Social Justice in HCI: A Systematic Literature Review

Given the renewed attention on politics, values, and ethics within our field and the wider cultural milieu, now is the time to take stock of social justice research in HCI. We surveyed 124 papers explicitly pursuing social justice between 2009 and 2022 to better reflect on the current state of justice-oriented work within our discipline. We identified (1) how researchers understood the social justice-relevant harms and benefits, (2) the approaches researchers used to address harm, and (3) the tools that researchers leveraged to pursue justice. Our analysis highlights gaps in social justice work, such as the need for our community to conceptualize benefits, and identifies concrete steps the HCI community can take to pursue just futures. By providing a comprehensive overview of and reflection on HCI’s current social justice landscape, we seek to help our research community strategize, collaborate, and collectively act toward justice.


INTRODUCTION
"If you have come here to help me, you are wasting your time, but if you have come because your liberation is bound up with mine, then let us work together."-Aboriginal activists group, Queensland, 1970s The field of Human-Computer Interaction (HCI) is envisioning and building a more socially just future.There has been incredible growth in the area of social justice, as evidenced by the proliferation of justice-related terms and studies (e.g., racial justice, design justice, feminist studies, decolonial HCI, and more).At CHI 2022, where this project began, there were several workshops focused on collectively imagining what justice could look like in HCI research [12,114,179].The keynote from Dr. Kishonna Gray urged us to "pay attention to minority voices" and look for "precursors" to harm [78]. 1 There is a strong drive from within our community of technologists and researchers to move forward with equity and care.
Still, there is a disconnect between our future intentions and the field's current state.The disconnect is present because, as a field of practice and scholarly research, HCI does not have a shared understanding of how to work toward justice.This is evidenced by our ongoing discussions and reflections on our own and broader communities' ethical, scholarly, design, and research practices.For example, scholars have raised concerns about extractive practices when collaborating with community partners [119,173], about who is cited [104], and about our review systems favoring authors with more privilege [57].Junior researchers across many campuses have organized due to precarious living situations [99,186,196,197] and, at the time of writing, many authors and reviewers are boycotting CHI 2024 [84].We see these ongoing discussions as a sign of healthy and vibrant engagement with questions of justice.While our intentions toward justice are clear, we have some work to do in our scholarly and designerly practices and within our academic and professional systems.There is an opportunity to think more deeply about the horizon we are working towards and how we might take steps towards it [54].
The need for deepening our understanding of social justice is called out by Bellini et al.: "there appears little critical discussion around what is meant by the justice inherent to social justice, despite there being calls to adopt such a lens" [12].These definitional tensions reflect the vibrancy of our ongoing conversations and indicate our discipline's commitment to the area and practice.We do not believe that HCI requires a single, coherent, unified definition of justice.Still, there is work to be done in our discipline as we strive towards justice.It is challenging to understand where our field should go without understanding where we are.What are our goals as a field when conducting such work?What commitments do we make?What practices do we espouse?What strategies can we use?How does a new researcher get started?Without a shared understanding, we limit our opportunities to reflect, collaborate across the field, strategize, and take collective action.
In this paper, we seek to understand how HCI research currently engages with concerns relevant to social justice so that we might better identify harm, acknowledge potential consequences, and work towards better futures.We review 124 full papers to understand (1) the landscape of harms and benefits that are currently being addressed, (2) the strategies researchers are taking to pursue justice, and (3) the tools researchers use to support thoughtful and equitable research processes.We contribute a set of key considerations that researchers can use to investigate justice-oriented questions and concerns and a discussion of directions the HCI community can take to address gaps and move towards more just futures.We hope this work sparks imaginations and seeds new ideas about how to support social justice work in HCI. 1 Looking for precursors to how our sociotechnical systems can be used to perpetuate societal harms and create new injustices allows for prevention instead of only reaction.

About our framing
To understand how contemporary HCI researchers conceptualize social justice, we planned to investigate how researchers explain and define social justice.We were surprised to find little explicit conversation of justice.Instead, we commonly found descriptions of the harms that require justice and social change, including marginalization, exploitation, oppression, and vulnerability.Uncovering how authors considered dimensions of social justice, including these harms and benefits (Section 4) and what level researchers acted on (Section 5), became the focus of our work in understanding how HCI researchers conceptualize social justice work.In doing so, we captured how authors in HCI pursued justice by reacting to problems of social injustice and the core tools they used to do so (Section 6).
We were conflicted in taking this approach.Centering injustice in this review could reduce people to the issues they face, leading future researchers into the trap of the deficit view [49,168] or damage-centered design [183,198].We are further hesitant to use this framing as we do not want to implicate all authors within our corpus as reproducing harmful deficit or damage-centered narratives.At the same time, harm can serve as an entryway to begin considering social justice issues and measuring the impact of sociotechnical systems.Future research agendas can engage with injustice while focusing on joy, care, and wholeness [88].With this work, we elevate the harms and injustices our community works to rectify, as well as the concerns plaguing certain historically marginalized groups.

BACKGROUND 2.1 Turning Towards Justice in HCI
In this section, we share a trajectory of social justice work in HCI to demonstrate how the field's core concepts relevant to justice have evolved.As early as the 1980s, HCI adopted Participatory Design from Scandinavian workplace environments to democratize design practice and strengthen end-users' influence on system development [26].Participatory design emerged as a response to unilateral decision-making by management and made commitments to "rebalance power and agency" in the workplace [5].Building on this work and recognizing a need to incorporate value commitments into the design process more intentionally, Friedman introduced Value Sensitive Design in 1996 [73].While Friedman did not focus explicitly on questions of injustice, early work examined bias in workplace systems and found that biased computer systems can be "difficult to identify let alone remedy because of the way that technology engages and extenuates them." Friedman and colleagues concluded that "biased computer systems are instruments of injustice" and that "freedom from bias" should be amongst the select few values considered when evaluating computer systems [74].
As HCI moved beyond traditional workplace environments, HCI scholars showed increased interest in addressing complex social problems such as those concerned with economic and social development and environmental sustainability [54].Early work in this area, however, promoted individual narratives of behavior change [50], technological progress at the cost of political and social sustainability [59,89], and disempowering models of charity [54,79,152].For example, some ICTD projects which aimed to support socioeconomic development internationally were critiqued for viewing "poverty" as a thing to fix [54,64,89] and "perpetuating already-uneven economic relations by transforming individuals into consumers" [54,59,89].Critical scholars argued that without attending to system-level factors such as the distribution of power and privilege, design risked reproducing and perpetuating harm [40,54].Lin et al. asserted that HCI's tradition of building "useful" things was particularly insidious because it masked various forms of violence and social injustice behind narratives of progress [119].
Since the 2010s, HCI scholars have more explicitly taken up social justice as an orientation to doing HCI work (See Figure 2).Dombrowski, Harmon, and Fox describe social justice as an approach that attends to the "ways that individuals experience oppression, including how benefits, burdens, obligations, power, opportunity, and privilege have been (in)equitably distributed within society" [54].When related to HCI, an approach to justice often means a concern for how systems of "oppression, such as racism, sexism, ableism, ageism, classism, and so on, impact people's experiences with technology, information, and design" [53].When doing this work, HCI scholars have turned to different strands of justice, such as Disability Justice or Transformative Justice [12].These strands challenge complex systems of power and oppression and offer visions of more equitable futures.For example, Transformative Justice is a community-based form of justice developed by the LGBTQ+ community that centers on healing, accountability, and the transformation of harmful structures as a response to state-sponsored violence [51,127].Disability Justice interrogates how disability is conceptualized in order to advance inclusion and equity for disabled individuals [88,97].Bellini and colleagues refer to these various strands of justice as "mosaics" because while distinct, they are "tightly interconnected" and "demonstrate the complex patterned nature of social justice HCI" [12].A focus on social justice in HCI often refers to a focus on the complex pattern of marginalized experiences, identifying systematic oppression and pressures, and identifying new, prefigurative futures.HCI scholars have proposed various frameworks to explore the role design can play in helping us move towards these mosaics or visions of justice.Irani and colleagues, for example, introduced postcolonial computing as an orientation to design work that centers the role that "global power, wealth, economic strength, and political influence" play in shaping cultural encounters in "the developing world" [89].We discuss more about these frameworks as a tool for pursuing social justice in Section 6.
While the HCI community is having encouraging conversations about justice, they remain surprisingly disparate [12].Independent progress is being made within separate strands of justice, making it challenging to build shared understandings and learn from one another.It is a great time to reflect on the current landscape and consider how we might want to move forward as a scholarly community.This paper reviews how relevant scholarship describes and engages with justice-oriented concerns.We identify the harms and benefits currently being addressed, the strategies researchers employ to address injustice, and the tools they use to support justiceoriented work.Collectively, these offer us a set of key considerations with which to think about social justice in HCI.

Conceptualizing Systems of Power
Social justice work requires attention to the broader social and political systems that produce injustice because technology often embeds and reproduces these systems.Black feminist scholar Patricia Hill Collins introduced the Matrix of Domination to examine how society structures power.Collins conceptualized race, class, and gender as interlocking systems that shape the lived experiences  [158] from Morgan's Axes of Oppression [135] of Black women and other intersectional identities [36].Collins also explains that individuals experience and resist oppression on three levels: the personal, the communal, and the institutional [36], which we will further unpack in the findings section.
Black feminist legal scholar Kimberlé Crenshaw introduced Intersectionality as a theory in 1989 [41].Yet, as Rankin and Thomas acknowledge [155], Intersectionality's history originates before Crenshaw.The Combahee River Collective Statement is often acknowledged as one of the first recorded discussions of Intersectionality [35].Crenshaw discusses how those with multiple marginalized identities experience oppression as an additive, compounding phenomenon rather than a singular, disconnected phenomenon.Indeed, "the intersectional experience is greater than the sum of racism and sexism, so any analysis that does not take intersectionality into account cannot sufficiently address the particular manner in which Black women are subordinated" [41].' The idea has picked up momentum within academic and activist circles.For example, Disability Justice activists list intersectionality as their first and primary principle, along with a refrain from poet activist Audre Lorde: "We do not live single issue lives" [88].
Morgan, a gender studies scholar, developed the Axes of Oppression in 1996.The axes visually represent how systems of oppression privilege some identities and oppress others [135] (the top half of Figure 3 radially displays identities that hold power and privilege while the bottom half displays identities that have the potential to be oppressed due to the systems of oppression, such as racism and ableism).Privilege and oppression are unevenly distributed amongst individuals and communities based on their position within the axes of oppression.Morgan writes that this position is "simultaneously a locus of our agency, power, disempowerment, oppression, and resistance" and shares the necessity for "both awareness and honesty with respect to our own positioning on the various axes of this grid" [135].
Systems of power are intertwining and additive in ways that make it challenging to see, engage with, account for, or attempt to remedy the unequal distribution of benefits and harms that they produce [40].By focusing on lived experiences of benefits and harm, we deepen our understanding of how systems of power manifest for individuals and populations.Furthermore, scholars have noted that a focus on the lived experience "is a useful way to establish what is meant by harm, and indeed in gauging or deciding ways to measure it" [27].In this paper, we examine the distribution of benefits and harms as a way to conceptualize and do the work of justice.

METHODS AND CONTEXT 3.1 Data Collection and Analysis
To develop our corpus, we collected articles that explicitly discussed "justice." We built on Bellini et al. 's observation that the many mosaics, or strands, of justice"demonstrate the complex patterned nature of social justice HCI " [12].Bellini et al. list five such mosiacs: citational justice [104], research justice [3,113], disability justice [88], restorative justice [203], and environmental justice [126,164].
Building on this insight, we created a search query by concatenating these five mosaics with four others that we are familiar with as HCI researchers: transformative justice [51,127], reproductive justice [160], economic justice [4], and racial justice [101,171].We searched the ACM Full Text Collection using this initial query, selected a random 20% of the resulting set, and iteratively added search terms as they appeared in this subset.We identified six additional mosaics (structural justice [131], distributive justice [156], information justice [128], intergenerational justice [132], language justice [106], design justice [40]) and searched the ACM Full Text Collection using this final query in February 2023: "social justice" OR "racial justice" OR "disability justice" OR "restorative justice" OR "citational justice" OR "research justice" OR "environmental justice" OR "transformative justice" OR "economic justice" OR "reproductive justice" OR "structural justice" OR distributive justice OR "information justice" OR "intergenerational justice" OR "language justice" OR "design justice" We acknowledge that this list is not exhaustive.During the coding process, we came across additional mosaics that were not included in our final query, including mob justice [177], testimonial justice [151], and hermeneutical justice [151].Furthermore, we found that terms such as "marginalization" and "oppression" were commonly used in justice-focused research but were not included in our query.Our approach only captures research that clearly labels itself as social justice-oriented, and while this is a limitation of this survey, we were able to identify a meaningful number of papers that nevertheless provide valuable insights.
To bound the corpus to a manageable size that still covered topics across HCI, the research team chose to restrict our search to four keystone conferences: CHI, DIS, UBICOMP, and CSCW.While we acknowledge that justice-related conversations are happening in many different venues, we believe the four chosen venues cover the breadth of research in HCI and capture our core conversations 2 .We further excluded papers if the primary goal of the work was not justice-oriented.Papers were excluded from the corpus based on four criteria: (1) Justice was discussed strictly by participants (e.g., as a participant quote).(2) The justice-related term was used strictly as an example (e.g., "...representation, beauty, language, self-transcendence, subjectivity, creativity, interpretation, identity, self-determination, and social justice, among others" [7]) (3) The justice-related term was only used in the discussion or conclusion as a direction for future work rather than as a goal of the current paper.(4) The justice-related term was used without a discussion of the "ways in which people experience oppression and marginalization" [53].For example, ICTD papers that lacked a discussion of colonialism or marginalization were excluded.
to include or exclude were made by consensus.Our final corpus included 124 total papers, 93 from CHI, 22 from DIS, 8 from CSCW, and 1 from Ubicomp, which spanned the years 2009 to 2022.We have attached a complete citation list of all publications in our corpus in Appendix B. An overview of the corpus can be found in Appendix A: corpus at a glance.Four members of our team went through three rounds of coding to ensure consistency across the corpus.In the first round, we coded 10% of the corpus.For each paper, four authors independently coded the paper and met weekly to discuss challenges and necessary changes to the codebook.In the second round of coding, we tested the codebook on another 10% of the corpus where each paper was independently coded by two of the first four authors.Our iterative process helped us refine our questions and address the most salient factors.In our final round of coding, all 124 papers in the corpus were independently coded by at least one of the four first authors.The final set of codes collected for each paper included: (1) the title, author, year, publication venue, and contribution type; (2) the problem statement; (3) where or if the paper explicitly called out justice; (4) frameworks employed; (5) methods; (6) participants; (7) the type of oppression addressed by the paper (guided by the Axis of Oppression); (8) the approach used by the researchers to resist oppression; (9) the material benefits to the participants; and (10) self-disclosure, reflexivity, or positionality statement.
Guided by Braun and Clarke's approach [23], we then conducted a reflexive thematic analysis.Reflexive thematic analysis is an approach to analyzing data that acknowledges and "fully embraces the subjective skills the researcher brings to the process" [23].Through the screening, data extraction, and analysis processes, the first four authors kept independent memos to note questions, reflections, and insights.The notes in these memos, particularly those involving edge case papers, seeded rich conversations during weekly team meetings over five months and helped evolve our understanding of the corpus.We analyzed our final set of codes inductively, where each researcher took two or three of the ten codes listed above and analyzed them using affinity diagramming.We used weekly team meetings to gather feedback on our analysis, share insights, discuss alternative interpretations, and ultimately settle on the core themes reported below.

Positionality
Our driving motivations behind this paper stemmed from our collective interest in understanding how best to pursue socially just aims in our own work and positioning that work within the broader context of social justice in HCI.As community members, we have been inspired by the changes occurring in our field.We wanted to write a love letter to the community highlighting our progress and making room for more.As individuals who care deeply about the uptake of justice, we hoped to orient ourselves and others to the many discussions in our field.
Our team consists of eight researchers, four graduate students, and four faculty members across six academic institutions in the United States.Members of our team have struggled with the effects of marginalization in a multitude of contexts as researchers and beyond.We are each pursuing social justice research in different contexts, including civic technology, community activism, workplace technologies, mental health as well as race and disability.Some of us were more comfortable with taking on the identity of being a social justice researcher, while others were coming to terms with framing their work in that manner.
We are each a part of the HCI community which we have surveyed.Several authors wrote papers that were included in our corpus.Additionally, we knew many of the authors in our corpus as colleagues, which may have influenced how we perceived their work.
Recent calls from within the community have been made to avoid CHI 2024 in Hawaii based on a history of colonialism, overtourism, environmental degradation, and the recent devastation in Maui [84].Our decision to submit to CHI this year was not taken lightly.While our team members have chosen to withhold several other publications, we struggled with the irony of withholding a social justice paper in this context.The call to support Hawaiian's desires highlights our community's need to contend with justice.It is a reminder that we are part of a system that can be harmful to communities and perpetuate power imbalances.We also weighed the professional impact on our careers and academic pursuits as a blended group of junior and senior researchers, some of whom felt pressured to publish as early-stage professionals.The professional, political, and ethical tensions around publishing illustrates the potential precarity of enacting justice-oriented work for those with less privilege.
Throughout this project, we grappled to understand the current state of social justice research and its practices.In the following sections, we share how our field conceptualized social justice through harms and benefits, explain how researchers approached change on multiple levels, and show some concrete tools they used.

HARMS AND BENEFITS
"Most design processes today therefore are structured in ways that make it impossible to see, engage with, account for, or attempt to remedy the unequal distribution of benefits and burdens that they reproduce."-Sasha Costanza-Chock, activist, researcher and designer [40] In this section, we map conditions of injustice that authors responded to by sharing how authors discussed: (1) the form of harms and benefits, (2) who is being harmed and who benefits, and (3) where researchers locate the cause of harms and benefits.
While conducting this analysis, we found that authors disproportionately framed their work around harms rather than benefits.While 102 out of 124 papers included a discussion of harm in their paper's introduction, only 34 included a discussion of benefits, which remained limited even when included.Even though there was limited data on benefits in our corpus, each aspect of harms and benefits introduced in this section includes a discussion of how benefits were brought to light.

Harms & Benefits Addressed
4.1.1Forms of Harm.In order to identify what forms of harm are being addressed by HCI researchers, we turned to zemiology, the study of social harms [27,85].Zemiology emerged from critical criminology as a way to highlight the range of harms experienced by society that extends beyond those that are caused by crime.Canning and Tombs provide a provisional typology of social harms, which we use to understand the lived experiences of injustice that are addressed by the papers in our corpus [27].The typology included physical harms, psychological and emotional harms, financial and economic harms, cultural harms, harms of recognition, and harms of autonomy.In addition to their typology, we have included environmental harms, which was a category addressed by several papers in our dataset and reflects HCI's nascent turn to non-human impacts of computing [21,120,166].A full description of each category, along with examples of how these harms appeared in our corpus, can be found in Table 1.In line with Canning and Tombs, we emphasize that these categories of harm are not mutually exclusive, but rather, are complex, interrelated, and synergistic.
In our corpus, researchers paid the most attention to recognition harms, referring to one's ability to engage in society (n=45), and autonomy harms, referring to blocked capacities and self-actualization (n=43).The attention to these two forms of harm reflects a broad interest in the field of power, identity, and access.While recognition and autonomy harms are important and require extensive attention, other forms of harm have been understudied.Less attention was given to environmental harms (n=8) and cultural harms (n=3).The lack of discussion around these forms of harm indicates a dearth of language for talking about lived experiences under environmental and cultural turmoil.Colonialism, despite being widely addressed within our corpus, was often related to recognition harms, contending with identity aspects of colonialism."Harm that arises through the destruction or undermining of particular cultures or ways of being" [184], and "harms that result by the imposition of a particular culture" [27] were rarely attended to in the framing of social justice work.In addition, we observed little attention to environmental harms, particularly harms that impact non-human individuals, including animal and plant life and broader ecologies.Without explicitly naming these harms, we constrain our discussion of environmental justice to a human-centered one.
Many papers also carefully attended to the complexity of harm.Of the 102 papers that included a discussion of harm in the introduction, 52 discussed multiple interrelated forms of harm.For example, Lu et al. contributed a rich description of the ways surveillance manifested in forms of psychological and emotional harms (student feelings of discouragement and embarrassment), harms of recognition (reducing students actions to misbehavior), and autonomy harms (penalizing students, especially students with disabilities) in the classroom context [122].Additionally, we found cases where one form of harm contributed to harm of a different form.Harms associated with racial profiling highlighted the possibility for recognition harms to produce physical harms in the form of injury or even death.Bosley et al. argued that technological solutions and design interventions in policing "are often rooted in the continued racial profiling of poor and socially marginalized communities" [22].In this case, racial profiling, a harm of recognition, combined with structural police violence, met to create physical harms that disproportionately impacted poor and socially marginalized communities.We encourage authors in HCI to continue to attend to rich, complex accounts of the many forms that harms take.[22], domestic violence [12,34], "suboptimal vaccination" [45], "lives lost due to violence" [176], "increase in diet -related diseases, as well as rising hunger and malnutrition rates" [153], "physical or bodily injury or harm, including gun violence that results from conflicts between individuals or groups" [58] Psychological and Emotional Harms Psychological states or emotional impacts of traumatic events or ongoing distress.

Financial and Economic Harms
Monetary harms that affect individuals, households, or wider communities.

20
"barrier to getting a job" [190], "hidden fees and social media gamification strategies that compel unwanted financial risk" [42], wage theft [194], "overproduction and free trade agreements flood and destroy local markets of developing and developed countries alike" [153], "low wages, and lack of hazard pay, " [

" [109] Autonomy Harms
Reduced capacity, opportunities, or potential for self actualization

43
"inequitable access to basic human rights (e.g., health care, education, housing, employment opportunities)" [145], "excluded from accessing care" [149], "impede disabled people's already limited access to public space" [14], "work is often invisible or not valued" [44], "lacking in some aspect of digital access" [147], "limiting Native American individuals' potential for political engagement through digital means." [189], "taboos generated by the marginalization of women directly inhibit speech and information seeking" [110], Environmental Harms Arise from human interaction with other species and the natural environment.

8
"local air pollution" [11], "Sea-level rise" [167], "urban heat" [105], "unsustainable food system, air pollution, contribution to climate change, loss of biodiversity, and low animal welfare." [153] Careful attention to forms of harm is important because forms of harm frame problems of social injustice.Shifting the problem frames we use to conceptualize social injustice shifts how we pursue social justice.Pei and Crooks, for example, argued that approaching the digital divide as a problem of distributing digital access constrains the solution space and ignores the social inequalities that root technical disparities, ultimately failing to produce equity [147].Framing the digital divide as an economic harm or a harm of recognition as opposed to a harm of autonomy produced a different understanding of the experience of digital connectivity and implied an alternative solution space.Thinking about how we frame problems, including their interrelated and additive dimensions, can illuminate new perspectives for thinking about social injustice and the approaches we use to address it.

Forms of Benefit.
Mirroring the discussion of harms, papers in our corpus also discussed benefits.The ability for technology to broadcast information and capture attention was discussed as a benefit of recognition, an ability that afforded power and increased the ability for individuals and groups to engage in society [129,141].In addition, it was widely recognized that technology has the potential to increase access to information [17] and resources [143,159] -benefits of autonomy.Other forms of benefits, such as positive health outcomes [86], financial benefits such as employment [157], and psychological and emotional benefits like feelings of safety [34] were also referenced in our corpus.
Discussions of benefits, however, remained limited.A common pattern was to acknowledge that technology produced benefits and then identify the limitations of that technology that also produced harm.For example, Tuli et al. discussed how menstruation trackers benefit people who menstruate by facilitating a better understanding of one's body, avoiding stigma around staining, and providing a resource for natural birth control [188].However, the same systems contributed to the medicalization and control of menstruating bodies.These examples illustrate a pattern in the discussion of benefits in our corpus: they emphasize the positive potential of technology while critiquing elements of its current form.
4.2 Who is Harmed and Who Benefits 4.2.1 Who is Harmed.Harms are not experienced evenly across populations; they disproportionately affect those on the periphery of power.67 of 124 papers named a clear group that experienced disproportionate harm (see Table 2).
The majority of papers in our corpus discussed single-axis identities.Researchers most commonly focused on working identities as well as social identities.Social identities such as gender [34,44,110,129,188], race [22,24,55,58,181,182,189], and ability [14,17,109,168] received significant attention, while there was less attention on the challenges that trans and nonbinary people [72,77,80], LGBTQ communities [75], and older adults [108,140] face.Papers which focused on people's working identities primarily discussed invisible work, such as that done by artists and crafters [157], mothers [123], and health workers [10,187], and hazardous work, such as that done by content moderators [170] and sex workers [174][175][176].Many marginalized identities have received little attention from social justice researchers in HCI.Some examples include indigenous people, unhoused people, non-citizens, people with large body sizes, and religious minorities.
Only 17 of 124 papers in our corpus discussed harms related to intersectional identities [41].In this category, we included papers that addressed how the distribution of harms manifests across multiple factors of identity.Attention to intersectional identities in social justice work is important because it often reveals uniquely felt forms of harm.For example, Kirabo et al. explore the unmet transit needs of disabled people in Kampala, Uganda, which do not match the unmet needs of disabled people in countries that are better represented in HCI literature.This mismatch frequently leads accessibility interventions designed in the Global North to fail when implemented in the Global South [100].As another example, Musgrave and colleagues explore the unique form of harassment experienced by Black women online [138].
This imbalance replicates the findings of a systematic review of identity in CHI between 1982-2016, which found that research during this time tended to address a single facet of identity at a time rather than intersectional identities [163] 3 .All 17 papers in our corpus that addressed intersectionality were published after 2018, and 11 of these were published between 2021 and 2022, indicating a growing response to the lack of intersectional analysis observed in 2017.Overall, intersectional identities remain understudied in the context of social justice work.

Who
Benefits.Opposite to those who are harmed on the Axes of Oppression (see Figure 3) are those who are privileged.In our corpus, it was rare for papers to engage with those who benefit 3 Work in HCI relating to intersectionality has been critiqued for failing to cite and recognize Black women researchers as well as its true origins (also discussed in Section 2.2) [155] from social structures.One exception was authors who sought to change the behavior of the privileged and the powerful.For example, Luckoff et al. aimed to promote paternal engagement in family life via technological cueing, addressing labor imbalance by gender in families by increasing the amount of labor performed by men [123].Any paper in our corpus that focused on the behavior change of researchers (via developing better methods, etc.) similarly engaged those who are in positions of power [92,111].Another exception occurred when authors in our corpus spoke of the uneven distribution of harm and benefit between groups.Matias et al. highlighted how male voices receive more attention than females in social media, resulting in silencing women's voices [129].Corbett and Loukissas set up a clear dichotomy between the gentrifiers and the gentrified: the distinction between which is rooted in systems of power that distribute privilege and power among dimensions of class and race [39].While these examples showcase researchers moving towards engaging groups that receive social and material benefits, many questions remain.How do benefits contribute to the perpetuation of social injustice?Can those benefits be distributed more evenly?How can groups that benefit be engaged in working towards social change?More work is needed to understand the role technology can play in engaging the powerful and the privileged in concerns of justice.

4.
3.1 Sources of Harm.70 out of 124 papers included an explicit explanation for why harm is occurring.Authors located the harm within sociotechnical systems (n=19), processes of design, research, and making (n=19), and social and political structures of oppression (n=35).Full descriptions of each category, along with examples of how they appear in our corpus, can be found in Table 3.
The most common explanation of harm was broader social and political systems of oppression (n=35).Felice et al. identified menopause as a matter of social justice due to the "intersecting gender-and agebased marginalization that people going through menopause still experience" [33], and Musgrave et al. explain that "Online harassment relies on underlying hierarchies of power, privilege, and discrimination based on characteristics like sex, race, and gender" [138].These explanations situate the causes of harm not in technology but rather in social and political systems and, ultimately, a histories of oppression.

17
"Black women and femmes" [138], "Low resourced job seekers, specifically those with neither social networks nor digital literacy" [190], "Persons with disabilities from the Global South" [100], "Women in the Global South" [177,178] 19 papers in our corpus identified processes of design, research, and making as a source of harm.In presenting a framework for social justice-oriented interaction design, Dombrowski et al. argued that "explicit engagement with social justice can help guard against" the tendency to "design for the status quo," which "often leads to the re-entrenchment of problematic inequalities and power relations" [54] and Spiel et al. argued that "Technology research for neurodivergent conditions is largely shaped by research aims which privilege neuro-normative outcomes" and characterize ADHD experiences as disruptive [168].This set of papers focuses on the role of researchers and designers in causing harm.
The corpus demonstrates a clear awareness of the links between different explanations of harm.In the 70 papers that gave explanations of harm, we saw thoughtful links between sociotechnical systems, the processes of design, research, and making that shape them, and the broader social and political systems they operate within.Musgrave et al., quoted above arguing that online harassment is rooted in a long history of discrimination in the US, go on to explain that "These historical trajectories are important for contextualizing online harassment not as a contemporary issue caused by technology, but as existing structural inequalities that are boosted and bolstered by technology."We hope to see HCI researchers continue to engage with this complexity, even as they focus their work on addressing harms located at these different sites.When we discuss harm without identifying the source, we absolve ourselves of responsibility for contributing to harm and obscure the possibility of alternative futures.We encourage authors to be diligent about explaining how harm occurs and to be thoughtful in choosing where they want to have an impact.

4.
3.2 Sources of Benefit.15 of the 34 papers that included benefits discussed benefits as produced by technology.For example, virtual reality has created new opportunities to "present, express, and experiment one's identity" [72], assistive technology has increased "access to information and participation on social media," [17] and sales platforms like Etsy have provided opportunities for employment [157].Despite awareness of the potential for technology to cause harm, there remains optimism within the field about the positive potential of technology.
The 19 other papers that included benefits either did not explain the sources of those benefits (10 papers) or the source of the benefit did not clearly fit into the framework we used to organize sources of harm (9 papers).There was a range of benefits that came from various social practices, including street outreach work [58], conversation facilitation tools [87], paternal involvement in heterosexual parenting [123], civic engagement [90], and education [161].More work is needed to unpack sources of benefit and examine the ways in which they may be fundamentally different from sources of harm.It is especially important to understand systems, structures, and processes that allow some to benefit from harm to others, as these systems are key in the continuation of harm.

APPROACHES TO PURSUING JUSTICE
"...people experience and resist oppression on three levels: the level of personal biography; the group or community level of the cultural context created by race, class, and gender; and the systemic level of social institutions.Black feminist thought emphasizes all three levels as sites of domination and as potential sites of resistance." [36] -Patricia Hill Collins, Black Feminist Sociologist We turn to Patricia Hill Collins to understand the levels at which HCI is resisting oppression [36].Collins emphasizes that oppression is not limited to one level but operates at multiple levels simultaneously, and by recognizing how oppression manifests at the individual, community, and systemic levels, we can develop more effective strategies for resistance.Our corpus shows researchers pursuing justice at all three levels, sometimes independently and "As crucial responsibilities are increasingly transferred to computer systems, however, systems of public consequence are "black boxed" by closed-source code, security-by-obscurity policies, outsourcing to private companies, or simply closed door agency processes that exclude public participation and oversight." [191], "AI-powered analysis of faces, bodies, and associated data" [15], "Apps are being imbued with their designers' interests, opinions, biases and assumptions about self-care."[169], "The biometric point of sale (POS) machine in the administration of food security in Indian's public distribution system" [137] "HCI is increasingly working with 'vulnerable' people, yet there is a danger that the label of vulnerability can alienate and stigmatize the people such work aims to support."[75], "Research aims which privilege neuro-normative outcomes." [168], "The incorrect assumption that knowledge produced is applicable to all genders when the data only justifies generalization to one gender group." [142], "Community based participatory research" [37], "researchers working with marginalized communities" [113], "community based research" [111], "Rushing to build and deploy AI systems, without first examining the knowledge, needs, and perceptions of the paraprofessional workers that will be expected to operate these systems within marginalized communities" [144], "Neoliberal design logic" [38]

Social and Political Systems of Oppression
Authors locate the cause of harm in broader systems of power

35
"Structural racism, classism, patriarchy, and other systems of oppression have rendered breastfeeding a luxury good, more easily accessible to privileged families."[86], "Domestic work is visible or invisible in a society due to many cultural, social and conventional factors."[44], "Legacies of structural racial inequity."[42], "Rooted in violent inceptions from the human trafficking and exploitation of Africans and subsequent economic and social inequality and discrimination.These historical trajectories are important for contextualizing online harassment not as a contemporary issue caused by technology, but as existing structural inequalities that are boosted and bolstered by technology." [138] interdependently.Here, we discuss the strategies researchers use to pursue justice at the individual, community, and systemic levels.

Individual
At the individual level, people experience harm due to relationships, individual experiences, and personal beliefs [36].Researchers in our corpus explored how sociotechnical systems can shift individual experiences of harm by influencing interpersonal relationships or individual belief systems.For example, some researchers investigated the role technology can play in shifting personal dynamics in interpersonal relationships.Dhaundiyal examined the distribution of domestic labor in India during the COVID-19 lockdown and prototyped a tool that makes the invisible work done by women more visible to the rest of their families [44], thereby mitigating the harm of recognition.This tool was designed to "initiate discussion among the family members that could break the cycle of continued gender inequalities." Sultana et al. also designed a tool to support women who are victims of online harassment.This tool captures evidence of the harassment and empowers them by providing authenticity and credibility to their claims [177].While underexplored in our corpus, these examples demonstrate the role that technology can play in building awareness about the inequitable distribution of power between individuals.We also saw a very small number of studies (n=3) where researchers used technology to support victims in coping and sense-making in the aftermath of a harmful experience, reducing psychological and emotional harm.To and colleagues used novel prototypes to explore the role ICTs can play in helping BIPOC individuals cope and respond to racist micro-aggressions [182].Similarly, Dimond and colleagues explored how online storytelling can help individuals make sense of a traumatic experience [48].Studies at the individual level represent what Collins refers to as changes to the "individual consciousness," an important site of resistance [36].Approaches targeting the individual level were the least represented, making up less than 10% of our corpus.There is an opportunity for HCI to explore further the ways technology can shift experiences of harm at the individual level by building awareness, supporting reflection, and educating processes that change how we think about ourselves and each other.

Community
Each individual is a member of certain groups based on their factors of identity.Those groups have their own cultures, norms, histories, knowledge, and social institutions.People experience harm at the community level when the group's knowledge, norms, and ways of being are oppressed or controlled [36].Researchers working at this level surfaced how sociotechnical systems can contribute to such oppression or serve as a tool for resistance [40].Approaches targeting the community level were the most represented in our corpus, appearing in over half of the studies.Researchers often sought to understand how a specific technology does not work for a certain community.For example, Bennett and colleagues explored how smart scooters and autonomous delivery robots create obstacles for those with disabilities [14], an autonomy harm, and Cunningham et al. investigated the ways that mobile banking applications inequitably serve Black communities [42], a financial and economic harm.We also found examples where researchers asked a specific community about their experience of harm, healing, and joy online and the design interventions they would like to see [138,201].Additionally, researchers highlighted examples of communities self-organizing and using technology to pursue justice.
In contrast to research exploring technology's role in supporting or oppressing groups of people, this work sought to understand how communities are already adopting and appropriating social technologies to further their own visions of justice.This included understanding how older bloggers are combating ageism online [108] and the way that Venezuelans are using Facebook to barter for basic needs in a troubled economy [61].While these studies were less common in our corpus, they demonstrate a bottom-up approach to doing community-level justice work.Instead of asking what HCI researchers can do for the communities we work with, bottom-up approaches sought to identify and amplify the work that is already being done by these communities.

Systemic
When Patricia Hill Collins discusses oppression at the systemic level, she specifically discusses formal organizations such as educational institutions, which "represent the dominant groups' standpoint and interest" and subjugate others to those interests [36].Fortunately, we saw many examples of researchers problematizing existing knowledge, frameworks, and methods used within HCI.For example, Dourish and Mainwaring critiqued ubiquitous computing, highlighting how colonialism is entwined with "how we think, how we talk, and how we work" [56].Pei and colleagues illustrated how the "digital divide framework" perpetuates inequity by ignoring startup, maintenance, and affective costs accompanying digital access.Researchers also discuss structural challenges in the practice of research and how, for example, community-based research can be extractive and constrained by the funding and timelines of academia [37,113].Researchers working at the systemic level introduced new methods and frameworks that centered the viewpoints of historically marginalized groups.For example, Lazar and colleagues introduced "making" as a method that can center the viewpoints of people with complex communication needs [109] and Chen and colleagues introduced "trauma-informed computing" to better account for the role that trauma plays in people's interactions with technology [29].At the systemic level, a rich amount of critical and generative work is helping us as a community question our epistemologies and reflect on alternatives [8].Next, we discuss specific tools that researchers used to pursue this work.

TOOLS FOR EQUITABLE PROCESSES
"For the master's tools will never dismantle the master's house.They may allow us temporarily to beat him at his own game, but they will never enable us to bring about genuine change." [121] -Audre Lorde, Black feminist lesbian activist and poet Researchers further their social justice goals throughout many stages of their research projects.Throughout our corpus, we observed that researchers' considerations of justice extended beyond a project's topics or outcomes and into the process and methods.The methods researchers chose can be reviewed in Appendix A.4.To capture why they were chosen, we included a code to capture any "commitments" to social justice made through methodological choices.A commitment occurred any time the project's methodology was explicitly bound to a social justice paradigm.For example, a diary study might include a commitment to reflexivity.This code garnered limited quantitative information as most papers did not write about explicit commitments.Those we did collect showcase the many considerations made throughout a project.These commitments are not tied to qualitative or quantitative practices.
The most frequent commitments were direct engagement with the community (n=11) and self-reflection (n=11).The second most common were those having to do with team roles.These commitments included having researchers participate as members of the community of interest (n=6) and taking on participants as collaborators (n=6) (which may serve to reduce power dynamics (n=3) or empower participants (n=3)).We also saw commitments to engaging with complexity in difficult topics (n=2), making space to support participants and researchers with those difficult topics (n=4), and commitments to care (n=2) throughout the process.
In the remainder of this section, we elevate three tools researchers employ to accomplish such commitments to social justice.Though there are countless tools, we chose three which we believe to be a good starting point for researchers interested in making their process more just.We cover how projects within our corpus use reflexivity, directly reward participants, and employ justice-oriented lenses.

Reflexivity: An Individual Level Tool
Since the 1970s, design scholars have turned towards reflexivity to navigate the "fragile encounters" between designers and participants [69].HCI is embedded within larger systems of power and capital, and reflexivity allows designers to reflect on how these systems shape their research and relationships with participants.48% (n=59) of papers in the corpus included reflexivity statements in dedicated sections or disparately throughout the paper.In this section, we discuss the information included in these statements and how it enabled researchers to grapple with their individual roles within larger systems of oppression.
63% of the papers that included reflexivity statements discussed researchers' membership or the (mis)match between authors' and participants' identities.This involved researchers sharing their race, gender identity, sexual orientation, education status, ability, language, nationality, and class.Consistent with prior work, membership alignment afforded authority to the data collection and analysis process [117].Authors often pointed to areas where membership was aligned.For example, in their investigation of "harm, healing, and joy among Black women and femmes on social media, " Musgrave and colleagues explain their decision to have "two Black coauthors conduct the focus groups because of their shared identities" [138].They share that "the Black coauthors bring educational expertise and lived experiences to how they approached the study design and analysis" and that they "prioritized those over perceived norms and expectations of a "typical" CHI paper or of the white-presenting coauthor." When membership was not aligned, authors most often acknowledged that as a limitation.For example, Ismail and Kumar share that "despite our sincerest attempts to understand and portray the perspectives of women from the marginalized contexts where our research is located, we acknowledge that our lived realities are starkly different from those of our participants, and we can at best offer a partial perspective," [91] and Strengers and colleagues share that "due to our own cultural positioning as WEIRD (Western, Educated, Industrialized, Rich, and Democratic) society scholars, the scope of our scenarios biased towards studies and interactions we have experience with through our respective research areas, and should be read with this perspective in mind." [172].In one instance, however, membership misalignment served as an opportunity to continue learning.Hope and colleagues share that because their team was "majority white (4 out of 6 co-organizers), college-educated, cis-gender, and heterosexual", they undertook personal work to educate themselves, including forming an advisory board to serve as mentors and participating in anti-racist training [86].They share that this work was "transformative in helping white members understand and confront the ways racism, oppression, and other forms of white supremacy manifest themselves." This is an example of what Liang and colleagues refer to as "personal work" [117].Liang and colleagues caution that the HCI community should not rely on reflexivity statements and membership as a shortcut for validating (or invalidating) work with marginalized people.Given the intersectional nature of oppression, it's unlikely that participants' and researchers' identities will match completely.Instead, mismatches in researcher membership can serve as opportunities to investigate and understand oppression in its many forms.
41% of the papers that included reflexivity statements discussed researchers' previous experience with the community.Prior work has noted the challenges of gaining access to marginalized communities [111,113], and researchers spoke about the multiple hats they wore to build trust and gain access.This included roles such as researcher/activist, researcher/ally, researcher/community member, researcher/volunteer, and researcher/organizer.Many of the reflexivity statements focused on the community-facing role.For example, Chopra and colleagues describe how one of their coauthors "lives within the neighborhood, engages with Green South activities and is closely involved with the food growing community, giving the researchers exclusive access, reliability, and convenient recruitment of participants, " and Berns and colleagues describe how the first author's involvement as a researcher and a participant "was helpful to gain access to the field.She could, for instance, contact an acquaintance among the community volunteers who suggested some participants for initial interviews" [18].Fewer statements discussed the challenges researchers faced in their academic-facing role.In one paper, Leal and colleagues discuss the challenges related to funding structures, citing practices, traditional metrics of success, timelines, and expectations they face from their academic institutions when doing community-oriented work [113].They see the act of sharing these tensions "as an act of care -an act of critique because we care and because we want to initiate the change from within." While this discussion is largely missing from the papers in our corpus, Leal and colleagues demonstrate that sharing the challenges community-oriented researchers face from their academic institutions can highlight structural barriers within the academic community.
The third most popular reflexive element was the authors' political and ethical stances (observed in 25% of papers with a reflexivity statement).Researchers in our corpus often made strong, explicit commitments.For example, Chopra and colleagues shared that they "align themselves with social justice and environmental citizen-led movements driven through grassroots and feminist perspectives, " [31] and Brewer et al. align themselves "with other design activists who value an explicit orientation to social justice goals" [25].At the same time, participants did not always share these commitments.Tseng et al. reflected that their vision for healthcare aides "may not necessarily map onto those aides may pursue for themselves" [187], and Battega et al. describe how when they presented digital commons alternatives, they received very little interest from participants [19].Chopra et al. made strong commitments to environmental and social justice but were disappointed to see that participants "unwittingly replicate and reproduce some of these more normative ways of imagining food futures, " including by suggesting that "sustainability can be achieved through efficiency gains that limitless technological advances and growth can provide" [31].Okolo et al. felt conflicted because while they "view HCI research from a social justice-oriented design practice, " the community health workers in India with whom they partnered shared sensitive personal data of patients without their consent, raising questions about the role of researchers as allies in culturally unfamiliar contexts [144].While questions about negotiating value tensions between diverse stakeholders are not new [73], social justice work raises new questions about the ethics of making normative political commitments when working with historically marginalized communities.

Direct Support: A Community Level Tool
In human subjects research, researchers are asked to contend with how their research may positively or negatively impact their participants by institutional internal review boards (IRB) [66].Beyond the IRB process, harms and benefits produced by the research process are generally not reported.Traditional HCI research has been described as fundamentally extractive [113], and without attention to how research processes produce benefits for participants, we risk reproducing harm.
In our corpus, we looked at how researchers provided benefits to their community partners through direct support.Benefiting participants was front of mind for some researchers: "the most important component of the initial planning phase was that every interaction we had with residents in the neighborhood had to result in some direct benefit to the residents working with us" [167].Overall, this approach was rare.Ninety-four papers (76 percent of the corpus) involved direct work with participants.Out of this subset, less than half (n=39) discussed direct support that participants received from engaging in research, while the remainder did not discuss any type of support.We highlight this data point not to suggest that this work is not happening but rather that it likely does not make it into our research papers.Here, we draw attention to the many ways researchers are directly supporting participants in the hope that it sparks imaginations and provokes conversation.[187].In one case, we saw researchers pay participants who were teachers but not the students who participated.We recommend researchers consider compensation based on the value participants offer to the research process rather than defaulting to minimum standards.One way to do this, for example, is to offer all participants a living wage.within social movements or with community organizations to help build capacity for political action or growth.This included establishing symbiotic relationships between community organizations [161], building the capacity to better advocate for themselves [191], building resources and skills to lead or manage grant proposal development [111], and more.Whitney and colleagues, for example, supported their community partners to better advocate for themselves: "Our HCI knowledge contributed to this struggle for democratic control over technology, but in ways that went beyond design, user studies, or systems building.By analyzing legal and technical documents, we helped coalition members speculate about the functions of the technology and the intention of company and city that animated it.By creating "Slightly Dystopian" demos, we began with the desire to demonstrate latent harms but ended with new insights into the messy, material operation of the streetlights API that proved politically consequential.Finally, our report forged a tool with which coalition members confronted the claims that the City had assembled a coalition supportive of the streetlights, including business owners, technologists, and environmental activists" [191].• Emotional Benefits -At least 4 five papers reported emotional benefits.Bosley and colleagues organized workshops and activities to support healing justice for their participants.They share that the "practice of Healing Justice gave participants a safe space to discuss the impact of traumas like policing and the ways healing could be integrated into their daily life" [22].• Other Types of Support -Five papers note other types of support.Those include providing participants feedback on how they did on research measures [124], providing access to personal informatics [129], providing food [91,200], and creating digital archives on behalf of the community [176].
Ismail and Kumar provided benefits on a case-by-case basis [91].They note that their participants were "compensated for their time and care when possible and appropriate as a token of gratitude, such as with chai and snacks, sweets, mobile recharge cards, and transportation costs, but their contributions to our research cannot be measured in material terms." Other forms of support that were not included in our corpus but that researchers might consider include sharing the use of physical spaces for meetings, purchasing or sharing access to technology or tools, and creating lasting infrastructure for the community by, for example, writing purchases into grant proposals.

Justice-Oriented Frameworks: A System Level Tool
We collected information on what frameworks authors used and their motivations for choosing them.Frameworks are a valuable tool to decenter dominant ways of thinking and highlight alternative viewpoints.The authors in our corpus used a variety of frameworks to foreground a population, a set of harms, or a specific set of values that may otherwise be marginalized in design work.For example: Rankin and Han use Intersectionality "as a framework for inclusivity [value], positioning Black women gamers [population] as legitimate members [implied harm of recognition] of the gaming community." [154] We highlight this tool because applying a different framework can fundamentally alter a project's process and outcomes.The framing used by Rankin and Han impacts the rest of their research practice, including their study design.Instead of individual interviews or a survey, they organize game events for Black women on campus as a site to gather data.This creates an intentionally inclusive "social atmosphere for gaming."They note that "Because intersectionality invites the inclusion of those who are often silenced, ignored or just invisible, the inclusion of marginalized populations challenges the status quo." This study design reifies Black women's "experiences are relevant within the gaming subculture" [154].
Researchers who used frameworks to foreground certain populations often sought to center voices that have been marginalized or excluded from that space.We most frequently saw feminism applied in our corpus.Other frameworks that were used to foreground specific populations included Intersectionality [41], Black Feminism [180], Afrofuturism [195], crip technoscience [81], and more.
Authors who used frameworks to foreground harmful conditions often highlighted harms that the HCI community has overlooked.Chen et al. use trauma-informed computing as a framework because there is "no cohesive accounting of the role of trauma in people's interactions with technology and what, if anything, those responsible for the design, deployment, and support of digital technologies should do to account for the potential effects of trauma" [29].In another example, when doing research internationally, Corbett and Loukissas share how they use post-development critique and transnational feminist theory to attend to the "power inequities between developers and target communities" [39].
A third set of papers used frameworks to foreground a set of values or commitments.These papers discuss how a specific set of values shapes their own positionality or the approach they take.For example, Hope et al. share that, "our work within the design space of breastfeeding has been guided by the epistemic and emancipatory commitments of feminist HCI, which accounts for situated knowledges and lived experiences and supports innovations that are imbued with sensitivity to the central commitments of feminism-agency, fulfillment, identity and the self, equity, empowerment, diversity, and social justice" [86].Berns and colleagues use a Community Economies framework to embrace "interdependence" and the "commonality of all beings" while also highlighting "marginalized modes of livelihood that co-exist under the umbrella of mainstream economic models" [18].
Researchers can also use multiple frameworks in the same project.For example, Spiel and colleagues use both interdependence and crip technoscience to understand how technology research considers neurodivergence [168].They use interdependence to encourage "researchers to explicitly understand the resulting technological artifacts and implications as political," and crip technoscience as a framework for "centering disabled people as well as making commitments to access, interdependence and disability justice." Collectively, these examples demonstrate how frameworks can help researchers intentionally center marginalized populations, account for overlooked harms, and provide value commitments that can guide the research process.

DISCUSSION
In this review, we map the landscape of social justice research in HCI as a basis for our field to reflect, collaborate, strategize, and take collective action.We present the harms and benefits that practitioners investigate (Section 4), who they work with (Section 4.2), and their reasons for intervening where they do (Section 4.3).We also discuss the social levels where researchers can intervene (Section 5) and the tools they can use to do so (Section 6).In this discussion, we first reflect on how we used these five key considerations to review the field and how HCI researchers can leverage them in future work (7.1).We then encourage the HCI community to attend to what is under-researched (7.2), make concrete changes to better recognize and reward social justice work (7.3), and shift towards pursuing justice rather than responding to injustice (7.4).Moving forward, the HCI community can take these steps to address gaps in existing social justice work and move towards more just futures.

Key Considerations for Social Justice HCI
Within our community, it is not always clear what we, as HCI scholars, mean by justice.We found that researchers used justice terms in passing (see our exclusion criteria, Section 3.1) and that justice was rarely defined explicitly (only 21 of 124 papers detailed what they meant by justice).While we see this lack of standardization as necessary and useful for intellectual work with a concept that encapsulates a wide range of concerns, topics, and lives, we also want to highlight the key considerations that underlie existing social justice research and can provide scaffolding for future work.
We identified five key considerations that can help researchers frame social justice problems and decide how to approach them.These are by no means an exhaustive list of the considerations of justice but rather a set which we found valuable to work with.There are three key considerations that can help researchers frame problems of justice: (1) harms and benefits (2) who is harmed and who benefits (3) sources of harms and benefits The taxonomy of harms in Section 4 offers language for HCI researchers interested in understanding the lived experiences of injustice, the harms and benefits that individuals and communities experience.Looking at who is harmed and who benefits from that harm allows us to zoom out to view systems with multiple actors and to contemplate power differentials within those sociotechnical systems.Identifying sources of harm (such as research practices, sociotechnical systems, and broad systems of oppression) allows us to consider the various mechanics by which harms and benefits are enacted.These three considerations can support HCI researchers to explore and identify harms in their own work, question why and how people accrue benefits, and interrogate the mechanisms of how these harms come to be.Taken together, these considerations offer a more complete picture of conditions of injustice along with the mechanisms by which they persist.
While the first three considerations help researchers frame and map critical issues, the last two can help researchers decide how to address these issues.There are two key considerations that can help researchers determine a path forward: (4) site/s of intervention (5) tool/s for intervention In Section 5, we build on the work of Patricia Hill Collins to describe three sites of intervention.Importantly, one project might take action at all three sites, depending on the tools used.Action can be taken throughout the research process and does not have to be solely represented by the final research outcome.Instead, researchers might work towards justice in the forms of direct support they offer and their everyday interactions with the communities they work with.The set of tools discussed in Section 6 provide examples of ways researchers can enact justice through the mundane choices they make throughout their research processes.By thinking about justice as a process, rather than an end goal, researchers can intervene at multiple sites of action over the course of a single project.
Using an article from our corpus, we now present an example of how these five key considerations take form.We hope to illustrate how they can be used in interpreting existing work and setting out on new work.Spiel et al offer a critique of HCI approaches towards ADHD in their paper "ADHD and Technology Research -Investigated by Neurodivergent Readers" [168] 5 .They share that experiences and behavior of folks with ADHD "are perceived as disruptive to neurotypical standards of behavior," and these individuals are rarely invited to the table to co-construct technology.These are harms of recognition and autonomy since individuals with ADHD are misrepresented and excluded from design and research.The people who are harmed are participants, researchers with ADHD, and potentially the broader public due to biased technology design.Standard HCI research on ADHD benefits the researchers, medical practitioners, and parents, who are all in relative positions of power and are likely to hold and perpetuate dominant ideologies.They may stand to gain long-term career growth (e.g., funding, publication opportunities, etc.) or short-term benefits that stem from power dynamics, such as retaining control or avoiding the discomfort of questioning a non-dominant narrative.The source of the harms and benefits is the type of ADHD research that occurs within academic institutions, and that is also the site of intervention. 6Action on the systemic level raises issues of injustice to the broader world of HCI and reifies the experiences of researchers with ADHD on the communal level.For their tool selection, they perform a critical and reflective literature review as member researchers in the field in order to problematize past work.
While this paper has a strong social justice component to begin with, the five considerations presented here allow us to look deeper into why these harms exist at all and who benefits from them, to recognize the labor that has not been explicitly stated by the authors, and to imagine alternative sites of action to take on in future work.We offer these considerations as a starting place, not as a full description of what it means to do social justice work in HCI.As eluded to in the opening quote of Section 6, new tools and ways of thinking are still needed.We hope that others in our field will continue strategizing and developing tools and techniques that leverage our unique domains.

What's Missing from our Social Justice
Conversations?
While conducting this review, we found areas of social justice work in HCI that appeared underdeveloped.We draw attention to critical gaps in the field that require further attention to move forward an agenda of social justice.First, we found a set of harms that are under-researched.Cultural harms, the destruction or imposition of one culture over another [27], have not been fully explored in HCI research.For example, the cultural harm that resulted from building the Dakota Access Pipeline through the Standing Rock Sioux Tribe Indian Reservation is described by Mike Faith, the chairman of the Standing Rock Sioux Tribe, "Every day the pipeline operates represents a threat to our way of life and an insult to our culture and traditions that have withstood so much."[27].Our field has long recognized that computing reproduces colonial systems of power [56,89,148], has built on postcolonial lenses to understand the cultural impacts of colonialism [89], and has proposed steps for researchers working towards decolonization [1,2].In our corpus, we found that HCI researchers contending with colonialism primarily focused on recognition harms rather than cultural harms.Importantly, colonialism includes the subjugation of knowledge, traditions, languages, and practices [20,62,96], and the impact of colonialism cannot be fully understood without unpacking these cultural harms.Anthropologist Arturo Escobar describes resistance to cultural harms as a form of "ontological struggle" and asserts the importance of addressing cultural harms so that we can move towards a pluriverse, a world where many ways of being, knowing, and relating can co-exist [60].By naming and mapping cultural harms, we can identify places where there are tensions between worlds.
We also found that environmental harms resulting from human interaction with other species and the environment are similarly under-explored in HCI.The negative changes impacting the planet are disproportionately due to human activity, leading to the demise of coral reefs, uncontrolled fires, depletion of animal and plant habitats, and more [164,192].While environmental harms are inevitably wrapped up in other forms of harm like increased illness, psychological and emotional distress, and financial and economic burden that constitute important issues of environmental justice, attending to environmental harms specifically allows us to go beyond the human impacts of climate change to understand the damage to the natural world as harmful in itself.Decentering humans is a crucial step to understanding and contending with the Anthropocene [67,68,118,166], an unofficial name for the current geological era in which human activity plays a significant role in shaping Earth's climate and ecosystems [28,116].By naming and mapping environmental harms, we, as a field, can more effectively advocate for non-human species and ecosystems.To do this work might require perspectives outside HCI, including fields such as environmental sciences, geography, biology, animal studies and more.
Second, the role that intersectionality plays in shaping experiences of harms and benefits has been understudied.Although we did see an increase following calls to attend to intersectionality in HCI [43,155,163,199], few papers engaged with how complex, overlapping identities uniquely contribute to the harms that people experience.The small number of intersectional papers in our corpus could be a result of the standard tools we use as a field.Normative views are often baked into our research design and the tools we use [32,40,92].For instance, single-axis identities are easier to model and generalize since statistical models become more complex when multiple identities are involved.In many studies, those who do not neatly fit into single-axis identities are excluded (this has been voiced by AI ethics scholars [16,146,185] as well as accessibility researchers who have observed that less than 1% of articles discuss participants with multiple disabilities [125]).This exclusion perpetuates inequities and does not recognize the wholeness of participants.In contrast, injustice is brought about by complex interactions between multiple factors of identify as well as sociopolitical contexts.Committing to recognizing and studying intersectional identities may require alternative models and strategies for analysis.Further, it may require new recruitment strategies and bringing together intersectional research teams.
Third, we find that the bulk of our attention as a field has been focused on approaching justice at the communal and systemic levels, while there are fewer interventions at the individual level.This may be because researchers are not necessarily identifying individual beliefs as sources of harm (see Section 4) or because researchers are not reporting individual-level work in their writing.Researchers in our corpus located harms within sociotechnical systems, processes of design and research, and larger social and political structures of oppression but did not explicitly discuss individual beliefs as a source of harm.While systems of oppression undoubtedly influence individual beliefs, individual beliefs and actions also make up systems of oppression and serve as a unique site of intervention.Furthermore, there may be fewer interventions at the individual level because the field of technology and design research may be less oriented toward shifting personal beliefs and behaviors, instead grouping individuals as "users," rather than focusing on complex individual experiences.For example, To and colleagues find that they need to rely on psychology literature when exploring the role that ICTs can play in helping BIPOC individuals cope with racist micro-aggressions [181].By relying on other fields, such as psychology and education, HCI can more effectively build awareness and support reflection and education processes that change how people think about themselves and each other.
Fourth, we see that HCI research currently does not account for these benefits' role in the continuation of social injustice.Injustice (and the harms and oppression that come with it) exist because people within systems of power stand to benefit from it.Without considering who benefits and how they benefit, we risk placing the burden of change on those who are already burdened.We echo previous calls to "study up" [139] -to study institutions, administrations, and networks that "create the preconditions necessary for specific marginalized and peripheral subcultures to emerge in the first place" [6].Anthropologists have documented structural barriers to such work, such as gaining access to elite and powerful institutions [139].In our work, we also faced challenges identifying who benefits and how due to the dearth of language to understand benefits.While Canning and Tombs provide a provisional typology of harm, we found no such typology for benefits [27].A deeper understanding of the kinds of benefits that powerful groups reap due to systems of power and oppression can shed light on how and why these systems continue.

Recognition for Social Justice Work
Researchers in our corpus moved the needle on social justice issues despite operating within unjust systems themselves.As we saw in Section 6, common challenges like negotiating membership (mis)alignment, engaging value tensions with participants, and navigating academic institutions are all forms of invisible labor which are not necessarily rewarded by traditional HCI career progressions [113].This messiness, however, is the day-to-day work of enacting justice.For HCI researchers pursuing social justice work, social justice is not only a research outcome, but can also be a part of their process-much of which goes unseen and unreported.Our field must evolve to make social justice-oriented work sustainable for the researchers who do it.Inspired by Leal and colleagues, we share these tensions "as an act of care -an act of critique because we care and because we want to initiate the change from within" [113].While we do not have the solutions to these critiques, we are hopeful that we can work together to make this research sustainable.
We identified a few opportunities for our research institutions to better support social justice work.First, we hope that HCI can measure research impact not only by the work's contribution to the field, but also its impact on the communities of focus.Reporting direct benefits is not a standard practice in HCI research (as evidence by the dearth of papers reporting on this), but has the potential to provoke conversation about the myriad of ways we can support our community partners.By expanding our field's ideas around research impact and contribution to include direct impact on community members, we can create more spaces for these conversations.
Second, the fast-paced world of technology development does not fit research involving complex, multidimensional social issues.Building equitable research relationships is thorny requiring reflexivity and thoughtfulness.In 2016, Dombrowski et al. recommended the field adopt a commitment to reflexivity and conflict, encouraging designers and researchers to grapple with tensions that arise during the research process [54].Patricia Hill Collins argues that this work is especially challenging because although most individuals have no problem in recognizing their own victimization, "they typically fail to see how their thoughts and actions uphold someone else's subordination" [36].In our corpus, we found that Hope and colleagues undertook personal reflexive work [117] when they ran into complexity due to not being members of the community they were studying [86].Their reflexivity allowed them to confront the systems of power that they themselves were part of.These moments of reflexivity, where researchers grapple with conflict, power dynamics, bias, and more, make up the everyday micro-moments where justice is negotiated and enacted.Ultimately, committing to slow, reflexive, and complex work may result in publishing nothing at all.Not all work is appropriate to report on or pursue because it creates more possibility for harm [9] 7 .We hope to see recognition of this labor and the development of incentive structures that reward enacting and sharing messy and conflict-ridden processes.
Issues of recognition and reward structures are compounded for researchers from historically marginalized groups.These researchers are more likely to do social justice-oriented work, advocacy, and partake in additional emotional labor [94,150], all while also facing institutional barriers and epistemic violence [52,65,95,202].Epistemic violence occurs when contributions made by these marginalized members of the community are silenced, undervalued, erased, and disregarded [202].Epistemic violence and long-term barriers to access have left historically marginalized voices out of our field and its history.For example, Harrington et al. "call out the lack of attention and blatant disregard for Black women's contributions to the design canon," specifically around design speculation and futuring [82].Work remains to be done in taking down the walls of the ivory tower so that we can retain and uplift historically marginalized voices [52].More work still remains in untangling the damage done by a written history and praxis of HCI (and academia more broadly) that has left out, denied access to, and disregarded the labor and worth of historically marginalized thinkers [130,155].
Doing justice-related work as a researcher with a historically marginalized identity is further complicated by fluctuating institutional politics.In section 3.1, we highlight that not all justiceoriented work takes on the label by explicitly discussing social justice.Identifying oneself as a social justice researcher comes with its own risk (as evidenced by recent events [76,83,93,107,133]).This risk impacts who can safely claim justice work in academic writing.Hence, efforts towards social justice is likely to continue under the radar.At the same time, there is something beautiful about our collective efforts happening both under the radar and out in the open.
A commitment to slow work is necessary for social justice contexts, but our inflexible systems of career progress punish researchers who engage in slow, caring work, and the labor it takes to carry it out is often invisible to traditional academic research structures.Without making meaningful changes to our research institutions, we all but guarantee that the technologies we build will perpetuate the harmful systems they are built under [13].

Beyond Responding to Injustice
As our field continues this work, we want to ask what it looks like for HCI work to explicitly pursue justice instead of responding to injustice.We found that much of our corpus focused on a damage-centered view of justice [183], and most papers centered their investigations of justice around the harm people experienced.But what do just worlds look like that go beyond responding to harm?How can we be proactive about building those just worlds?
One way to start answering these questions is by listening to historically marginalized communities and privileging their needs and desires.We must continuously ask people who they are and what they dream of, moving beyond aspects of identity towards humanizing communities and developing a real, whole understanding of what it looks like for those communities to self-actualize [183].The frameworks we highlight in Section 6 offer one tool to center the voices and visions of historically marginalized populations.For example, Black Afrofuturism centers the experiences of Black Americans and provides a tool to speculate futures that could have existed had Black existence been historically celebrated [82].Similarly, crip technoscience critiques medicalized or damagecentered views of disability and instead views 'disability' as a part of one's "self-determined identity" [168].Dreaming is an act of resistance [150], and there are many different ways people have envisioned better futures and articulated their dreams.Part of our work as HCI scholars is to develop the skills and experiences needed to hear and attend to them.
To and Smith et al. highlight that supporting BIPOC flourishing also requires shifting away from design's tradition of solving problems and moving towards actualizing desires [183].A growing group of scholars have already argued that the focus on need-finding and problem-solving rests on assumptions of deficit [136,202] and have called for a shift towards alternative approaches, such as asset-based design [30,47,102,198] or design for human flourishing [183].In the social justice space, the impulse to identify and solve problems allows researchers to react to conditions of injustice but also ultimately reinforces deficit and damage-centered narratives.In authoring this paper, we struggled against replicating these views in our own approach, even as we reported on a body of work that often took this approach to solving problems of injustice.Moving forward, we argue that social justice work must go beyond reacting to problems of injustice to explore, understand, and actualize just futures.To and colleagues reflect that concretely, this shift requires moving the focus of design from meeting baseline needs to identifying and prioritizing desires; "we must seek opportunities for self-actualization" [183].Furthermore, moving away from needfinding and problem-solving calls for a fundamental shift in the role of designers from problem-solvers to "facilitators of design" that work to understand and lend expertise in service of the communities they work with [47].This reframing may challenge our traditional understanding of design; "we must realize that today's problems hail from yesterday's "solutions" [183].Exploring, understanding, and actualizing just futures requires skills and tools to look for, listen, and understand other ways of being in the world, especially when those narratives oppose dominant discourses.Together, we can design the futures we dream of.

CONCLUSION
The field of Human-Computer Interaction (HCI) is envisioning and building more socially just futures, yet there is an opportunity to think more deeply about the horizon we are working towards and how we might take steps towards it.In this paper, we conduct a systematic review of 124 papers explicitly pursuing social justice between 2009 and 2022.Through this review, we present a landscape of social justice research in HCI, including the harms and benefits addressed, the approaches researchers use to pursue social justice, and tools that can support thoughtful and equitable research processes.Collectively, these key considerations serve as a useful tool for the HCI community to think about justice-related questions and concerns.We conclude with a reflection on steps the HCI community can take to address gaps in existing social justice work and move towards more just futures.reading papers outside our corpus and those in it.To those who are doing the work and are not represented in the corpus: we see you.Keep it up.We hope readers move forward with care for themselves and the communities they work with and within.
This work was supported in part by the National Science Foundation Graduate Research Fellowship Program under Grant No. DGE-2039655.Any opinions, findings, conclusions, or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

A CORPUS AT A GLANCE A.1 Publication Venues
The first instances of social justice-oriented papers in our corpus originated from CHI in 2009 (See Figure 2).We see the topic begin to gain momentum after 2016, following a foundational publication from Dombrowski et al. introducing "social justice-oriented interaction design." Publication numbers also seem to follow larger social, cultural, and political conditions.There is evident growth following 2020 which aligns with the Black Lives Matter protests and the COVID-19 pandemic.Similarly, a period of growth happened in 2017 and 2018, which may relate to the #MeToo movement.When we conducted our search, CSCW publications had not been released yet for 2022.Overall, CHI has the most contributions (n=92) to our corpus, followed by DIS (n=23) and CSCW (n=8).UbiCOMP has only one contribution in our corpus [56].It follows that the low incidence of papers from UbiCOMP might be explained by the conference's focus on devices and technology development.However, ubiquitous computing revolves around computing in our everyday lives -it seems that this area should be a great fit for social justiceoriented research.As discussed in the next sections, we saw a low incidence of papers that engaged in technology development and or within the domain of emerging technologies.

A.2 Contribution Types
Wobbrock and Kientz define HCI contribution types as empirical, artifact, methodological, theoretical, dataset, survey, or opinion [193].We coded for these contribution types directly.The most common research contribution type in our corpus was empirical work (See Figure 4).Of the papers included in our corpus, 83 made empirical contributions, 15 contributed artifacts, 13 made theoretical contributions, 11 made methodological contributions, 6 were survey studies, 4 were opinion pieces, and 2 were dataset contributions.The glaring difference between empirical contributions and others (5x more) highlights the exploratory landscape of the corpus, rather than oriented towards defining (theory) or actualizing (artifact, dataset).This exploration could represent either that the topic is still emerging within the field (following the design approach of exploring, defining, and then designing in sequence), or that empirical work is somehow better suited to social justice at this junction.

A.3 Domains
Half of our corpus is represented by the civic technology and social computing domains (See Figure 5).There were considerably fewer paper that had to do with emerging technologies such as VR (1) [72], IoT (2) [10,71], or AI & ML (9) [17,115,200].It is surprising that these domains are not explicitly discussing social justice concerns given that these emerging technologies are reshaping the social structures we live in today.

A.4 Methodologies Used
The most employed data collection strategies were interviews (n=53), research through design (n=23), and workshops (n=22) (see Figure 6).Many projects employed a mix of methods (n=51).Very few included gathering data from device usage like biometric data (n=2) [112], logfiles (n=4) [129], or even web scraping (n=3) [108].One standout (though still limited) category was those who created and ran community events (n=5) as sites for data collection and observations (n=19).We highlight these papers for their efforts to organize community within the research context.For example, Rankin and Han's game nights (further discussed in Section 6.3), and Strohmayer et al's collaboration in organizing a public activist march on International Day to End Violence Against Sex Workers.

Figure 2 :
Figure 2: Increasing Number of Papers Using the Term "Social Justice" in HCI Publication Venues from 2009-2023.There were no papers before 2009 found.

Figure 3 :
Figure 3: The intersecting Axes of Oppression as adapted by[158] from Morgan's Axes of Oppression[135]

Figure 4 :
Figure 4: Contribution Types in Our Corpus

Figure 5 :
Figure 5: Domain spaces represented in our corpus

Table 1 :
[27]ples of each form of harm found in our corpus.Definitions are based on Canning and Tombs' typology[27]

Table 2 :
Examples of populations or groups of people that were identified as experiencing harm in our corpus

Table 3 :
Examples of how authors in our corpus explained why the harm was occurring

Table 4 :
A full collection of papers included in our corpus.

Table 4 :
A full collection of papers included in our corpus.

Table 4 :
A full collection of papers included in our corpus.A vehicle for research: Using street sweepers to explore the landscape of environmental community action Aoki, Paul M.; Honicky, R. J.; Mainwaring, Alan; Myers, Chris; Paulos, Eric; Subramanian, Sushmita; Woodruff, Allison

Table 4 :
A full collection of papers included in our corpus.

Table 4 :
A full collection of papers included in our corpus."Genderfluid" or "Attack Helicopter": Responsible HCI research practice with non-binary gender variation in online communities

Table 4 :
A full collection of papers included in our corpus.Strangers at the gate: Gaining access, building rapport, and co-constructing community-based research