What is femicide? The United Nations and the measurement of progress in complex epistemic systems

Femicide is a key global indicator of progress towards gender equality. The occurrence of some but not all five gender dimensions in the indicators of violence used to measure progress towards United Nations Sustainable Development Goals 5, 11 and 16 are analysed as resulting from the tension between divergent feminist strategies that focus either on women-only or on mainstreaming intersecting inequalities. The tension between universalist and particularist projects underlies the contestations over the construction of these gendered indicators. The analysis develops a conceptualisation of indicators as assets in order to capture the social relations of power involved (rather than as boundary objects), supported by platforms (which can be public as well as corporate) and generated by dynamic epistemic systems (rather than stable epistemological infrastructures).


Introduction
What is femicide? Femicide has become an important global indicator of gender inequality and violence. The United Nations (UN) is a key location of the measurement of progress, most recently in the Sustainable Development Goals (SDGs). Global indicators of progress have emerged as significant forms of power relations during processes of digitalisation.
The conceptualisation and measurement of femicide are shaped by tensions between strategies for progress prioritising universalism or particularism. One tension concerns whether femicide should be distinguished from homicide and another whether feminist strategies are centred on women or on mainstreaming intersecting inequalities. The negotiated outcome of these tensions is found in the various ways the gender dimensions of violence and femicide are conceptualised and measured. A further tension is between strategies of progress centred on a model of modernity driven by economic growth and on a broader model that includes multiple inequalities, concepts of justice and forms of violence. The UN is a platform on which indicators of progress undergo contested development. These contestations include attempts to insert gender inequality and violence into globally dominant practices to measure and encode progress.
The increase in digitalisation highlights concerns as to whether indicators and associated quantitative procedures deepen and entrench power inequalities or support science and justice by speaking truth to power. This analysis of indicators builds on concepts of 'boundary object' and 'epistemological infrastructure' to develop a new approach that analyses indicators as 'assets' rather than 'objects', and on complex 'systems', rather than foundational 'structures'.
This article analyses the presence of gender, violence and femicide in the development of UN indicators of progress. It shows how, despite the misgivings of some feminist theorists, ending gender inequality and violence were mainstreamed into the UN SDG indicators as part of a broad coalition seeking to widen the concepts of development beyond economic growth. It offers a new approach to indicators that use the concepts of data 'assets' to capture the social relations of power and 'epistemic systems' to facilitate analysis of the dynamic nature of the processes.
The article reviews feminist strategies to end violence against women, the gender dimensions of femicide, the nature of the UN, concepts of progress and the relationship of numbers to society. It offers an empirical analysis of the development of the UN SDGs on gender, violence and femicide. It assesses the implications of the empirical investigation for theories of gender, violence, and data in society, with a focus on femicide. It offers a mainstreaming of feminist approaches to femicide, thereby re-visioning the universal.

Universalism and particularism
While the tension between universalism and particularism in the pursuit of development and justice is a classic debate, the inclusion of gender inequality and violence is more recent. This tension is found in feminist debates on how to include the gendering of violence in indicators of development established by the UN.
It has been argued that 'Only a universalist conception of humanity can provide the political and philosophical basis on which to build a struggle for equality' (Malik, 1996: 8). This approach is found in the Universal Declaration of Human Rights (UN, 1948).
The universalist approach is challenged on the grounds that it insufficiently addresses the rights of specific groups and multiple versions of modernity. The particularist approach seeks to make specific rights more visible as, for example, in the Convention on the Elimination of Discrimination against Women (UN, 1979). The tension is articulated in debates on identity politics, mainstreaming and intersectionality (Crenshaw, 1991;Walby, 2005;Walby et al., 2012).

Gender equality strategies and multiple gender dimensions
The tension between gender equality strategies that are centred on women (particularist) and on mainstreaming intersecting inequalities (universalist) shapes the way gender is included in the concept and measurement of femicide. The particularist strategy focuses on women, whose human rights are violated by violence (Bunch, 1995). The universalist strategy mainstreams gender and intersecting inequalities into the concept and measurement of violence. The strategies share the attempt to make gender visible in the analysis of violence (Strid et al., 2013) but differ in how this is done. Five dimensions of 'gender' are relevant to violence: sex of the victim, sex of the perpetrator, the relationship between perpetrator and victim (domestic (intimate partner, other family member), acquaintance or stranger), sexual aspect (e.g. rape) and gender motivation (e.g. dowry death) (Walby et al., 2017).
There is debate as to whether the sex of the victim is sufficient to identify a killing as femicide, or whether other gender dimensions are needed, especially but not only, gendered motivation (Corradi et al., 2016: 981). ' [T]here is an open debate that can be divided neatly into two positions: whether the term femicide should encompass all murders of women or, alternatively, be restricted only to some' (Mujica and Tuesta, 2014: 5). The main axis of the debate is whether a female victim is sufficient to identify a killing as femicide (Dawson, 2016) or whether a gendered motivation is essential, killing women 'because they are women' (Radford and Russell, 1992). Further issues include whether a 'domestic' relationship between perpetrator and victim counts as a gender dimension (Stöckl et al., 2013) or not (Johnson, 2015).
Some work on femicide does not clearly identify which dimensions of gender are necessary or sufficient. For example, Fregoso and Bejarano (2010) do not clearly distinguish between the various gender dimensions in their discussion of feminicide in Latin America. Even Radford and Russell (1992: 157), often treated as a reference point for defining femicide in terms of gender motivation (Corradi et al., 2016: 978), refers to femicide as both 'the misogynous killing of women by men' and, more ambiguously, as part of 'the overall oppression of women in a patriarchal society'. The UN Special Rapporteur on Violence against Women variously refers to femicide as a form of violence against women (the title of her post), as the gender-related killing of women (para. 25), of which 'intimate partner or family-related homicide of women' is an example (para. 31), and as the gender-motivated killing of women (para. 26) (UN General Assembly, 2016). This fluidity may have the advantage of increasing debate and impact (Corradi et al., 2016) or the disadvantage of lack of consensus on a single, and thus more powerful, indicator.
There are hybrid positions that combine principles and pragmatism. One of these seeks data on femicide concerning gender motivation but pragmatically uses data on the sex of the victim if only that is available. Dawson (2016) states 'femicide is defined as the killing of women by men, consistent with the definition used by most international quantitative data because such acts are the most obvious and easy to document for prevalence and comparative purposes' (p. 997). She draws this conclusion after noting the 'recognized difficulties in establishing motive . . .and understanding of what would be 'femicidal' offences. . .coupled with lack of consensus on gender indicators'. Dawson and Carrigan (2020: 1-2) consider the 'utility and accessibility of sex/gender-related motives and indicators' and conclude that while they may be desirable, data are too poor to make them useful: 'many SGRMIs [sex/gender-related motives and indicators] do clearly distinguish femicide from other types of homicide, but current available data do not consistently allow for their documentation'. Weil, while preferring to define femicide by gender motivation, pragmatically, when mobilising existing statistics, uses data that defines femicide by the sex of the victim only. In Standish and Weil (2021: 6), femicide is defined as 'the intentional murder of women because they are women' and in Weil and Keshet (2021: 40) 'an extreme form of gendered violence, is the killing of women and/ or girls because they are perceived as female'; however, when mobilising statistics for the analysis of the killing of old women (over 60 years of age), pragmatically uses the sex of the victim (Weil and Keshet, 2021: 46). Similarly, Campbell and Runyan (1998) state that 'femicide refers to all killings of women, regardless of motive or perpetrator status' (p. 348) while noting that discovering whether some femicides are hate crimes is important for developing prevention policies though reference to motive should be avoided unless there is clear data.
The perpetration of violence by a domestic relation, either an intimate partner or other family member, is treated by many writers as constituting gender-related violence even when simultaneously utilising a definition of femicide based on the sex of the victim. For example, Stout (1992) equates femicide to the killing of women by male intimate partners (although does not define it explicitly). However, Johnson (2015) considers that a focus on the domestic relationship alone amounts to 'de-gendering'. Dawson and Gartner (1998) note that 'the relationship between a victim and offender is critical to understanding the context and dynamics of homicide' (p. 378). Some combine an analysis of femicide, defined in terms of the sex of the victim, with a focus on a subset where the perpetrator has a domestic relationship with the victim, thereby treating domestic or intimate partner homicide as a subset of femicide. For example, McFarlane et al. (1999) define femicide as the killing of women while focusing their work on those committed by intimate partners, which they call: 'intimate partner homicide' (p. 300). For Taylor and Jasinski (2011), 'the term femicide refers to the killing of a woman generally'; and they examine the sub-set that are intimate partner perpetrated femicides (p. 341). For Faqir (2001: 65) 'honour killings' are 'a particular type of intrafamily femicide', thereby combining gender motivation, 'honour killings', with a definition of femicide as the killing of any woman. For Messerschmidt (2017: 70), there are 'two types of femicide: intimate partner femicide and so-called 'honor femicides', thereby using the term femicide to indicate the killing of women, while focusing on subsets related to intimate partners and honour. Some researchers analysing the multiple intersecting inequalities that shape the distribution of the killing of women use data on the sex of the victim to define femicide and data on the intersecting inequality to define the subset of women affected. These include Weil and Keshet (2021) on gender and age.
Further issues concern exclusions and indirect harms (Smit et al., 2013). The deaths of some women, especially minoritized women, may not be included as a homicide but attributed to other causes (Dawson and Carrigan, 2020). The accumulation of harms due to gender inequalities that lead indirectly to unnecessary death (Galtung, 1996) is sometimes counted as femicide (Walklate et al., 2020).
The literature contains multiple definitions (Corradi et al., 2016;Weil et al., 2018) and references contexts that are 'complicated, historical and context contingent' (Abraham and Tastsoglou, 2016: 517). Most researchers combine principle and pragmatism in which the sex of the victim is sufficient to define a killing as femicide while treating data on the additional four gender dimensions as desirable but not essential.
The definition of femicide developed in theory is necessarily revised and clarified when confronted with the practicalities of empirical data. 'Indicators' are the place where theory and data collide and the UN SDGs are key global indicators.

Progress and development
There is tension between strategies of progress centred on economic growth and broader conceptions of development. Modernisation theory understands progress to be driven by economic growth that transforms the rest of society in progressive directions. This is contested by the conceptions of modernities as multiple and conceptions of progress that recognise multiple inequalities, such as gender and ethnicity, and dimensions of life, including violence as well as economy (Robeyns, 2003;Sen, 2004;Walby, 2009Walby, , 2012. The contested development of the UN's indicators of progress and development is an example (Blaustein et al., 2018). Transnational feminism has been active in UN developments (Moghadam, 2005).

Digitalisation and complex epistemic systems
Digitalisation is a process in which numbers come to have increased significance in organising social processes. Its implications have been considered variously progress, because of the enhanced capacity of science to speak truth to power, or detrimental, because it restructures control over knowledge towards power elites (Porter, 1995). There is a gender dimension to this debate, which addresses whether datafication is detrimental to women (Buss, 2015;Merry, 2011Merry, , 2016 or makes visible forms of gender inequality that facilitate challenges to inequalities (Cohen et al., 2011).
'Indicators' are a special instance of a number (Berger-Schmitt and Jankowitsch, 1999). Indicators summarise complicated information and encode it in institutions, with consequences for processes ranging from public policy to prestige to value extraction. The investigation of their development requires the conceptualisation of numbers as social relationships (Day et al., 2014;Lury, 2020). The negotiation of the construction of indicators can be participatory and interactive (Marres and de Rijcke, 2020) or highly contested.
Indicators and data have been conceptualised as 'objects' (Star, 1999;Star and Griesemer, 1989) in an attempt to better insert them into the ontology for social science analysis. Key indicators are 'boundary objects' with relevance across more than one field of power and practice. However, there is a limitation to the concept of object because of the rigidity and restricted ontological depth of the connotations of the term object. An indicator is not a neutral technical matter but an entity with consequences and causes.
An alternative concept of indicators emerging in the literature is that of 'asset' (Birch and Muniesa, 2020). This parallels the theory of money by Keynes (1936): While some treat money as an object that is neutral in relation to power, Keynes demonstrates that money is always a social relationship, entailing relations of power, obligations and entitlements. The extension of the vocabulary of property to non-tangible social relations, such as brands by Lury (2004) and legal entitlements to identity by Cooper (2018), are further parallels. Selecting and mobilising one data summary rather than another as an indicator has implications for the social groups and projects affected by the field of power in which the indicator is located. Conceptualising indicators as assets rather than objects invokes the social relations of property and power rather than technical neutrality. Whether indicators are sufficiently important in its effects that they are better thought of as a form of property, entailing social relations parallel to those of ownership or nonownership of property, than an 'object' is investigated below.
An indicator requires a platform to become an asset. A platform constitutes the assemblage of power that enables this particular number to be selected and mobilised as an indicator and have effects on the distribution of power. The concept of platformisation captures the process through which some forms of data become more important than others through elevation by powerful organisations in a digitalised society (Lury, 2020). A potential limitation of the concept of platform is that it is too narrow to capture the complexity, while its simplification is simultaneously an advantage in communicating the issues at stake. The definition of the concept of platform could be revised to be wider than its earlier formulations, which focused on new technology in for-profit settings (Gillespie, 2010;Plantin et al., 2018;Poell et al., 2019) to include not-for-profit entities including public bodies. This is investigated in the context of the UN.
The development of indicators takes place in a wider system of knowledge, or epistemic system. This has been previously conceptualised as an 'epistemological infrastructure' (Star, 1999;Star and Griesemer, 1989) in which a range of institutions support the development of key numbers; and has been used and developed to address the development of databases (Nakazora, 2016). A limitation of the term 'infrastructure' is the connotation of stability associated with the concept of structure. If the elevation of key numbers is a fluid and contested process, then the connotations of stability within the concept of infrastructure are disadvantageous. An alternative conceptual vocabulary derived from complex systems thinking (Kauffman, 1993) has a greater capacity to grasp the fluidity and mix of cooperation and contestation of complex adaptive systems. Complex systems can also address the way an indicator may not be at the 'boundary' between entities but actually part of each of the multiple entities involved in its construction. While ostensibly it is the same thing, actually, within each of these diverse entities it is, logically, different. It is better to consider an indicator as an 'asset' not an 'object' at a 'boundary' as if it were on the edge of different entities or somehow between them; but to note its joint constitution by multiple systems, within each of which it has meaning, even different meanings. Systems concepts aid the development of this thinking.
The analysis of global indicators on gender and violence by the UN builds on and goes beyond existing concepts concerning indicators. The article investigates whether 'object' should be replaced by 'asset', whether 'epistemological infrastructure' replaced by 'epistemic system' and the UN considered a 'platform'.

Method
The article investigates the way dimensions of gender and violence have been included during the development of the UN SDGs. The main components in the empirical analysis concern the SDGs relevant to gender and violence and the history of indicators in the UN.
Data concerning the nature of the SDGs and the history of UN development of indicators of progress are available on UN websites. The analysis uses analytical categories derived from the review of the literature which generated a fivefold classification of the gender dimensions of violence and homicide and a twofold classification of feminist strategy (Walby et al., 2017).
The analysis identifies the gender dimensions of SDGs indicators concerning violence; analyses the history of the development of UN indicators of progress; and tests the relevance and applicability of proposals to develop concepts of asset and epistemic system.

Gender, violence and femicide in UN indicators of progress
The UN created a set of goals, targets and indicators of progress in the SDGs, which include reducing and ending gender inequality and violence. The variations in the way gender and violence are included in indicators for Goals are investigated here. The development of the UN as a platform creating indicators as assets is investigated in the context of its wider environment.

Sustainable development goals
The UN has 17 SDGs (UN, 2021). The intersection of gender and violence is found in three Goals: Goal 5 gender equality; Goal 11 safe cities; and Goal 16 peace and justice. The relevant goals, targets and indicators are identified: Goal 5 concerns gender equality, 'Achieve gender equality and empower all women and girls' and has several targets, of which Target 5.2 is to 'Eliminate all forms of violence against all women and girls in the public and private spheres, including trafficking and sexual and other types of exploitation', and measured by two indicators: 5.2.1 'Proportion of ever-partnered women and girls aged 15 years and older subjected to physical, sexual or psychological violence by a current or former intimate partner in the previous 12 months, by form of violence and by age', and 5.2.2 'Proportion of women and girls aged 15 years and older subjected to sexual violence by persons other than an intimate partner in the previous 12 months, by age and place of occurrence'.
Goal 11 concerns safe cities 'Make cities inclusive, safe, resilient and sustainable'. Target 11.7: is to 'provide universal access to safe, inclusive and accessible, green and public spaces, in particular for women and children, older persons and persons with disabilities'. Indicator 11.7.2: 'Proportion of person victim of physical or sexual harassment, by sex, age, disability status and place of occurrence, in the previous 12 months'.
Goal 16 concerns peace and justice, 'Promote peaceful and inclusive societies for sustainable development, provide access to justice for all and build effective, accountable and inclusive institutions at all levels', with target 16.1 to 'Significantly reduce all forms of violence and related death rates everywhere', measured by indicators, 16.

The development of the UN as a platform for indicators of progress
The UN is an international institution with multiple legal instruments and agencies. The gender equality projects within the UN have changing locations and diverse links to transnational feminist projects. The UN system is a 'platform' on which key indicators of progress are given a privileged position in the world. When the UN endorses a particular indicator, other entities adapt, even when they are not legally obliged to. States, acting within an international community, appear to take pride in their relative positioning in the hierarchies created by the SDGs, while national and international non-governmental organisations (NGOs) use them to legitimate access to financial resources and to promote preferred policy options. The UN has ridden the wave of digitalisation.
The UN is an institution legally based on Treaties signed by most states. It has multiple legal instruments, including Conventions (e.g. Convention on the Elimination of Discrimination against Women) and Declarations (e.g. Universal Declaration of Human Rights); multiple agencies, including United Nations Development Programme (UNDP), World Health Organisation (WHO), UN Women, UN Statistics Commission and UNODC; multiple forums and councils, including the General Assembly of all Member States, Security Council and Human Rights Council; ad hoc conferences (e.g. Beijing Conference on gender equality); and privileged positions for some post-holders (e.g. UN Secretary-General, UN Special Rapporteurs).
The gender equality project in the UN developed two major feminist strategies -a focus on women and on mainstreaming an intersectional coalition. The UN gender equality projects, including political participation, education for girls, economic participation and ending violence against women (UN Women), have been achieved with the support of transnational feminist civil society, NGOs, academics who contribute to ad hoc expert groups and some states (Bunch, 1995;Moghadam, 2005).
From an early 'universalism' that purported to include women, for example, the Declaration of Human Rights (UN, 1948), there has been increasing and successful pressure to make gender equality more visible in the UN legal instruments, for example the Convention on the Elimination of Discrimination against Women (CEDAW) (UN, 1979), CEDAW Recommendation 19 on gender based violence (UN, 1992), and the 'Palermo' Protocol to Prevent, Suppress and Punish Trafficking in Persons, especially Women and Children (UN, 2000), in UN General Assembly Resolutions, for example, the Declaration on the Elimination of Violence against Women (UN General Assembly, 1993), special conferences, for example the Beijing Platform (UN, 1995), the work of specific agencies, for example that of the World Health Organisation on violence against women (WHO, 2005), reports from the UN Secretary-General on gender-based violence (UN Secretary-General, 2006), the appointment and work of the UN Special Rapporteur on Violence against Women (Ertürk and Purkayastha, 2012; United Nations (UN) Special Rapporteur on Violence against Women, 2008), and the attention of the UN Statistical Commission to indicators for violence against women (UN Statistical Commission, 2010). The UN Special Rapporteur has called for femicide to be more seriously addressed (Domazetoska et al., 2014;UN General Assembly, 2016), using her report to the General Assembly to call for the development of femicide observatories.
The process of the development of indicators to measure gendered violence was both cumulative and contested. While the gender equality projects generally supported the inclusion of indicators on gendered violence, they did not agree on what they should be. Some documents included multiple positions simultaneously (e.g. Beijing Platform, UN, 1995). The controversies can be seen in the disparate positions taken in papers by academics and stakeholders presented to the Expert Group Meetings called by the UN agencies to settle these issues, such as UNDAW (UN Women, 2007). These represent two strategies: a human rights-led focus on violence as a violation of women's human rights that concerned violence against women only and a CEDAW-led focus on violence as a form of discrimination against women that called for gender disaggregation of all statistics.
When the UN was initially formed, it endorsed no quantified indicators of progress. For example, the Declaration of Human Rights did not include numbers. However, adjacent global institutions (World Bank, International Monetary Fund) adopted Gross Domestic Product (GDP) as an indicator of progress. The goal of economic growth, measured by GDP, assumed hegemonic status in many UN member states.
The use of GDP as the lead indicator was successfully contested in a series of steps. The contestation from radical social scientists (Sen, 2004) and the Global South became institutionalised in the UNDP (1990), which offered multiple additional indices or indicators of progress, including longevity, education and literacy. The first set of UNDP indicators did not include gender, but, following feminist pressure, later sets included separate gender inequality indices (Fukuda-Parr, 2003;Robeyns, 2003). The Millennium Development Goals drew on this work and incorporated a wider range of indicators of progress than economic growth (UNDP, 2015). They included gender equality; but did not include violence, despite calls for this. The later SDGs included reducing and ending violence as well as gender inequality (UN, 2021).

The UN and digitalisation
The power of the UN to set global standards on key statistics has been increasing. This is linked to changes in the epistemic systems associated with globalisation, digitalisation, IT capacity, communications systems and global connections between institutions of knowledge including universities. The UN system operates in a wider environment that structures its capacities and capabilities, constituted by states and other polities, global civil society including international NGOs, a global economy structured by changing forms of capital accumulation and regional hegemons, systems of violence including inter-state war and inter-personal crime, and regimes of inequality including gender, ethnicity and class. This global environment is dynamic, unstable and crisis-prone. It is better understood as a complex system than as an 'infrastructure'. In a period of globalisation, the UN has increased some of its powers but not others.

Discussion
The development of the UN measurement of progress in the SDGs is addressed in relation to multiple dimensions of gender relations and the UN as a platform for the development of indicators. The multiple gender dimensions illuminate debate on the nature of gender. The development of gender equality strategy and practice, in the UN, illuminates debate about the tension between separatist and mainstreaming intersectional strategies. The development of indicators to measure development and justice in the UN illuminates debates on the tension between universalism and particularism and debates on processes of digitalisation.

Multiple gender dimensions in gender equality strategy and practice
Multiple gender dimensions of violence are included in the SDG indicators. They address the sex of the victim (sometimes women only, sometimes compared), gendered relationship (intimate partnership), and the sexual dimension. Some indicators of violence in the SDGs have no gender dimension. In no indicator is there inclusion of the sex of the perpetrator, domestic relations other than intimate partners, or gender motivation. There is no gender disaggregation in some indicators of violence (16.1.3, 16.1.4, and 16.2.1). The gender concepts, where used, include women and girl victims only (5); sex of the victim (male and female) (11, 16); perpetrator-victim intimate partnership where victim is female (5); sexual violence and harassment (5,11,16).
There is no 'One UN' on issues of gender. Multiple approaches coexist in the SDGs, even within the same Goal. The four most frequent approaches are no gender dimension; female victims only; sex disaggregation of victims and intimate partnership. Goal 5, which is the focus of UN Women, is focused on women and girls only. Goal 11 has sex disaggregation of the victim. Goal 6, which has three different approaches to the gender dimension has multiple UN agencies responsible for reporting on its indicators: UNODC, WHO, OHCHR and UNICEF.
The indicators on gendered violence included in the SDGs represent two different strategies: some measure only violence against women, following a human rights' focus on violence as a violation of women's human rights (Goal 5), and some provide gender disaggregation of priority forms of violence, following a CEDAW-led focus on violence as a form of gender-based discrimination (Goals 11 and 16). The tension between the strategies is unresolved.
There is tension between feminist strategies to create a greater political focus on violence against women in that one focuses on women while the other mainstreams intersecting inequalities. The utilisation of different concepts of gender and violence is linked to the differences between these strategies: in one, women are the political subject and in the other, an intersectional, mainstreaming approach to gender equality is linked to a wider justice project. This tension between strategies is articulated in different definitions of femicide: the term may mean only the gender-motivated killing of women or it may be wider, including all homicides where the victims are female.
The identity strategy with its focus on women has been narrowed further by an ideational approach to gender relations that leads to motivation being a sufficient signifier of gender/patriarchal relations and hence to femicide including only those women killed when there was a gendered motivation. When the focus is on ending violence against women only, the preferred measurement of violence against women is presented as if there is no need for a comparator with men.
The intersectional/mainstreaming strategy has a focus on gender inequality and gender disaggregation of all data on violence and including homicide, with gender defined across a range of dimensions beyond the ideational to include sex of the victim, sex of perpetrator, gendered relationship (domestic: intimate partner or other family member, acquaintance/wider kin, stranger), sexual component (rape), as well as motivation. The sex of the victim is sufficient to identify a homicide as femicide (though more dimensions are important for analysis) and aids intersectional analysis, since it offers a simple empirical point of reference that can be combined with other inequalities.
The tension between the two strategies can generate a dynamism that is productive of change, even if the result might appear somewhat messy. The unresolved tension between feminist strategies leads to the diversity of gender dimensions included in the SDG indicators. The selection of one gender dimension or another is a consequence of the specific balance between universalism and particularism, between feminist strategies, found in different situations.
Femicide and other forms of violence have five gendered dimensions: the sex of the victim, the sex of the perpetrator, whether the perpetrator is a domestic relation (intimate partner or other family member) of the victim, whether there is a sexual dimension (e.g. rape) and gender-motivation. The definition based on the sex of the victim is the simplest, most widely used, has the most available data, and facilitates the building of intersectional coalitions. A definition based solely on gender motivation makes it hard to identify cases reliably and results in fewer cases being counted. A definition based on the notion that violence can be 'gender-related' could potentially include any, some, or all of these five gender dimensions. For the purposes of a global indicator, the sex of the victim is sufficient for the definition, although information on the other four dimensions is desirable.
The SDGs vary in the way in which violence, including homicide, is gendered, if at all. They include no gender; women only; sex of the victim; domestic (intimate partner) relationship between perpetrator and victim; sexual (sexual violence, sexual harassment). UN Women 'owns' Goal 5 which is focused on women only. Multiple UN agencies contribute to Goals 11 and 16, which vary in whether they are gender-neutral or gender-disaggregated. This means that the UN overall does not have a consistent gendering practice for its SDGs. The UN indicators of violence with their varied gendering represent the messy reality of the contesting and cooperating entities in this system of knowledge which are differently balanced in different locations within the UN system.

The development of global indicators
The strategies to achieve gender equality vary as to whether they focus on women as the political subject or engage in wider coalitions with intersecting projects. The gender equality project in the UN uses both separatist and mainstreaming practices. While some components of the gender equality project in the UN are centralised in UN Women, other components are dispersed across other agencies. In each agency, there is a different balance of projects supporting universalism or particularism. This has implications for the SDGs which are rooted in different UN agencies.
The early gender architecture of the UN involved several small units dispersed across several entities. This was restructured in 2009 to form a central entity, UN Women, that drew together gender specialists previously dispersed across UNDAW, Secretary-General's Office and INSTRAW. Yet, there remain gender units in most large UN entities. Jointly, this UN gender machinery has become influential for gender equality globally. The tension between separatist and mainstreaming intersectional strategies of the gender equality project has been productive of innovation in the gender equality project despite the inconsistency in gender dimensions included in indicators.
The UN develops globally relevant indicators of progress. The early attempt at universalism was revised to deliver greater attention to groups and projects that felt neglected by that particular construction of the universal. This is represented in the movement towards multiple indicators of progress in the MDGs (which included gender equality) and in the SDGs (which include both gender equality and violence).
When the UN was first established, quantitative indicators of progress were rarely used. However, economic growth represented by GDP (Gross Domestic Product) became an authoritative measure of progress promoted by the World Bank and International Monetary Fund. The contestation of this indicator of progress from the left and the Global South, led within the UN by the UNDP, led to a wider range of indicators in the Millennium Development Goals, which included gender equality, and the SDGs, which included violence and gender inequality. The successful challenge by UNDP to the World Bank and IMF may be understood as a conflict along a South versus North axis; but was also a conflict along a left versus right axis in political economy, in which academics employed in the North as well as South aligned with the South-facing UNDP. There are some competing platforms, but they are not as important. The Global Burden of Disease contests the WHO, while Walk Free contests UN definitions of trafficking in human beings. There are differences within the UN, which might potentially weaken the claim that it constitutes a platform. Goal 5 and UN Women gender violence by a focus on women; while Goals 11 and 16 include both de-gendered categories and those that are gender disaggregated (the UNODC (2015) uses both approaches).
The UN is an increasingly important platform, through which global standards for key statistics on progress are authoritatively established. During the same period in which the UN became a platform for global indicators, the gender equality project in the UN increased in organisational capacity. Indicators developed from those proposed by the UNDP to the MDGs (which included gender) to the SDGs which included gender, violence and sustainability. The feminist project in the UN has ridden the wave of the digitalisation of knowledge and ensured that ending gender equality and violence are included in the SDGs.

Conclusion
A different universalism is possible. A universalism that takes notice of gender and violence, not only of economically led progress. This article offers a rethinking of universalism through the lens of femicide that is more encompassing and therefore less particular. It situates gender equality strategy in the tension between a focus on women and the mainstreaming of an intersectional perspective. This provides the framework in which to understand why the five gender dimensions to measure violence are unevenly represented in the UN indicators of progress. The theorisation of the development of these gendered indicators supports the re-conceptualisation of data as an asset, rather than a boundary object, in an ecology of epistemic systems rather than epistemological infrastructure. The UN is conceptualised as a platform constructed by systems of power that contest and cooperate to shape the priorities for global development.
There is tension between universalism and particularism in the development of UN indicators of progress. There is tension between strategies of progress that are centred on economic growth and on broader conceptions of development that include reducing gender inequality and violence. The feminist project has challenged universalism on the basis that this is merely a particularism that inscribes a patriarchal practice.
The reasons for the selection of various gender dimensions in defining femicide/ homicide and more generally violence in the SDGs are related to the tension between universalism and particularism, between feminist strategies that are mainstreaming and coalition oriented (universalist) and those that are more focused on women only (particularist).
The approach most productive of change is that of gender disaggregation of data on all forms of violence. Measuring outcomes for women only or narrowing the definition to gender-motivated killing of women, in the particularist approach, reduces the capacity to build the alliances that facilitate movement towards a hegemonic position that includes ending violence and gender inequality. By contrast, gender disaggregation, including the simplest and widest definition of femicide as the killing of women, provides greater capacity for mainstreaming intersecting inequalities, coalition building, moving from the margin to the centre and to creating a new hegemony that includes these concerns within a reworked understanding of universalism.
All five gender dimensions of homicide are relevant to including femicide in global indicators of progress. The simplest and easiest to combine with other data is that of the sex of the victim. Further gender dimensions are the sex of the perpetrator, the relationship between perpetrator and victim (intimate partner or other family member as compared with acquaintance or stranger), sexual aspect, and gender motivation. The best single measure of the gender dimension of femicide is the sex of the victim.
Despite ambivalence towards or critique of quantification as a mechanism of coordination, feminism has been successful in getting gender and violence included in the emergent quantification of progress by the UN platform. The development of indicators as assets in a dynamic epistemic system is not a matter of simple progress or regress but rather a multifaceted restructuring of power. On this occasion, the gender equality and anti-violence projects have successfully inserted themselves into the measures developed by the UN, thereby contributing to the reshaping of the global agenda by participating in an authoritative claim to measure progress.
Indicators became increasingly important in this fluid and changing epistemic system. The conceptualisation of these processes needs to grasp their dynamic nature, which is better achieved through the concept of complex systems than infrastructure. Indicators do not sit at the boundaries between entities but rather are constructed so that they have meaning in each of the relevant contesting system. The same term (indicator) may represent different concepts (meaning) in these diverse systems.
The UN is a platform in an ecology of complex epistemic systems. Contesting and cooperating entities drive the selection of some numbers rather than others as indicators, which are assets in multiple epistemic systems. The indicators encode the social relations in these assets into global governance systems.
This analysis concludes that SDG indicators are better conceptualised as assets, than as boundary objects, because of the social relations of power through which they are constituted and have effects. The UN can be conceptualised as the platform that elevates these assets to global significance. The indicators are assets in a fluid ecology of multiple contesting and cooperating epistemic systems.
This reconceptualization builds on and goes beyond the earlier conceptualisation of 'boundary objects' and 'epistemological infrastructure'. The rather static notion of infrastructure is replaced by the more dynamic concept of an 'ecology of complex systems' that better recognises the topological complexity involved. The digitalisation of knowledge leads to changes in the social institutions most privileged within this ecology of epistemic systems, facilitating the development of key platforms. This is a move from the concept of infrastructure to that of complex systems. While epistemological infrastructure has been a productive concept, it under-signifies the complexity and instability of the processes involved. Complex systems thinking facilitates a more productive engagement with the fluid, dynamic and conflictual processes involved.
The concept of platform is helpful in conceptualising the way that the UN has the capacity to elevate and prioritise its preferred set of indicators. The simplicity of the concept is both a strength and a weakness. It elegantly summarises the privileging involved with a concept that is well understood. Potentially it underestimates the complexity involved. But that simplicity is simultaneously its strength; and for this reason, is adopted.
The indicators are conceptualised as assets rather than objects. This is to engage the connotations of the social relations of the power of property in the concept of asset, rather the technocratic neutrality implied by the concept of object.