral Court her candidacy for an office not of federal, but state deputy. The victim claims that the convention’s minutes had been rigged and the party's conduct would constitute political violence against women.

In a decision, the Electoral Court of the state of Tocantins considered the negative of the party in letting the victim run for federal deputy as a party’s "internal matter", and not as a case of gender political violence:

"[...] disagreement about the application of the democratic principle regarding the form of statutory choice of candidates by the Partido Progressista is classically characterized as interna corporis matter, which can be questioned only by the members themselves in another judicial field, and it is not for the Electoral Court to interfere and question the criteria of choice, since the electoral legislation itself confers on the statutes of the parties the prerogative to define the form of choice and the criteria of candidates for the election pledge. Precedents of the Superior Electoral Court".

Thus, the challenge was dismissed by the Regional Electoral Court of Tocantins.

b.2. Harassment or threats directed at candidates or representatives

Eight cases involving offensive attacks to women candidates based on Art. 326-B of the Electoral Code were identified. Of these, four cases involved gender political violence in face-to-face environments during the electoral campaign. Only one of the identified processes occurred on digital platforms. The other three cases are still under investigation, with initiation of inquiry and filed complaints. Because they were in early stages, until the publication of this report it was not possible to have access to detailed information about the processes.

The cases are described below.

1. Offline gender political violence
   

The first case (case no. 0606277-77.2022.6.19.0000) is before the Regional Electoral Court of Rio de Janeiro and has as legal basis Art. 22 of Supplementary Law no. 64/90 and Art. 326-B of the Electoral Code, which typifies the crime of gender political violence.

On this case, the claim is that the persons investigated would have committed acts of gender political violence against the candidate, making unfounded offenses and criticism aiming at her withdrawal from the candidacy. Also, one of the investigated persons is alleged to have used his condition as a city councilor to attack the candidate, as well as allegedly used resources from the Special Campaign Financial Fund to finance his election campaign, which would constitute abuse of power. In the action, the conviction of the defendants for abuse of economic and political power is required and for committing the crime provided for in Art. 326-B.

The court dismissed the charges of abuse of power, on the view that there was no evidence of wrongdoing. In relation to the crime of political violence, the file was forwarded to the MP - the body responsible for investigating and filing the complaint - to investigate possible offense to Art. 326-B of the Electoral Code. The case was declared extinct without a resolution of merit.

The second case identified also proceeds under the Regional Electoral Court of Rio de Janeiro (case no. 0606052-57.2022.6.19.0000) and has as legal basis the Arts. 248, 326-B and 331 of the Electoral Code, and Art. 110 of Resolution No. 23,610/2019 of the Superior Electoral Court. In the process, the claim is that the candidate was threatened while performing acts of electoral campaign. According to the case, the accused were accompanied by armed security and demanded that the candidate move away from the scene and stop the campaign acts.

In the lawsuit, the representative argues that there was a violation of the Arts. 248 of the Electoral Code and 110 of Resolution No. 23,610/2019 of the Superior Electoral Court, which provide that no one can prevent electoral propaganda, as well as the crimes provided for in the Arts. 326-B and 331 of the Electoral Code, relating to gender political violence and the prohibition of "impairing, altering or disturbing properly employed means of propaganda", respectively.

The court ordered the referral of the whole case-file to the Regional Electoral Prosecutor's Office for the adoption of the appropriate measures; and the redistribution to one of the Electoral Judges designated to judge representations of electoral propaganda.

The third case (case no. 0600975-41.2022.6.02.0000) concerned a complaint that is proceeding in the Regional Electoral Court of Alagoas. The candidates claim that they were the targets of speeches aiming to prevent or hinder the election campaign during the inauguration event of the Electoral Committee of the candidate for Governor, Rui Soares Palmeira, in the city of Arapiraca. According to the plaintiffs, at the time, the competitor in the electoral race attacked their campaigns due to their condition of being women. They claim to have been victims of gender political violence according to Arts. 326–A, § 3, and 326–B of the Electoral Code.

The Court referred the case to the State Prosecutor’s Office for it to become aware of the complaint. The Regional Electoral Prosecutor confirmed and registered the complaint within the State Prosecutor’s Office and asked for formal filing of the pieces within the Electoral Justice.

Finally, the fourth case (case no. 0601448-86.2022.6.07.0000) refers to a candidate who claimed to have suffered verbal attacks against her candidacy in a face-to-face event by a senator, in a statement on behalf of the Regional Collegiate Federation PSDB Citizenship of the Federal District. In the action, the candidate requested the suspension of any statement of the offender on behalf of the Collegiate, until referred the registration of the candidacy of the victim, stating that the practices of the senator fall within Art. 8 and 326-B of the Electoral Code.

In its decision, the Regional Electoral Court of the Federal District recognized violence on grounds of gender, however, there were no criminal consequences due to lack of express request. The victim merely pleaded that the offending party be prevented from interfering with her candidacy registration. At the time of the decision, the registration of the candidacy had already been granted, so that it was understood that there was a loss of procedural interest. Although the decision mentions the existence of a specific channel for the investigation of the crime, that is, via Electoral Public Prosecutor's Office, the process was extinguished without resolution of merit.

1. Online gender political violence

Of the 12 cases about gender political violence, only one was about violence online. The case (case no. 0602410-12.2022.6.07.0000) is now before the Regional Electoral Court of the Federal District and was filed by then senate candidate Damares Alves, against an Instagram profile. In the lawsuit, Alves argues that the profile is responsible for promoting personal offenses against her, with negative propaganda, dissemination of false information, public contempt and crime of electoral violence against women. The senator required the identification of the person responsible for the profile, the removal of the reported publications, and that the profile and the responsible refrain from running publications with the same content. In addition, Alves asked for the payment of a fine for dissemination of misinformation under the Elections Act and Resolution 23.610/2019-TSE.

Until November 30, in the absence of an express request for the case to fall in the crime provided for by Law 14.192/21, the decision was limited to determining the identification of the agent and the removal of the content on the mentioned platform.

As portrayed in the cases above, on three of the five lawsuits identified - both those that occurred online and offline – the Electoral Court considered that the lawsuits filed on the basis of Article 326-B should be referred to the Public Prosecutor's Office due to their unconditioned public competence, guiding the victim to request the establishment of an investigation to the Prosecutor’s Office. As consequence, there is no decision of merit on such actions. Although the two other cases mention gender political violence, they do not include the conviction for committing the illicit within their requirements.

c) Cases that are not yet under investigation, with no detailed information on the complaint

In addition to the five cases described above, available on Electoral Justice portals, there are three other, namely:

Case number: 0600028-48.2022.6.19.0150

Court: Regional Electoral Court of Rio de Janeiro

Class: Police inquiry

Case number: 0600040-77.2022.6.21.0030

Court: Regional Electoral Court of Rio Grande do Sul

Class: Police inquiry

Case number: 0600071-42.2022.6.17.0149

Court: Regional Electoral Court of Pernambuco

Class: Complaint

The cases are related to violence against women candidates or in the exercise of elective office, however, they were submitted to the Electoral Prosecutor's Office for investigation, with no detailed information about the object of the complaint or its progress.

b.3. Complaints under investigation by the Public Prosecutor's Office

Due to the low number of legal actions related to gender political violence, we sought to understand the reasons for such result. In addition to the talks with the Work Group for Combating Gender Political Violence (WGCGPV), we obtained information from the Public Prosecutor's Office about the complaints that were being investigated at an administrative level.

In conversations with the coordinator of WGCGPV, Raquel Branquinho, the prosecutor acknowledged that there are still flaws in identifying and dealing with allegations of gender political violence by prosecutors in all states, and that the Secretariat of Women has been essential to press the action and resolution of these cases. In this sense, she emphasized the importance presenting the complaints through the correct channels, so that they are received and dealt with agility, still in the election period.

The meeting with Branquinho contributed to understand the results obtained by data collection. Given the information resulting from this meeting, we chose to send a request for information to the Federal Public Prosecutor's Office about the number of complaints involving political violence on grounds of gender.

In a letter to the Federal Public Prosecutor's Office, we requested information on the number of complaints involving the subject of gender political violence. In response, we received a report that identified the existence of 62 cases under investigation by the PPO. The list did not consider cases that under investigation by State Electoral Public Prosecutor's Offices which had not yet reached the PPO.

It is noteworthy that these complaints do not fall within the scope of the research, since they were still under investigation, that is, there was still no complaint in the judicial scope. Although there are no ways to monitor these cases, it is worth mentioning some relevant information from the report:

The complaints were classified according to five kinds of violence, as described in the guide produced by WGCGPV on gender political violence:

1. Physical: as bodily injuries; arbitrary imprisonment; torture; ill-treatment; and feminicide;

2. Moral: associated with crimes against honor, such as defamation, slander and libel.

3. Economic and structural: such as damage to property, deprivation of campaign resources by refusal of access or diversion, extortion; associated with the lack of transfer of funds from the Party Fund (PF) and the Special Campaign Financial Fund (SCFF), especially in view of the gender and racial quota; non-broadcasting of free electoral propaganda on radio or TV;

4. Psychological: associated with violation of intimacy, isolation, threat against the victim, family member or close person, humiliation, manipulation;

5. Symbolic: such as intimidation, silencing, demerit, use of exclusionary language, restriction of speech, imposition of tasks foreign to the office, restriction of access to justice, omission of information, questions about clothing, appearance, weight, personal life and sexuality, absence of a seat in parliaments.

Most reported cases involved more than one type of violence, as described in the table below:

Although the majority of cases could fall under gender political violence as defined by Law 14.192/2021, these are under investigation by the Public Prosecutor’s Office. No channels were found that allow the monitoring of processes before the PPO, so that the intended analysis was impaired.

d. An insufficient legislation

Despite the indisputable achievement of the enactment of a law designed to look at and to take specific care of women's candidacies in order to ensure the space of this portion of the population in electoral pledges, it became evident that the current scenario does not corroborate the intended effectiveness.

The plurality of channels able to receive complaints of violation of legislation disperses the filing of complaints by the victims. The consequence is a delay in the initiation and progress of investigations during the short electoral period.

Victims, in their turn, not always count on the tools - emotional, legal and of information about the law - to recognize that the violence suffered can be framed as gender political violence. Still, even those who understand that they have suffered political violence do not necessarily find sufficient motivation to denounce their offenders, either because of lack of knowledge of the appropriate channels, or for fearing the consequences of such actions. This inertia is evident when we compared the amount of attacks on women candidates on social media during the electoral period to the small number of complaints filed at the Electoral Prosecutor's Office.

The survey also evidenced the variety of understandings of the Electoral Justice about cases of gender political violence, either to understand the limits of their competence, or to identify the subjectivity common to the cases. The equity sought through Law 14.192/21 is given precisely in response to a difference in gender structurally built in society, which, in general, will not be as obvious as in cases of crimes not motivated by gender, race or ethnicity.

The current procedure delays the resolution of cases involving gender political violence, which explains the existence of 62 complaints to the Federal Public Prosecutor's Office, disregarding any cases reported to the PPOs of the 27 Brazilian Electoral Zones, compared to the identification of only 12 cases in the Electoral Court. As we can see in the decisions studied, the competence to file an action at the Electoral Court is public and unconditional to the Public Prosecutor's Office. However, for a case to reach this stage, it is necessary for the PPO to conclude the investigation.

We did not identify cases between August 16 and October 30, 2022 (electoral period of 2022), whose investigation had been completed and had resulted on electoral criminal action before the courts. That is, even if the victims reported the alleged violence suffered during the election period, none of them would effectively have the legal support brought by Law 14.192/21, because the current procedure is not sufficiently rapid to ensure it.

If the type of criminal offense specified by the legislation aims to suppress gender violence that hinder the candidacy and the exercise of elective office by women, this objective has not been achieved by the available means, since a decision given after the election period or the simple opening of a complaint at the PPO without any outcome are not able to prevent the effects of violence. In this sense, the law does not support the parity on number of women in elective positions, since its effects will only have practical repercussions in the next electoral period.

In addition, if the actions of this type of criminal offense are classified as public competence and unconditioned to the PPO, it is necessary to invest, at least in electoral periods, in larger teams, thus enabling a constant, focused and proactive supervision, ensuring the rapid identification of cases and promoting the speed in the course and outcome of investigations. This is because the law establishes in its Article 2, single paragraph, that, "competent authorities shall prioritize the immediate exercise of the violated right, giving special importance to the victim's statements and to evidence elements".

Political violence against women and historically minorized groups permeate the political and electoral environments, undermining the participation and diversity of voices in Brazilian institutions. Although there have been recent advances in the Brazilian legislation and in the performances of political parties and platforms towards the debate and creation of new regulations and good practices, there is still a long path ahead to effectively confronting and mitigating gender political violence.

We have gathered, based on a perspective directed at historically subalternized groups and on data collected throughout the 2022 elections, 13 recommendations to sectors involved in the topic that may enable us to envision a path of combat, prevention and monitoring of political violence.

Platforms and Internet Application Providers

1. Commitment to the development policies and guidelines for protective usage in face of political violence

Online gender-based political violence and violence against historically marginalized groups involve the combination of multiple factors, such as hate speech, discrimination, democratic guarantees, specific electoral contexts and political disputes among different actors. In this context, the construction of usage policies and guidelines that aim to combat and prevent online political violence needs to consider all the variables related to this specific kind of violence, how it is spread online, and how it affects the candidacies and the political-electoral performance of the affected groups. Often, policies aimed at hate speech, bullying and electoral integrity fail to account for these diverse aspects surrounding political violence.

We therefore recommend the development of specific usage policies and guidelines for political violence that consider:

a. Differences between the kinds of offensive contents directed at candidates, with contextual distinctions regarding what platforms understand as hostile language (which does not violate internal policies), and what platforms understand as aggressive content (which cannot be tolerated). We take as example the distinction between insults and attacks proposed on this report;

b. Political and electoral local contexts, so that the parameters for taking or not certain content for political violence are localized and specific to the country and/or region;

c. The construction of swift and efficient responses, facing the possibility of massification of hostile content directed at historically marginalized groups along more critical periods, such as electoral campaigns. This may apply not only to contents understood as attacks, but also to those understood as hostility, which would equally result in damage to the candidates involved;

d. The need to build specific policies to confront political violence, in dialogue with civil society and the academia, which can provide input on the local context, on political changes in course in countries and regions, and on specificities of different forms of discrimination. Here, special attention to the way in which relations are established among platforms, civil society organizations and the academia is required. It is also necessary to think about relations that may bring gains to all parties. There must be access to data or to justifications for removing (or not) a content from the platform, as well as funding for research and continuous dialogue, beyond electoral periods. Those are examples of points to be appreciated by the parties.

e. Transparency in relation to the platforms' criteria for classifying different kinds of political violence and to the protection practices aimed at historically marginalized groups;

f. The need to understand the combat against gender political violence apart from content moderation practices, with the inclusion of provisions on media education, such as the production of educational guides on practical aspects of restraining political violence against minority groups; and

g. The understanding that political violence must be faced as a continuous phenomenon, which does not occur only as a result of elections. The monitoring of parliamentarians, especially those belonging to historically marginalized groups, must be constant, as well as the application of policies focused on combating and mitigating political violence.

2. Improving practices of data access and transparency for researchers

Understanding the ways in which political violence takes place online, during elections or in other periods, requires the collaboration of researchers, platforms and public authorities. Practices of transparency and data access, granted by platforms, are an essential part of the construction of diagnoses, assessments, comparisons and recommendations for confronting violence.

Gender-based violence and violence against historically marginalized groups on social media are widely known facts among researchers, either by previous research or by empirical experience. Detailed information about how political violence occurs and how platforms moderate offensive content, however, are still largely unknown. The reason for this difficulty relates to limitations in access to data from digital platforms. The lack of significant transparency measures and of practices of data access for research are an obstacle to the advancement of scientific production on online violence and to the articulation of civil society organizations.

In addition to quantitative data, it is essential to extend the practices of transparency and data access to other aspects of the platforms, such as contact, meetings and interviews with the platforms' teams of content moderation, human rights and public policies, as well as access to internal policies and content moderation criteria, among others.

Thus, to ensure transparency and academic and knowledge freedom, we recommend that platforms build and apply data access policies and programs for researchers. Knowing more deeply the dynamics of violence and the decision-making criteria of platform teams will also enable us to contribute to the mitigation and confrontation of political violence in other social sectors.

3. Commitment to protective enforcement and to transparency of usage policies and guidelines

In order to address violent speech online, especially during the electoral period, it is essential that platforms and internet application providers act in an agile and proactive manner in the moderation of violent content, with a swift application of policies aimed at hate speech, bullying and election integrity.

It is worth emphasizing that, in addition to the removal of contents that violate the platform’s terms of use, there needs to be more transparency about the procedures related to content moderation. It is recommended that platforms inform the decision-making procedures for the removal of a content, with the reasons for the decision and informing those who had the content removed of which policies were violated by the content. Furthermore, it is recommended that platforms have appeal instances for challenges of content moderation.

We also suggest that platforms inform, on their transparency reports, the quantity, the reason for removal, and the kind of content removed during the electoral period. In addition, we suggest that the platform’s actions for confronting violent discourses should also be qualitatively explained and, whether in general elections or municipal elections, that these data are separated by region of the country.

4. Improving policies aimed at public figures and political office holders

Platforms’ policies aimed at public figures, especially at those holding political offices and functions, allow, in behalf of public interest, the publication of comments containing harsher criticism, as long as they are not severely abusive. A more permissive stance towards people with political offices is justified as a way to guarantee freedom of public debate and to enable voters to make demands to people who act politically.

A more hostile space for political figures, in reason of their public office, however, may also mean an even greater barrier to the democratic participation of historically marginalized groups. As an example, moral offenses which, when isolated, would not be considered seriously abusive and which could, therefore, be allowed by the platforms, when massively repeated, may have a largely negative and abusive impact, just as severe as a direct attack.

If public office holders should receive a differentiated attention because of their public function, it is necessary that the moderation of hostile content directed at them take into account the differences of the various groups of people who hold public office and of their political contexts. Thus, platform policies should not be restricted to offensive words or to the posted content, but should also observe the broader dynamics and the offensive potentials of that content, analyzing how massive the attacks and insults are, as well as the narratives there mobilized. Platforms could, for example, monitor candidates and policies in order to analyze the cases in which insults are massive, in which there are waves of offenses directed at office holders, which differ from the daily patterns of messages/posts received by the public figure. This monitoring would enable, for example, a more thorough analysis, in order to understand whether the content is legitimate, or it constitutes a massive attack to be removed from the platform. It is essential that this kind of monitoring take into account the different social groups affected by mass insults or attacks, since historically marginalized groups, such as women, LGBTQIAP+ people, black people and people with disabilities are more vulnerable, and a hostile environment can hamper their participation of the political-electoral debate.

It is worth mentioning that the monitoring of content published by public office holders and by public figures should identify violent content produced and disseminated by them, especially one directed at historically marginalized groups. The reach of contents with political violence published by public figures can be significantly broader, and can generate a massification of attacks and insults to the candidacies of women and minorized groups.

5. Improving platforms' internal reporting channels

Platforms' reporting channels are central for candidates and users to take action towards actively removing violent content and protecting themselves from online abuse and harassment. It is recommended that, for a more efficient and protective operation, reporting channels include:

a. Prioritizing reports based on risk of harm, with the use of artificial intelligence to speed up the application of policies;

b. Permission to report multiple contents simultaneously; and

c. Inclusion of the categories of political violence and hate speech against historically marginalized groups as specific items of reporting, to facilitate the identification, combat, and prevention of political violence against historically marginalized groups.

Legislative Branch

6. Improving legislation on gender political violence, with the inclusion, on the Law of Gender Political Violence, of protective measures and of liability of political parties

The approval of the Law on Gender Political Violence was an important acknowledgement, by the Public Power, of the need to implement actions for the protection of women who work or who intend to take part on institutional politics. However, the mapping on the enforcement and effectiveness of the legislation points to an insufficiency of the law in achieving its goals. In its current wording, the law provides for electoral crimes and establishes duties for political parties; however, it does not establish protective measures or hypotheses of protection for victims in the course of the judicial process, especially considering the electoral period, in which candidates are more exposed and more susceptible to violence.

The period required for the trial of an action, until the deliverance of a sentence or interlocutory decisions, is often incompatible with the typical swiftness of the electoral period. Thus, the absence of protective measures aimed at candidates and politicians along the course of trial results in insufficient legal protection prior to the decision on the merits by the competent judiciary organ. There is no provision, for example, to cases in which the aggressor holds elective office in the same power stance as the victim.

In this sense, it is necessary that the Law on Gender Political Violence include measures for the protection of the candidate, when she is at risk, in the light of similar measures provided for by Law Maria da Penha (Law No. 11,340/2006) and by the Statute of the Child and Adolescent (Law No. 8,069/1990), which establish hypotheses of prohibition of certain conducts, such as the approach or contact with the offended (Art. 22 of Law No. 11,340/2006).

In addition to protective measures for women candidates and politicians, the Law on Political Violence must provide for measures to hold accountable those political parties and candidates that use political violence as a strategy for electoral campaigns. Violence against candidates and women who act in politics is often perpetrated by people who also run for elective office or hold public office, through online and offline attacks against their political opponents. To address this kind of case, the legislation on gender-based political violence may include hypotheses of civil liability, with the application of fines or decreased access to campaign funds, for example, both to candidates and to parties.

Finally, there needs to be an improvement in the wording of the Law on Gender-Based Political Violence in relation to the groups protected by the legislation. Currently, the legal text refers to contempt for the condition of woman, associating it to discrimination or inequality on the grounds of “sex”. Using the term “sex”, however, is limiting and potentially exclusionary to transgender people. As explained on the report, a significant portion of political violence is directed at trans candidates and politicians. Thus, it is recommended that the law refer to gender-based discrimination, rather than sex-based, for the due inclusion of multiple gender identities.

Judiciary Branch and the Public Prosecutor's Office

7. Improving access channels and transparency mechanisms

The administrative and judicial bodies of the Judiciary Power must guarantee the right of access to information, in a transparent and clear manner, to individuals and legal entities (Art. 2, Resolution No. 215/2015 of the National Council of Justice). This transparency duty stems from the constitutional principle of publicity of procedural acts (Art. 5, LVXX of the Federal Constitution), which establishes the duty of transparency of the Judiciary Power in the acts and judgments by its bodies.

Access to information involves, therefore, the availability of data about the filing and registration of complaints and follow-up of cases, in a unified portal, with easy access for citizens and respecting the rights to privacy and data protection of the offended and offenders. We recommend, therefore, the creation of a Unified Portal for receiving and registering complaints, which can be accessed by any citizen and which contains constantly updated information on the progress of investigations involving cases of gender-based political violence. The unified portal enables social control, enshrines the principle of publicity and enables the production of knowledge and research on the phenomenon of gender political violence in the country, in addition to facilitating communication and understandings among different organs of the Judiciary Power in different states of the federation.

8. Collection and compilation of data on gender-based political violence

Currently, there are no consolidated and publicly accessible data on judicial cases of gender-based political violence in the country. Data on the receipt of complaints, on the monitoring of processes and on judicial decisions are not available for collection and compilation in their entirety and through unified and public portals. Civil society initiatives, such as MonitorA, monitor this information based on scattered data from some agencies in different states, with a number of methodological limitations. The absence of consolidated and unified databases prevents the construction of analyses about the totality of the phenomenon of political violence in the country.

In this sense, it is necessary that Public Power organs - such as the National Council of Justice -, which have the technical and legal capacity to access, collect and compile these data, periodically develop and publish reports on monitoring the judicialization of gender-based political violence cases.

9. Public Prosecutor's Office: improving reporting channels

Cases of online political violence should be swiftly addressed, since they not only represent a practice that affects the dignity of victims and that directly and fully harms the democratic system, but also because of the usual speed of content dissemination on digital environments and of the short term of electoral processes and dynamics. Thus, it is essential that cases of gender political violence be addressed during the period intended for electoral campaigns, since any delay in judicial decisions may pose risks to women's political campaigns.

Therefore, we suggest the creation of teams majorly composed of women within every Electoral Public Prosecutor's Office, which specialize and focus on the identification and investigation of cases involving gender political violence by Electoral Zone during the electoral period.

In addition to gender-based offenses and attacks, we also recommend that there be careful attention, in the judicial sphere, regarding demands involving analyses of irregularities in the transfer of funds and in the distribution of a minimum percentage of participation of women in the records of party candidacies, through the creation of teams focused on this specific analysis, suspending the party's registration in cases of suspected fraud.

10. Building strategies, within the scope of the Electoral Justice, to confront gender-based political violence and misinformation

In the 2022 elections, the Electoral Justice played a central role in combating misinformation along the electoral period, through the creation of strategies such as the Alert System for Misinformation Against Elections and the Permanent Program to Confront Misinformation, in which partnerships were established with social media platforms and other institutions, with the aim of combating the spread of misinformation.

The cooperation with platforms, information checkers and other institutions demonstrated the potential of articulation of the Superior Electoral Court to build strategies aimed at electoral integrity, gathering different social actors. Similarly, multi-sectorial strategies can be articulated by the Electoral Justice, with the aim of preventing and combating political violence against historically marginalized groups and against gender-based misinformation.

Executive Branch

11. Inclusion of combat against political violence in the governmental agenda

Confronting inequalities of gender, race and sexuality should be a central agenda for the Executive Power. The Ministries of Justice, Women, Racial Equality, Human Rights and Indigenous Peoples can - and should - work together to develop technical support and to provide human and financial resources, aiming to develop and consolidate public policies for the Internet, for ICTs and for combating political violence. The joint work of different organs of the Executive Power promotes a technical and theoretical plurality, enabling the construction of more precise and efficient policies to confront online and offline political violence.

In this sense, we reinforce the recommendation of the technical group of women during the transitional government for the creation of a Secretariat to Combat Gender-Based Political Violence, which would be an agency for coordinating, centralizing and monitoring debates around gender political violence, within the scope of the Executive Power.

Political Parties

12. Improving supporting tools for women candidates in cases of political violence, creating mechanisms for reporting and for internal liability

According to the Law on Gender-Based Political Violence, political parties must include on their statutes norms for preventing, repressing and combating political violence against women (Law No. 9,096/1995, Art. 15, X). In effect, however, parties have not yet significantly incorporated anti-violence practices into their internal structures.

It is critical that parties develop preventive and transparent policies to ensure the participation of women and historically marginalized groups, including provisions on how to prevent internal cases of political violence. In the cases in which the perpetrator is a partisan member, these organizations need to establish mechanisms for internal liability.

Finally, political parties should provide support for candidates who suffer gender-based political violence, offering reporting channels and supporting the prosecution of civil and criminal liability of aggressors.

Civil Society

13. Creating and strengthening networks about elections and violence (political and electoral, gender, race, etc.)

Civil society, through organizations such as Marielle Franco Institute, Terra de Direitos e da Justiça Global, Criola, Alziras Institute and many others, plays a central role in data collection and analysis, as well as in the construction of recommendations about political violence against historically marginalized groups. This work has been fundamental for the legislative and political advancement of strategies to combat gender political violence, as well as to aid candidates in seeking liability for their aggressors. Thus, it is essential to strengthen networks and coalitions dedicated to research, advocacy and lobbying on elections, and to ensure the safety and participation of historically marginalized groups in politics.

Among the joint actions that civil society organizations can lead are:

(i) Strategic litigation in cases of gender-based political violence, as a form of direct incidence on the Judiciary Power;

(ii) Mediation between victims of political violence and platforms, to ensure greater protection for candidates, politicians and women who act politically on social media; and

(iii) Partnerships with platforms to improve policies directed at hate speech, bullying and electoral integrity.

Louca, doida, maluca: misoginia domina ofensas a candidatas nessas eleições

Crazy, mad, wacko: misogyny dominates offenses against candidates in these elections

On the first analysis of MonitorA in 2022, the misogynistic practice of associating candidates to madness, hysteria or mental illness prevailed on social media. In addition to the terms "crazy", "mad", "wacko", "unbalanced", "hysterical" and “uncontrolled", there were suggestions for them to “undergo treatment", "meditate" or "get committed to a psychiatric institution". The discourse corroborates the idea that women are inadequate to institutional politics, and is violent to the point of leading many of them to withdraw from political life.

“Você é uma vergonha” e “mimimi”: ofensas a candidatas à presidência reproduzem falas de Bolsonaro

“You’re a shame” and “mimimi”: offenses to presidential candidates reproduce Bolsonaro’s speeches

Presidential candidates Simone Tebet (MDB) and Soraya Thronicke (UB) were targeted by 6,661,000 twitter offenses in only two days. The narrative was dominated by the expression "you're a shame", present in 1,050 tweets. In addition, nearly 1,000 (993) tweets used "mimimi" and "victimism" to offend or delegitimize candidates. More than 35.7% of the offenses identified on the tweets are misogynistic, and other 30.3% call the candidates "ridiculous", "pathetic", "liar", "hypocritical", "false", "scoundrel", "immoral", "shameless", "bloodsucker", "idiot" and "imbecile".

Mulheres negras e indígenas resistem à violência política no estado mais branco do Brasil

Black and indigenous women resist political violence in the whitest state of Brazil

In the State of Santa Catarina, 81.5% of the population self-declared as white, and the state has no black deputy. Its largest female bench in history had only five women. Within MonitorA, Portal Catarinas interviewed fourteen candidates in the 2022 elections. 92.9% reported episodes of political violence, 21.4% motivated by race.

Conflitos de terra e ataques de grupos bolsonaristas: a violência política de gênero no Centro-Oeste

Land conflicts and attacks by groups supporting Bolsonaro: gender political violence in the Center-West

The candidacy of two women from the Center-West region, Simone Tebet (MDB-MS) and Soraya Thronicke (UB-MS), has intensified political moods in the region. Both have faced a series of attacks on social media since the campaign began. The candidates to regional offices faced direct aggressions on the streets and even death threats. The scheme is repeated: attacks on candidates defending conservative and extremist positions on the right, with declared support for Jair Bolsonaro. The links between agribusiness - a leading sector in conflicts in indigenous territories - and bolsonarism also increase the tension in the region.

“A violência faz parte da nossa realidade, mas vivenciar isso como candidata me assustou”, diz Vanda Witoto, candidata indígena no Amazonas

"Violence is part of our reality, but experiencing it as a candidate scared me," said Vanda Witoto, indigenous candidate in the State of Amazonas.

The 2022 election campaign had 165 indigenous candidacies deferred - only 0.63% of the total candidacies. Of those 165, women candidates represent less than half. With the help of A Lente, we tell the story of Vanda Witoto (35), non-elected candidate to Federal Deputy. She gained notoriety in political militancy in Manaus, in Parque das Tribos neighborhood, which brings together more than 700 residents of 35 indigenous ethnic groups.

Ofensas nas redes sociais reforçam misoginia e xenofobia contra candidatas nordestinas

Offenses on social media reinforce misogyny and xenophobia against Northeastern candidates

"Shame," "cry" and "dumb" were common terms on tweets and comments directed at Northeastern candidates in 2022 elections. Although prejudice against Northeasterners is not new, it has been intensified since the 2018 elections, and is closely connected to the spread of hate speech and intolerance to divergent political positions. The attacks show that aggressors often have no knowledge about which are the Northeastern states, just as they do not know their social and economic realities.

Transfobia e violência sequestraram debate sobre candidatas trans e travestis nas redes

Transphobia and violence have hijacked debate about transgender and transvestite candidates on social media

We evaluated discourses directed at 11 transgender and transvestite candidates in the campaign for the National Congress, and found that the political debate they proposed on social media ended up hijacked by ideological-party conflicts, political violence, debates about public security and disputes over the representation of historically minorized groups. Transphobic comments represented an additional layer of violence.

“Basta ser mulher para sofrer violência política”, avalia candidata à vice-presidência, Mara Gabrilli

"Just being a woman is enough to suffer political violence," says vice presidential candidate Mara Gabrilli

Vice-president candidate on the group of Simone Tebet (MDB), Senator Mara Gabrilli (PSDB) found support on Law 14.192/2021, which typifies gender political violence, to denounce attacks received on social media. Gabrilli argues that gender political violence weakens democracy when it pushes women away from the political debate, but reinforces the need of facing fear.

Macumbeira e Micheque: nas redes sociais, violência da corrida presidencial sobra também para Janja e Michelle

Macumbeira and Micheck: on social media, violence from the presidential race also targets Janja and Michelle

In the most violent election in Brazil's recent history, the spouses of presidential candidates face intense attacks. Focused on religion and corruption, the offenses were almost three times more frequent against Janja da Silva, when compared to Michelle Bolsonaro. The president-elect's wife is most attacked on Twitter. During the analyzed period, offenses and insults were present in 15% of the interactions. The former first lady, who is not on Twitter, received 8% of the total number of analyzed offensive comments.

Keeping the tradition of previous editions, MonitorA 2022 served as material for victimized candidates to understand the dynamics of operationalization of online political violence. Candidate Soraya Thronicke, for example, used information from the observatory at the SBT presidential debate to question Jair Bolsonaro about the incitement of gender political violence.

The articles also served as support for debates of the Brazilian Academy of Electoral and Political Law (ABDEP), the main organization of professionals from the political and electoral fields in Brazil.

On the media

Só esse ano, publicamos nove reportagens, republicadas mais de 40 vezes. Nosso levantamento foi citado em 50 matérias em todo o Brasil. Entre elas, tivemos duas reportagens exclusivas na Folha de S.Paulo, uma n’O Globo e uma na coluna de Guilherme Amado no Metrópoles.

Our analyses at national, regional and local levels turn into news, content for social media and reports aimed at combating online gender violence, which offer support for demanding changes to social media platforms This year alone, we published nine news articles reports, which were republished more than 40 times. Our survey was quoted in 50 articles throughout Brazil. Among them, we had two exclusive reports in Folha de S.Paulo, one in O Globo and one in Guilherme Amado's column in Metropolis.

Republications

Clipping

This edition of MonitorA was released in September 2022. From there until the end of January 2023, the observatory of online gender political violence was quoted 53 times in Brazilian journalistic media, including articles on newspapers, magazines, TV, digital outlets and podcasts.

In these spaces, MonitorA’s data and analyses, as well as interviews with the team’s professionals, helped to qualify the national democratic debate and to broaden the discussion on the right of women to safely participate in institutional policy.

Among the outlets that published MonitorA's activities are Jota, Estadão, Folha de S.Paulo, Veja Magazine, Metropolis, O Globo, Yahoo and Marie Claire.

Access the English consolidate reports here