Critical Datafication and Data Feminism

The framework takes up two concepts from the first area of research that addresses data as tools of social mediation: critical datafication and data feminism. Both approaches reveal an unequal distribution of power, and critically draw attention to social problems through the production and analysis of data.

The premise of critical datafication is to redirect the data produced on social media to create divergent narratives that make social inequality and exclusion visible. Proponents recognize that, although government and businesses use data from social media to segment their public and for surveillance, they can also be used by research to bring attention to social problems and participate in placing these on media and government agendas (^{Ábrego Molina & Flores Mérida, 2021}).

Data feminism is defined as “a way of thinking about data, its uses and limitations, based on the experience, the commitment to action, and intersectional feminist thought” (^{D’Ignazio & Klein, 2020, p. 8}). Based on this definition, this approach focuses on recognizing systems of power and structures of domination that replicate inequalities embedded in the practices of gathering, communicating, analyzing, and presenting data.

Power and Algorithmic Visibility

On social media, the unequal distribution of power is expressed through systems that regulate the visibility of what is published. Although any user with an account can publish, not all content is made visible similarly (^{Cotter, 2019}). According to Twitter, their recommendation algorithms select, prioritize, and show content according to certain attributes of the tweets. These include their immediacy, their popularity as defined by circulation and reaction metrics, and the inclusion of multimedia content, such as images and videos.

Twitter has two ways of showing content to its users on their timelines: the first orders it chronologically, the second, according to recommendation algorithms (^{Twitter Help Center, 2022}). In the hierarchical organization of the algorithms, rankings tend to show fewer tweets with links to external sites, for example, to news sites (^{Bandy & Diakopoulos, 2021}). The automated operations of selection and ranking thus establish a relationship of mutual shaping: the algorithms select and prioritize content that corresponds to its parameters while, simultaneously, the content takes the shape of the formats preferred by the algorithms.

Besides the visibility logic, the importance of the reciprocal relationship between Twitter and the media in creating an agenda becomes evident. What is expressed on Twitter acquires greater visibility through the coverage of mass media and the press (^{Suh & Borah, 2019}), since one of its uses is, as has been shown repeatedly, a social gage in the face of a variety of circumstances, as well as an amplifier and replicator of a reduced number of messages that create an echo chamber (^{Colleoni et al., 2014}).

Social movements have been able to challenge the unequal distribution of power of algorithmic visibility, setting the stage for struggles and creating an objective that political efforts can strive for (^{Velkova & Kaun, 2021}). Studies have shown that the high algorithmic visibility of political messages shaped by online platforms, labeled vernacular because of their appropriation of platform communicative guidelines to produce viral messages (Sued et al., 2022). At the same time, it is important to consider messages with lower visibility, as these may contain other types of signifiers of interest to politics and activism through the manifestation of commonplace citizen political participation (^{Castillo-González et al., 2022}).

Femicides and Demands for Justice

In Latin America, two-thirds of women have experienced some type of gender-based violence (^{CEPAL, 2020}). Femicides are the result of the progression of this violence. According to ^{Lagarde (2004)}, femicide “occurs as the culmination of a situation characterized by repeated and systematic violation of women’s human rights” (^p.1). In the context of Latin America, impunity is also intrinsically linked to the phenomenon of femicide, given that the execution of justice carried out by the State is permeated with a “profoundly misogynistic perspective” (^{Lagarde, 2006, p. 223}). According to ^{Segato (2016)}, femicides are not only a consequence of impunity, but also their creators and replicators, because they seal pacts of silence between collective accomplices and patriarchal groups that express their power through aggression and the abuse of female bodies.

Facing the inefficiency of the State and the media, social networks have created an alternative space for women to express their claims, “as citizens [who] demand that the authorities treat them with dignity and justice” (^{Cerva Cerna, 2020, p. 181}). Thus, the demands for justice are presented, on one side, as a condemnation of the refusal of the media, authorities, and institutions to “make visible the existence of the distinct category of these crimes, that must be specifically differentiated, classified, and investigated” (^{Segato, 2016, p. 87}), and on the other, as a collective embodiment (acuerpamiento), understood as an action to resist the injustice other bodies have experienced and offer closeness and renewed strength “to recover joy without losing the outrage” (^{Cabnal, s.f., s.p.}). In the words of ^{Reguillo (2021)}, they allow for the creation of a counter strategy that proposes “remembering the victims and denouncing the consequences of violence” (^{p. 156}).

Visibility of Demands for Justice

Within Base 1, made up of 53 963 tweets, 927 were identified as demands for justice, occupying a minimal space within the general database, corresponding to 1.7 per cent. These tweets are related to a total of 96 particular cases, identified through the structure of the hashtag #justiciapor or #justicia para and a name (Figure 1).

Purple: femicides. Pink: cases of violence.

Source: Created by the authors.

Figure 1 Demands for justice with the name of the victim 

The specific goal is making the victims for whom justice is being demanded visible and naming them, and to avoid an academic objectivization, Table 1 is a composite of the women’s names for whom justice has been demanded, generated in Latin American countries using Twitter. The segmentation of the cases is thus shown, which are mostly linked to reports and cases denoted as femicides, but unresolved; only in three instances does the tag #justice refers to a verdict and sentencing of the aggressors. These trials were concluded between two to four years after the femicides, and in two cases there have been appeals filed. The other 81 cases show a current demand for justice during the period examined, since no resolution for the victims has been obtained.

The tweets that do not specify cases are divided among general hashtags such as #justicia (justice), #justiciaparatodas (justice for all females), and #justiciaparalasniñas (justice for the girls). The difference between the first and second group is that only the tweets that specify their demands reach a high, medium or low visibility. Those that call for justice in general terms are found in the low or null segments. Because of this, the analysis focuses on tweets of the first kind, whereas the others are only mentioned in sections related to low and null visibility.

Given the comparison that is shown in Figure 2, it can be deduced that, whereas the visibility of the feminist struggle against gender-based violence (pink) is highly visible in general, it increases until the apex on November 25th and the following days, there is a different pattern among the demands for justice (purple). In this second segment, visibility is related to the temporal proximity of the cases that occurred during that period, not the commemoration or recognition of murders of women that occurred in the past. Here, the correlation between the hashtags that demand justice and those that are more general -#niunamenos (not even one more), #25N- attract visibility of the case and relate it to others.

Source: Created by the author.

Figure 2 Visibility of #justicia (justice) in relation to #niunamenos (not even one less) 

Despite this, Figure 3 shows that when compared to the tweets from the general database, the demands for justice have higher visibility, above all in the high and medium segments. Whereas the high visibility publications make up 1.5 percent of Base 1, in Base 2 they increase to 5 percent. In the case of medium visibility, the tweets in Base 1 make up 4 percent, while those of Base 2 increase to 6 percent.

Source: Created by the author.

Figure 3 Comparison of visibility 

Characteristics and tactics of demands for justice

Returning to the need manifested by Salguero (^{D’Ignazio & Klein, 2020}) to stress the collection of data and details of each case of gender violence, including name, age, location, etc., we present the cases that gained the greatest visibility. Marisol Cuadras was an 18-year-old Mexican feminist activist, murdered November 25^th, 2021, precisely when she was participating in a protest against gender violence in Guaymas, Sonora within the context of 25N. 19-year-old Brisa Formoso was a victim of femicide on November 15^th, 2021 in Ranelagh, Buenos Aires, Argentina. The third case is related to the remembrance of the disappearance of Wendy Sánchez in the state of Nayarit, Mexico the day of her birthday. On January 9, 2021, as Wendy turned 33, she disappeared as she traveled to the city of Guadalajara.

When analyzing these high visibility cases, we identified the following characteristics. First, highly visible cases respond to immediacy, not the commemoration of the Day for the Elimination of Violence towards Women; this is verified by through the use of #niunamenos as represented in Figure 1. Immediacy here refers to the fact that the six femicides, the arrest, and the commemoration of Wendy Sánchez’s birthday occurred during the period within which the data sample was collected. These cases circulate quickly and massively amongst female users. They are broadcast and amplified by using a variety of resources available on the Twitter interface, above all a commented tweet, which happens when a user takes an existing tweet and adds her own comment.

Other resources include videos, especially one that captures the moment that the father of Brisa Formoso learns about the femicide of his daughter, appealing to a strong emotional response, and others that mention public figures, civil servants, politicians, and public prosecutors. Users that share and comment on these tweets show empathy towards the victims, and they include themselves in a collective “they are killing us all” that places them in a subjective position. Although the most visible cases are those that receive media exposure, neither the voice of the media nor the news item are amplified, but rather other stories, other voices that are present in the Twitter sphere. The former makes visible an autonomous and independent expression of traditional media and press. According to these characteristics, these demands for justice are visible because they adapt to Twitter’s algorithmic characteristics, as mentioned previously: immediacy, popularity, highly emotional videos that become viral, and links to online content. It can even be inferred that the higher visibility in Base 2 when compared to Base 1 is due to the timeliness of the tweets in reference to the events they mention. The high visibility metrics show isolated peaks in Figure 2 since, although the events in this segment are highly visible in terms of engagement, it must be said that this visibility is fleeting: high profile cases that achieve high visibility only last a few days.

Correlations were also sought between visibility and access to Twitter in different countries (Figure 4). To support this, the percentage of women who use Twitter in each country where the events included in the sample occurred were calculated.³ The results were that in Argentina female users of Twitter represent 5.88% of the total population, in Chile the number is 5.54%, whereas in Mexico the percentage is 4.02%. Although these numbers are lower than male users of Twitter (Mexico with 6.32%, Argentina with 9.25%, and Chile with 12.13%), they show a considerable force, mostly generated by female users interested in presenting an agenda and feminist demands in cyberspace. This is common in countries with fewer female users, as is the case of Guatemala (1.64%), where, despite the small number, cases are positioned in high and medium segments. The countries that bring most cases to light are Mexico and Argentina; both have a young and consolidated body of activists thanks to the struggle for the legalization of abortion and elimination of violence (^{Larrondo et al., 2019}). Data indicates that digital activism on Twitter is effective at placing demands, and it would be more so if the base of female users and activists were expanded. This is paradoxical given the level of violence and discrimination they experience on this platform (^{Luchadoras MX, 2017}). However, there are nuances in this correlation since, as seen above, the visibility of these demands also depends on elements like timeliness and the obstruction of the case, that can vary between countries during the studied period.

Cases of high and medium visibility may be repeated in lower segments.

Source: created by the author.

Figure 4 #justiciapor (justice for). Cases according to country and visibility 

Press media were also used to monitor the judicial progress of highly visible cases. It was found that only in four cases has there been a preventive arrest of those accused of femicide, despite the efforts of activists, academia, and civil society to position the need for justice in the context of the demands feminists have made to governments and judicial institutions.

The strategy that prevails within the medium segment, however, is the demand for dissemination and visibility of the cases. In this instance, family members or activists appeal to female journalists and public figures asking them to aid in broadcasting the case or, in the case of the public prosecutors, that they move forward in determining the facts of the situation. A quoted tweet does not produce results as quickly or spontaneously as in high visibility cases, but the resource continues to be valuable. This segment reinforces, through these quotes, performativity as a communicative act whose purpose is to provoke an action by the intended recipient, through a plethora of posts: the purpose of this reference is to request visibility and pressure to deliver justice. Although the visibility of these cases is lower, this segment presents a strategy of resisting the algorithm: asking each other for help, collaboration, and solidarity. Two of three femicides that are found in this segment, one of which is a transfemicide, are immediate cases, while only one refers to the posting of a memory of a prior event.

When considering the correlation between the level of access to Twitter and digital activism, the cases presented in this segment are mostly different. Within the low visibility segment, most of the sample cases are found. Sixty-one of the 96 cases identified are located here, nearly all of them reported as femicides. They generally correspond to earlier, low profile and unpunished cases, previous to November 2021. There are also more countries that present cases in this segment (see Figure 3).

Within the null visibility segment, smaller and alternative media publications are found that inform on the progress of causes or convocations to local protests. Very few cases appear for the first time in this segment, only 12 out of 96, the rest of which have achieved some level of visibility. They are also older or unpunished femicides; some posts indicate progress in the cases, for example, hearings and arrests. General demands are also found, such as #justicia (justice), #justiciaparatodas (justice for all women), and #justiciaparalasniñas (justice for the girls).

Low and null visibility is correlated with the immediacy that Twitter’s recommendation algorithm demands. According to what is known about the algorithm’s operations, claims for justice that are not related to a recent case do not fulfill the timeliness requirement. However, in these segments, the demands for justice represent an effort to keep the murdered women’s names visible despite the passage of time. A non-algorithmic visibility is created, but one that is human and persistent, based on memory and the permanence of the case in the public sphere.