Representing and Intervening in Swedish Education—Mediating and Adjudicating by Grading Numbers

Educational reforms are attempts to transform and reconstitute conceptions of knowledge, the practice of teaching and the forms of subjectivity associated with being a teacher, pupil, and parent. We will investigate a particular aspect of the extensive changes that have swept across the landscape of Swedish education over the last decades. We will show how the grading system has come to fulfill a mediating role between progressivist educational ideas and neo&#8208;liberal notions of individual choice and entrepreneurship, by way of adjudicating and mediating educational performance for individual pupils, schools, and the education system at large. The aggregated output of the grading system is used for instrumental purposes such as performance management, quality management, and bench marking in the school system. Grading and its associated practices extend throughout the entire school system, defining success and failure from individual classrooms to the national level as the dominant form of representing school performance. The grading system was initially intended to reshape the inner workings of education in Sweden but subsequently became the &#8220;gold standard&#8221; deployed in various managerial practices. Our analysis demonstrate the performative powers when seemingly innocent techniques are put to use in public sector reforms and new worlds are made and unmade by complex constellations of seemingly mundane techniques and ideational frameworks.


INTRODUCTION
There often appears to be a struggle in modern education systems given the function of the state schooling as: "the instrument and object of a technically formed, institutionally organized, and circumstantially driven governmental expertise, whose social distribution was inescapably limited and will remain so" (Hunter, 1996, p. 155), in light of the persistent critique of education, when run by large bureaucratic organizations. The state school systems of the real world, created through decades of piecemeal problem solving, are criticized as "instrumental," "technocratic," or "managerialist" for not realizing full emancipation (Hunter, 1996, p. 146). The foundation of this critique is to be found in the heritage of the Enlightenment and its commitment to the idea of "the pupil as sovereign subject who learns through freedom" and for whom the school system (should) function as a tool for emancipation. The two conceptions of education are, however, connected; state school systems routinely use practices, such as testing and grading etc., to constitute and realize the notion of the subject as pupil, although this has generally been in a more technocratic fashion than that desired by those who see education as a means of liberating humanity.
In this article, we will investigate an aspect of the extensive changes seen in Swedish education over the last few decades. Among a multifaceted and turbulent wave of reforms, we identify the grading system introduced in the early 1990s to be of particular interest. We show how the grading system has come to fulfill a mediating role between progressivist educational ideas and neoliberal notions of individual choice and entrepreneurship, by way of adjudicating educational success for individual pupils, schools, and the education system at large. The aggregated output of the grading system is used for highly instrumental purposes such as performance management, quality management, and benchmarking in the school system. Grading and its associated practices extend throughout the entire school system, defining success and failure from individual classrooms to the national level as the dominant form of representing school performance.
Rather than articulating a critique of instrumentality as such, we use our analysis of the Swedish grading system to show how technologies of representing and intervening might combine and intertwine different rationalities and ideals. Departing from the Governmentality tradition, we analyze how certain rationalities are imposed on a social domain through the establishment of symbolic and material borders, by establishing criteria of relevance and value and instituting certain forms of agency. However, such changes are not necessarily coherent. For instance, as our study shows, a neoliberal constellation of ideas and practices can be made to cohabit with arrangements and rules underpinned by an entirely different set of principles. We begin by discussing representing and intervening as the modus operandi of formal organizing in modernity. It is from that perspective we then discuss the framework suggested by Miller and Power (2013). It offers a more detailed conceptual apparatus for analyzing representing and intervening: the construction of calculable spaces ("territorializing"), the linking of actors, arenas, and goals ("mediating"), creating criteria of success and failure ("adjudicating"), and instituting prefabricated forms of agency ("subjectivizing"). We contend that the model specified by Miller and Power (2013) is relevant far beyond the study of accounting practices. It offers a promising means of studying all practices of representing and intervening in formal organizations, which operate via the imposition of equivalence.
Three empirical sections follow on from this, based mostly on secondary sources, including previous research, public committee reports, and public debate, although we also report some survey data. 1 The first empirical section outlines the initial reconstitution of the Swedish education system, in which education became a commodity through the "free school" reforms while, at the same time, a comprehensive set of reforms attuned to pedagogical progressivism was introduced. A section then follows on subsequent developments, in which the output of the grading system became the gold standard of performance management in the new school system, and became understood as a solid and indisputable measure of educational success, from the individual to the national level. In the third empirical section, we discuss the effects of the reforms on the actors involved. The reforms have in various ways invited, persuaded, and incentivized headmasters, teachers, parents and pupils to assume certain forms of agency. The different categories of actors face different types of challenges, risks, and possibilities related to the effects of the grading system. We conclude by summarizing our findings regarding the reconstitution and framing of the Swedish education system, and the role played by the grading system in that process. After that, we reflect on some of the limitations of our approach and consequently, what our study can and cannot tell us. Finally, we outline some tentative thoughts on how the seemingly irreconcilable logics of market liberalism and pedagogic progressivism have been institutionalized in the Swedish education system.

REPRESENTING AND INTERVENING IN FORMAL ORGANIZATIONS
Formal organizing is predicated on all forms of knowledge and practice, including seemingly insignificant and innocuous ones, which allow us to create representations of organized activities and to design interventions according to instrumental ideas and frameworks (cf. Hacking, 1990;Rose, 1999). In order to become susceptible to being organized according to instrumentally conceived interventions, which is to be organized must be constituted, that is, stabilized as cognitive objects and naturalized as true and self-sustaining. Modern forms of representing and intervening tend to follow a logic that selectively objectifies the world into dimensions, qualities, and logics that are not inherent in what is being represented (Kallinikos, 1995, pp. 118-119). All forms of representing are underpinned by a taxinomia, a conceptual basis for classification (Townley, 1995, 560 f.). Such orders can be, and often are, refined into a mathesis; modes of formally calculating differences between objects defined according to a taxinomia (Townley, 1995). Representing according to this worldview is far from a passive mediation of the world according to its previously understood proportions and its often overwhelming complexity. Further, representing according to this worldview is predicated on an instrumental approach to the world. The very act of representing is intimately connected to a will to master and manipulate to the extent that the world-as-represented is one that is amenable to systematic and controlled forms of intervening. Previous research has analyzed the means by which certain objects, processes, and qualities are constituted as real and amenable to action (cf. Dambrin, Lambert, & Sponem, 2007;Edenius & Hasselbladh, 2002;Revellino & Mouritsen, 2015).
The different modes of representing and intervening typical to modernity bring about presupposed objects, processes, and qualities, whether they are models of markets ("consumer markets"), notions of new forms of impairments ('the psychopathic manager'), qualities of manufacturing ("innovativeness"), or modes of national governance ("transparency"). The act of representing is directed toward pre-existing entities (bodies, production facilities, an area recognized as a country) but reshapes them according to new priorities, visibilities, and forms of knowledge. Technologies of representing cannot accomplish such remarkable effects by their mere existence (cf. Ahrens & Chapman, 2007). In the same way that rituals bring certain, sometimes vaguely, preconceived realities into being (Kunda,1992, p. 92-94), social practices of representing and intervening breathe life into new realities only when they are used, oftentimes supported by hierarchical power relations and symbolic systems of dominance. Reforms are a particular means of articulating strong value-laden statements about the higher aims of society (justice, prosperity, democracy etc.), enacted through plans for accomplishing the intended good. Reforms invariably have to pass through several steps of instrumentation to become manageable forms of intervening in the social domain (Hasselbladh & Bejerot, 2016). Educational reform is no exception, as shown by our empirical case study. Accordingly, the schemes for reinventing education in Sweden in the 1990s became effectual through extensive legislation, policies, rules, and new forms of governance.
The comprehensive framework by Miller and Power (2013) conceptualizes key aspects of practices of representing and intervening. Their "functions of accounting" are not unique to accounting as a form of representing and intervening, but accounting has become almost a masterclass in those endeavors in modern society (cf. Townley, 1995, p. 562). Miller and Power (2013, p. 557) summarize the performative roles of accounting as four functions: (a) accounting constructs calculable spaces ("territorializes"), (b) it links actors, arenas, and goals to each other ("mediates"), (c) it sets criteria of success and failure ("adjudicates") and (d) it offers or imposes distinctive, prefabricated forms of agency ("subjectivizes"). 2 The first function, to "territorialize," implies that social boundaries are drawn to define how calculations can be made with respect to costs and revenues (or any other calculable entity). These boundaries can be tangible and spatially located but also highly abstract, such as those pertaining to a "business area." The second function, to Electronic copy available at: https://ssrn.com/abstract=3601128 "mediate," refers to how accounting principles are used to transcend and homogenize differences in time, space, and qualities. What is represented is also exchanged into a new currency; the categories are used for representing. Everything represented in this manner can be moved through time and space, aggregated, compared, and evaluated. The third function, to "adjudicate," follows from the combination of normativity with calculation. Something is calculated, by simple counting or via complex, derived secondary and tertiary measures and quotas, and subsequently evaluated according to a specified norm. The normativity is the "magic" that attributes relevance to scores, quotas, or simple numbers. The possibility of calculating and combining according to a common currency (not necessarily monetary) incites a particular form of normativity, that is, more is better than less. The fourth function, to "subjectivize," relates to the ways in which accounting both presupposes and brings about a certain type of self. This is not to say that the foregoing social technologies homogenize subjects according to one predefined set of qualities, but rather that they presuppose that those operating the technology adopt or develop a generalized disposition to act and think in certain ways; to be aggressively expansive, to monitor risks carefully, or to regard the adjacent unit as a business partner, etc.
The four different functions of accounting are intimately connected and dependent on each other in different ways.
For instance, to adjudicate requires that a domain is instituted in which a certain mode of calculation can take place.
A mode of calculation requires actors to display a particular form of instrumental orientation to enact calculation as a practice, but unless the rights and obligations to calculate are specified, the same actors are unlikely either to find meaning in or be functionally committed to the practice of calculation. The causal effects of the combined functions can, thus, appear in the form that Deleuze describes as an "imminent cause"; a cause copresent and coextensive with its effects (Raffsnoe, Mennicken, & Miller, 2017). The first three functions are all based on selectively objectified representations (Kallinikos, 1995). To "territorialize" by means of creating boundaries relies on the existence of a measure that separates the inside from the outside; in accounting terms this refers to costs and revenues. "Mediating" across time and space by transforming the raw qualities of the social and material world relies on the existence of measures that allow this to be done. To exercise judgment and make decisions by "adjudicating" implies that normativity is anchored to calculable measures providing a singular dimension of evaluating advantages and disadvantages.
Compared to the above three dimensions, the fourth differs in type. While the first three functions all are predicated on selectively objectified representations, the formation of subjects, even under the harshest and most rigorously technocratic regime, does not adhere to a strict logic of difference and equivalence. The functions that Miller and Power have attributed to accounting are, thus, not entirely predicated on representing and intervening according to taxinomia and mathesis. The positioning of subjects into formal positions in organizations is irrevocably associated with meaning. The formation of subject positions in formal organizations proceeds by means of delineating expectations, specifying tasks, ascribing status, and setting priorities. Even though such work-related meanings are fairly transparent, they cannot be anchored in material arrangements, neither can they be summarized in a measure, quota, or sum.
Attempts to do this via the written word often lead to the finding that this is an imperfect means that can pervert the intention to solidify meaning. Furthermore, meanings cannot be expected to remain consistent across time and space.
It is no coincidence that research in accounting can testify to considerable homogenization of proper accounting practices, as related to the first three functions, while the way such partly fixed practices affect the actors occupying the positions linked to accounting systems are much more open ended.
Representing and intervening for instrumental purposes are in many ways the hallmark of modernity, despite being inevitably shaped by previous attempts at dividing, classifying, and measuring and the constitution of forms of agency.
Thus, the educational reforms we analyze must be understood as a form of reterritorialization, as the invention of new forms of adjudicating and mediating, and as a reconstitution of teacher, pupil, and parent subjectivity within a societal institution consciously molded by governments for more than 150 years. The changes presented here unfold as the reversal of a set of deeply embedded values and practices, linked to government control of schooling, educational success as an individual achievement and education as a common good, by reforms that set out to profoundly transform these patterns of thinking and doing. It will be shown that the deeply fractured education system born out of these reforms has been stabilized by the use of aggregated measures of grading, on the local as well as the national level, to mediate and adjudicate for managerial purposes.

UNFULFILLED HOPES OF EDUCATIONAL REFORM IN SWEDEN
The neoliberal constellation of ideas and practices associated with New Public Management (NPM) has often been pinpointed as the source of educational reforms in Sweden (as well as in other countries) (Ball, 2003;Arreman-Erixon & Holm, 2011). This is also a central conclusion in a recent influential official review of Swedish education reforms (SOU, 2016). Our analyses, however, present a somewhat more complex picture of the Swedish reforms rather than seeing these only as another case of neoliberal reform with their attendant audits, quasimarkets, and emphasis on measurable outcomes. While the later stages of the education reforms relied on well-known features of NPM-such as the mistrust of professions, the primacy of measurable outcomes, and conceptualizing social relations in terms of accountability and control (cf. Clarke & Newman, 1997), the initial reforms were quite different. 3 The content (curriculum and pedagogy), the espoused aims and prerequisites for learning, the view of teachers and students etc., were largely inspired by progressive pedagogy (cf. Morrison, 1989;Zhao, 2011, pp. 252-254). This general term for ideas concerns the development of children and teenagers underpinned by Enlightenment ideals, as articulated, for example, by the American pragmatist John Dewey.
Sweden saw the introduction of comprehensive schools, similarly to most other industrialized countries, in the mid-20th century (Antikainen, 2006). In a series of reforms, the existing school system that divided pupils according to achievement at several points in their educational careers, were refashioned into inclusive primary and secondary schooling for all pupils up to the age of 16. The reform act was preceded by almost two decades of intense debate and small-scale trials, increasing consensus on the general direction of reform until it was finally enshrined in law in 1962.
The new unitary school system was soon broadly accepted, perhaps because it managed to reconcile ambitions to level out social differences by providing equal opportunities to all children with a system of "soft streaming" of pupils in the last 3 years of compulsory schooling. The reform was, however, supported by those with far-reaching ambitions, many of whom saw it as merely a stepping stone toward subsequent reform to shift the emphasis further, from educational achievement to equality and democracy (Husén, 1989;Marklund & Bergendahl, 1979).
The comprehensive school and its compromises, frictions, and failures spurred an even more aggressive critique of the educational system for not being inclusive enough, for letting the weakest pupils down, and most importantly for failing to deliver on the promise of individual and collective emancipation (SOU, 1974, p. 64 ff.), a critique that became more prominent from the 1970s onwards. It was claimed that the promise of the great postwar education reforms remained unfulfilled, in that social mobility had not increased noticeably, teaching was still traditionalist, and the school system had not become the vehicle for social and political change it was intended to be (cf. Carlgren, 2009). The critique was inspired by developments in pedagogy worldwide, channeled through political and professional networks, and was almost a blueprint of the predicament of modern school systems as identified by Hunter (1996). The Swedish school system, the result of compromises and pragmatic statecraft, was criticized for not fulfilling the promises of the Enlightenment and, at the same time, functioning as a heartless instrument of the capitalist system by reproducing social inequality (Larsson, 2011, p. 113).
The progressivist critique became intertwined with one that was more recent, emerging in the 1980s, against excessive centralization of the school system and against a lack of influence and choice for parents and pupils. The different streams of critique became unified in the early 1990s and guided a plethora of reforms, the aftershocks of which are still being felt in Swedish society. The key components of the sweeping reforms of the first half of the 1990s were new curricula for primary and secondary education (Swedish National Agency for Education, 1994a,b), transferring the employment of teachers and headteachers, who were previously government employees, to local councils, and the abolition of the National Board of Education (Proposition 1990(Proposition /91:18, 1990). In addition, the Free School reforms (Proposition 1991(Proposition /92:95, 1991 allowed new providers to run schools and allowed parents and children to choose schools. These reforms were intended to alter the previous course of Swedish education (Carlgren & Kallós, 1997) and were generally decisive compared to practices in other countries. A large part of the reforms were directly related to education (concerning curricula, teacher education, and grading), and were the logical fulfillment of the introduction of the comprehensive school system in 1962 in the eyes of the reformers; with a 30-year delay, the idea of an inclusive Electronic copy available at: https://ssrn.com/abstract=3601128 education system was brought into existence. The effects of this reform were considerable and to this day continue to affect almost every aspect of Swedish education. Some of the new freedoms were, however, curtailed from the late 1990s when the government introduced auditing, benchmarking, and other mechanisms of centralized control in a manner typical of NPM-inspired governance.

Reterritorializing the Swedish school system
The educational reforms of the early 1990s were underpinned by two contrasting ways of constituting and framing education; first an ideal of the knowledge-seeking pupil from pedagogical progressivism, and second a market liberal conception of education as a commodity. These two differ in terms of what they select and objectify from education and schools and with respect to the guiding instrumental orientation-what is to be altered, by what means, and to what ends. Not only have these intertwined, but very different, ways of reterritorializing Swedish education affected policies and pedagogical language, they have also shaped the "rules of the game" in Swedish education, from the national to the classroom level (Wennström, 2016). The first approach to changing the Swedish education system was the previously mentioned progressive pedagogy, which in this case inspired an attempt to devise a totally inclusive school system, and ultimately to realize the unfulfilled promises of the 1962 reforms by introducing more distinctly progressivist curricula. The second approach was the neoliberal inspired Free School reform of 1993, which addressed the lack of freedom of choice for parents and pupils in large-scale mass education systems. This allowed parents and students free choice of schools, and allowed private entrepreneurs to establish schools freely and operate them on a commercial basis (Wiborg, 2013). Education was to be funded by local councils, arranged by public or private providers, and selected and "consumed" by parents and students (Björklund, Clark, Edin, Fredriksson, & Krueger, 2005). 4 Shortly before this substantial package of reforms was launched in the early 1990s, the responsibility for education was handed over from central government to local councils. This decentralization dismantled the previous tightly controlled regime overseen by the National Board of Education, which specified curricula and teaching methods, etc.
Local councils were given extensive freedoms, even to the point of designing their own local curricula and assessment criteria for grading. The reforms were implemented throughout the national education system, without any evaluation or other analysis of small-scale pilot reforms (SOU, 2014, p. 34 ff.). A few years later, a new leap into the unknown was taken. Without small-scale tests or systematic exploration of experiences from other countries, commercial providers were allowed to establish schools, all forms of differentiated teaching were abolished, children with learning difficulties were placed in ordinary classes, and finally a grading system was introduced underpinned by the assumption that all children could reach a relatively high level in all subjects. By 2014, 26% of upper secondary pupils and 13% of primary and lower secondary pupils attended free schools, most of which were owned by large private corporations. In the main cities and in some more affluent areas, the proportion of pupils attending free schools is substantially higher.
Establishment of new schools by private providers was to be funded by local councils, but the reforms left them with little or no ability to influence, far less to control, the free schools established in their areas.
Our analysis focuses on one of the major changes: the 1994 introduction of a new student grading system in primary and secondary schools (Wikström, 2005(Wikström, , 2007. The decades-old norm-referenced assessment system (where grades from 1 to 5 were allotted according to a normal distribution in statistical terms) was replaced by a criterion-referenced grading system measuring absolute levels of performance in every subject. 5 While the previous system had, in principle, been disconnected from students' absolute levels of performance, the new system was predicated on the notion of absolute performance, that is, in relation to stipulated assessment criteria. The cornerstone of the 1994 system is the notion of "goals to be attained." The pass mark introduced was roughly equivalent to grade 3 (of 5) under the previous system. 6 It is explicitly stated that it is the responsibility of the school to ensure that in every subject, every pupil should be given the opportunity to achieve a "pass"; (Swedish National School Agency, 1994a,b, p. 10 f.).
The new grading system was predicated on one of the fundamental assumptions of progressive pedagogy: all pupils can and want to learn. If they cannot or do not learn, the problem is located in the pupil's family or social situation, or the teaching context. It was stipulated that differences in aptitudes for learning different subjects did not exist; we can all successfully learn any subject equally well if the conditions and circumstances are right, at least to a certain level.
Time and commitment are crucial; gifts and aptitudes are irrelevant. The idea of equality, operationalized as the goal that all pupils can, and should, reach a pass grade in every subject, has since been in the basis of all evaluations and inspections undertaken by the National Schools Agency and the National Schools Inspectorate (SI) (cf . Labaree, 2005, see also Lundahl, Arreman, Lundström, & Rönnberg, 2010): The inspection operates with zero tolerance: All pupils are supposed to be given prerequisites to reach the minimum grade of Pass in all subjects. This means that SI criticizes practically all schools where not all the pupils are reaching these goals (Lindgren, 2015, p. 73).
While pupils had previously been able to "get away with" a low grade in one or more subjects, every pupil now had to perform comparatively well in all subjects in order to be promoted to the next class or to achieve an upper secondary school certificate. If a lower secondary school pupil fails in one subject, he or she is not eligible for upper secondary school, and similarly, a secondary school pupil who fails one subject is not eligible for university. There are no exemptions; all pupils are supposed to be able to pass in each of around 15 secondary school subjects. The new assessment system for grading stipulates an absolute difference between passing and not passing. Even a pupil who fails only one subject is, thus, deemed an educational failure. While the grades "passed" and "not passed" could not be assigned numeric values, these discrete terms nevertheless lent themselves to calculation and aggregation, something that would turn out to have considerable implications. On the micro level, the risk of not passing invited pupils and parents to develop counterstrategies within the new landscape of the freedom to choose schools.
The reforms have firmly reterritorialized the school system in two different ways. First, the free school reform granted new rights to local councils, school entrepreneurs, pupils and parents, and second, the grading system redefined educational success from a contingent outcome to an absolute right. By instituting a new, absolute measure of success combined with the option for pupils to leave their current school, the distribution of risks and opportunities among the involved actors has changed considerably. Parents and pupils now have more of a voice in terms of their right to choose any school they wish. Local councils, schools, and teachers have, thus, become exposed to new demands and risks; there is a strong temptation to try to avoid taking on pupils with a propensity to fail (although no pupil can in principle be denied access), and teachers are pressured by demands from headteachers, pupils, and parents. The totality of changes (school choice, new curricula, new assessment system etc.) made the education system a domain riven with contradictory forces and considerable uncertainties (Carlgren & Kallós, 1997). Within a few years, however, after experiencing a period of turmoil, the national governance of education was reinstated step-by-step (Rönnberg, 2011;Segerholm, 2009). One of the social technologies introduced during the great reform period-the grading system and its output in terms of pass and not pass grades-would then assume a quite different role; it would soon become the main tool for performance management in education (SOU, 2014, p. 21).

Adjudicating success and mediating performance across time and space
The lack of guidelines, evaluation, and structure that followed the reforms received increasingly harsh criticism in the late 1990s (Larsson, 2011;SOU, 2014). Education had become a social domain where progressivist pedagogy and neoliberal ideas were uneasy partners. The critique did not, however, pinpoint these new principles specifically, it only found a conspicuous absence of measures for comparing schools and pupil performance over time and space. The government, and not least various national auditing offices, criticized local councils for mismanaging the schools for which they were now responsible. Eventually, the new National School Agency, established in 1992, was restructured into a more traditional national education agency in 2003, responsible for audit and oversight of the national school system. The criteria for awarding particular grades were increasingly set at national level and the previous system of almost complete devolution of assessment has, therefore, gradually changed back to a system of national control.
This recentralization served to streamline assessment criteria, and made quality management mandatory (systematic improvements based on "process statistics"), resulting in the introduction of national tests and the imposition of strict demands for the documentation of pupils' test performance. It did not, however, change the inherent logic of the grading system.
The use of the grading system for adjudicating purposes has propelled the development of auditing practices throughout the Swedish school system, from the most remote individual primary school up to the national level. All schools, including private schools, deliver examination performance statistics to the local council and to the National Schools Agency. Differences in the pass rates between individual schools, local councils, larger areas, different social groups, ethnic Swedes versus immigrants etc., have been continuously measured, statistically analyzed, problematized, and fed into national and local governance and policy processes for the last 20 years. School performance measures have become a crucial part of local school governance, where schools are sometimes punished for not meeting targets or given extra resources to raise levels of performance. Since 1993, The Swedish National Agency for Education (2011) has made statistics comparing local councils publicly available. There are numerous different websites, tools, and reports available for comparing schools, local councils, and the whole nation over time with respect to the proportion of pupils who achieve different grades.
From the perspective pursued in this study, the new grading system offered new possibilities for representing educational success. The grading system, which was initially the offspring of a victorious progressivist reform, has gradually become the linchpin of performance management for schools, at the local as well as the national level. It now serves as the principal social technology for judging between good and poor performance, at any level of the education system. The proportion of pupils failing to reach the pass grade in a subject became the primary indicator of performance throughout the education system and is so to this day. The adjudicating function is tightly linked to the mediating function of a "core measure," considered to reflect something of the utmost importance that it is possible to measure, aggregate, and compare, from the level of the individual pupil to the national level. When different problems and frictions in education became publicly attributed to the reforms, the purist romanticist policies were soon sidelined by technocratic managers who introduced the same techniques for rationalized control as were commonplace in the Swedish public sector at large in the 1990s such as performance management, standardization, and quality management. In the end, reformers, dreaming of a bright future of equality and freedom, had unwittingly provided a reborn educational technocracy with a "gold standard" accountability measure for different levels of aggregation in the school system.
At the national level, aggregated examination results whirl back and forth in the political debate. Despite the emphasis on the goal of every pupil passing, recent statistics show that every fourth pupil still fails to achieve a pass in one or more subjects when they leave the ninth grade (at 16 years of age). Teachers, pupils, headteachers, parents, local councils, the private education corporations, national agencies, and central government, all participate in a never-ending cycle of exercising control and being controlled and questioned, a process understood as an exercise in accountability.
The grading of the individual pupil, aggregated in test scores, quotas, and trends, has come to define performance in the education system. Grading, translated into scores at different levels, has become a powerful tool for mediating across levels and contexts. The risk of failure and the accompanying strategies for avoiding it create new calculable spaces for pupils, parents, schools, local councils, and the nation. The functions of adjudicating and mediating afforded by the grading system operate at separate levels, in social transactions in the classroom, at the local school, and at the national level. Calculable spaces have been instituted wherein the individual pupil, the school, local councils, or the entire nation, are meticulously monitored in order to anticipate or address poor results according to the unquestioned norm that "all must pass." Grading permeates the school system at separate levels and in different forms; in social transactions in the classroom, at the local school and at the national level, in aggregated and disaggregated forms. This seems to have had certain effects on those who operate and are operated by the assessment machine.

New forms of agency
The most significant effects of the reterritorialization of Swedish education and the ensuing adjudicating and mediating functions of grading as a gold standard of performance is to be found in the forms of agency that the new system invites teachers, pupils, parents, school administrators, and politicians to assume. The constituting effects differ, however, for different actors. Parents and pupils have been represented as carriers of certain interests, rights, and responsibilities.
The right to move school, at any time, and the incessant monitoring of school performance, created affordances for parents and pupils to use "exit" or "voice" to influence results, and ultimately to use their exit to avoid bad results.
Teachers and headteachers, on the other hand, adapted to demands from the same pupils and parents, something that is evident in grade inflation (cf. Tyrefors, Hinnerich, & Vlachos, 2017 There are considerable secondary effects of the grading system and its particular embedding in a school system of "free choice." The right of pupils and parents to move from one school to another, coupled with the freedom of "educational entrepreneurs" to establish schools, has created intense competition between schools, particularly in larger cities. In our survey (method and material presented in note 1), 23% of teachers report that competition between schools affects their work to a high degree, and a further 17% say they are affected to a very high degree. Further, 26% of the teachers found it somewhat difficult and 11% found it very or extremely difficult to withstand pressure from parents and pupils to give higher grades. 7 Teachers are, thus, likely to be under pressure from their headteachers to retain those pupils who have chosen to attend their school, especially since all schools receive per-pupil funding.
A 10% drop in enrollment has considerable economic consequences for school finances, for example. Headteachers appear to respond to these pressures, as there is ample evidence of severe grade inflation, both to ensure that pupils pass and to satisfy pupils' and parents' appetite for higher grades (Vlachos, 2010;Wikström, 2005). 8 A quote from a teacher in our survey:

My headmaster keeps insisting that all pupils must at least have an E ('pass' in the present grading system) instead
of an F. Whether the pupils meet the requirements is of less importance, something we are reminded about at every grading conference. The inflation in grading has been present for many years and it is devastating.
The principle of parents choosing the "best school for their children" has bred an emergent and opportunistic form of calculative orientation, where parents and pupils not only compare schools with respect to quality but also with respect to how responsive the school is to their perceived need for better grades.
The teachers are both subjects and objects in the grading machinery. They grade and are at the same time evaluated on the basis of the totality of their grading. The pressure they experience follows from the role imposed on teachers, they are expected to simultaneously satisfy parents and pupils and deliver results to their superiors; that is, no pupil should fail. The adjudicating and mediating functions of grading numbers used for purposes of performance management creates a continuous game between actors following the scripts of their subject positions. The burden of responsibility placed on teachers to accomplish such a feat makes them prone to adapt, respond to harsh demands and generally acquiesce. 9 The machinery of grading, measuring, comparing, avoiding risks, seeking new opportunities, and foreseeing what actors might do has transformed teachers, head teachers, administrators, and politicians into "results watchers" and "number crunchers."

Conclusions
To summarize our analysis, the Swedish educational system underwent a decisive reterritorialization according to two different principles-education as conferring an absolute right to success and education as a commodity. The reterritorialization was brought about by means of consistent and detailed interventions in several spheres; such as legislation and policy, curricula and teacher education. Over time, the grading system has increasingly come to serve as a key mechanism for adjudicating and mediating purposes at different levels in the school system. The aggregated measures Electronic copy available at: https://ssrn.com/abstract=3601128 of proportions of pupils passing or failing, as captured in various accounts, guide interventions at all levels of the school system, a role that the previous grading system never had. The grading system is a machine that grinds on producing individual and collective measures of success and failure throughout the educational system. This has served to objectify the grading system and its output; disaggregated or aggregated scores and proportions of grades are seen as truthful representations of reality.
The results of the grading system, for individuals as well as aggregated outcomes, allow the nation and its regions to be compared over time, schools to be monitored and deemed as problematic or not, parents to choose schools according to historical or expected pupil grades, and the individual pupil to be monitored as a performing or potentially failing object. The main consequences are twofold. First, the way grading measures mediate and adjudicate seems to have had an effect on the form of agency exercised by pupils, teachers, administrators, and politicians. For the actors involved in Swedish education, the representations and interventions have shaped the parameters of a new world of education, now stabilized into something perceived as an objectively existing order. They navigate in an institutional reality with particular risks and opportunities derived from pedagogic progressivism, school markets, and imposed equivalence by the systemic use of grading measures. Failure and success is defined in relation to very high ambitions accruing from the dominant pedagogical ideology and to the particular way of defining "fail" following from the grading system.
Second, public debate has become increasingly occupied with aggregated academic results at the national level, to the extent that these were a major issue in the last three parliamentary elections. While interpretations of the continuously poor results at different levels vary widely (as reflecting social problems, bad teaching, immigration, commercialization, etc.), the grading system is nevertheless accepted as a silent producer of truth. Grade statistics have come to be seen as true reflections of performance, guiding action in everything from piecemeal tinkering at the school level to national reforms. The way education is understood in Sweden has changed fundamentally. Educational outcomes are represented in objectivized measures, scores, and statistical trends, understood as reflecting the relevant reality of education. Representing educational success in a single score, derived through a number of calculations and aggregations, has served to render education as something in principle manageable, despite the lingering results.
The analysis presented here shows that reterritorialization can combine widely different principles and ideologies such as those that had a decisive impact on the Swedish education system in the 1990s. In this case, the model we use might throw some light upon the conditions of such a feat. When the reforms created havoc in Swedish education in the 1990s, the mediating and adjudicating potential of the grading system had a pivotal role in transforming many (not all) dilemmas to calculable problems. Individual performance, school performance, the comparison of schools, and national performance became problems defined and understood in the language of grading results. Performance could be measured and units could be compared in one dimension. A more generalized conclusion could be that different logics, such as those guiding Swedish educational reforms in the early 1990s, are not likely to continue to coexist unless they have been placed within a simplifying frame that allows decisions to be made according to unitarist principles. In that sense, the adjudicating and mediating functions of exercising managerial control based on aggregate measures of grades became a secondary intervention, redefining the terms for the initial reform principles.
The logic of state bureaucracy forced itself upon the ideals of the Enlightenment through the back door, using the grade system as a ramrod. The absolute right to educational success, measured as a pass grade, became transformed to measures of scores and trends mobilized to allow the exercise of government control over a mass education system. Although the guiding principles of progressive pedagogy are much less to the fore today than they were when the reforms were introduced, they still serve as a guiding principle: every pupil can reach the pass grade if given the proper support. This romantic ideal is upheld nowadays by the Swedish education system's technocratic regime, however, rather than by adherents to pedagogic progressivism. This is perhaps less unexpected than it might at first appear.
There is an ever-present tension in the modern education system between, on the one hand, its bureaucratic instrumental logic and pragmatic statecraft, and on the other hand, the persistent ambitious view of education as the last hope of realizing the Enlightenment (Hunter, 1996).
The use of the analytical framework has some decisive limits. It can be used to map and understand how a social domain is constituted ("territiorialized") and, thus, forms an object of instrumentally oriented interventions to adjudicate and mediate. Such an analysis can only specify the parameters of a system; not all its inherent idiosyncrasies, shifts, and unforeseen consequences. We nevertheless find it important to make a clear analytical separation between the parameters set in the design and construction of a system and the situated interaction that takes place within it (Bejerot & Hasselbladh, 2013). For instance, when we identify new forms of agency, these reflect changing requirements and expectations related to the roles of headteachers, teachers, pupils, and parents. The actions of individual actors are always indeterminate and manifold, even though they are framed by a particular set of prescribed obligations, expectations, and markers of truth and relevance. As mentioned earlier in the article, the last function of accounting Miller and Power (2013) describe is of a different kind than the first three. To territiorialize, mediate, and adjudicate are strategies and technologies of representing and intervening, while to subjectivize pertains to the domain of meaning. There is certainly reason to believe that the way actors are constituted in a regime of priorities, goals, truths, and hierarchically sanctioned representations will have an effect on their subjectivity-how they understand and address the proper ways of being and acting as a professional. We are not likely, however, to see homogeneity in that sphere, in the same way as we can detect homogeneity between, for instance, local systems of control and practices immediately linked to them. Those teachers are placed in a position that invites (or even pushes) them to acquiesce to demands to award higher grades will not set all of them on that route. Any attempt to reterritorialize a social domain and make it work according to a coherent logic, or as in this case to competing logics, is bound to create frictions and tensions.
The attempts to direct action and fixate forms of agency are likely also to produce actions and reactions that cannot be controlled by the social technologies deployed to make the system work in an orderly and predictable way.

NOTES
1 We used closed-and open-ended items in our 2012 survey of 1,452 teachers (63% response rate). All quantitative data are available (in Swedish) on an interactive online database (www.pifokus.se). The survey included questions on management, work environment, and wellbeing. The two items that generated the data analyzed in this paper are (i) "In my workplace is there competition between schools to attract students?" (four response options, from a low to a very high degree) and (2) "Is it hard to resist assigning unrealistically high grades given pressure from students and parents, in order to make the school appear successful and, therefore, attractive to new pupils?" (five response options, from not at all to a very high degree). In addition, the open question placed at the end was formulated as follows: "Do you have comments about your work, your health, or about the questionnaire?"