key: cord-0927928-9lgr2xyg authors: Chen, Jia-shin title: Studying up harm reduction policy: The office as an assemblage date: 2011-10-13 journal: Int J Drug Policy DOI: 10.1016/j.drugpo.2011.08.008 sha: 29b63c8ad88ae5e19bbb8d12431154940389c848 doc_id: 927928 cord_uid: 9lgr2xyg BACKGROUND: By recounting the making of the office that contributed to the implementation of the harm reduction policy in Taiwan, this paper aims to answer two questions: Who and what assembled to make this policy possible? Which conceptual tool works best to understand what this policy-making was all about? METHODS: The research was designed as a multi-sited qualitative study whose materials were collected through archival research, in-depth interviews, and direct field observation. The data were analysed on the basis of the constructivist version of grounded theory. RESULT: Formulating the office as an assemblage with heterogeneous components and shifting territories, the present work endeavours to show how it was constituted by way of guanxi, or webs of social relationship that blur the boundary between the private and the public, the governmental and the social. CONCLUSION: This “studying up” approach is hoped to elicit more research on the office in which harm reduction policies are made into the backdrop of drug users on the street. Harm reduction became the policy of choice when the 2004 Health Statistics of Taiwan identified injection drug users (IDUs) as the most rapidly expanding subgroup of newfound HIV-positive cases. The Centres for Disease Control (CDC) in Taiwan (2004, not paginated) explained this policy move as follows: "[The CDC] found a pattern of infection that was regionally organised and clustered. Injection drug users contract HIV because they share needles or diluting solution." In August 2005, harm reduction was initially implemented as a pilot programme in four major sites (Taipei City, Taipei County, Taoyuan County, and Tainan County) by mobilising a wide array of personnel and by recruiting a great number of participating organisations, including public hospitals and volunteer pharmacies. With promising statistics that subsequently demonstrated its efficacy, this pilot programme soon expanded into a nationwide programme in the following year (Yang, Yang, Shen, & Kuo, 2008 ). Claiming its great success in preventing an impending HIV epidemic amongst IDUs and even the general public, CDC gradually withdrew its involvement in harm reduction policy around 2009. An ad hoc strategy at first, harm reduction policy was integrated into routine medical and public health practices and divided amongst several authorities before long. As the threat of HIV amongst IDUs wore off, the idea of harm reduction no longer aroused public discussion, although the Asian Harm Reduction Network (2008) showcased its success story as an exemplary case. The current paper, however, does not focus on how successful this policy is. Instead, it explores the sociological significance of the ways through which the policy was made as such in Taiwan. To be more specific, how did various actors assemble to make this policy possible? Which conceptual tool works best for us to understand the process? The current paper is the first to examine the shaping of harm reduction policy in Taiwan, and in doing so, attempts to answer the two questions mentioned above. First, it describes the formation of the office as an assemblage that brought forth public health measures as governmental technologies. Second, by describing the ways and elements of assembling the office, the paper attempts to exemplify how this notion of the office makes better sense of the messy conditions, in which collective yet heterogeneous actions and interactions take place in what is commonly over-simplified to be policy making. Overall, the office can be seen as a biopower apparatus that comes into being through the technologies of government it creates. The office and its technologies of government are co-constituted, and together they form a problem space that shapes the backdrop of IDUs' street life. Taiwan's harm reduction policy can be seen as a contemporary manifestation of biopower as it concerns itself with the well being of both the individual and the population at large (Foucault, 0955-3959/$ -see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.drugpo.2011.08.008 1976). However, there is no reason to believe that biopower works in Taiwan the same way as it does in Europe where it originates. Therefore, the very means by which policy components assemble illustrate the local character of biopower, thereby shaping the conditions of life with which drug users in Taiwan have to cope. The present work is inspired by the notion of assemblages from the works of Guattari (1983, 1987) . Deleuze and Guattari in A Thousand Plateaus (1987, p. 88 ) describe assemblages as aggregates of heterogeneous elements with aspects of "machinic" content (bodies, actions, and passions) and "enunciative" expression (acts, statements, and incorporeal transformations attributed to the bodies). In addition, the boundaries of assemblages are always shifting, a process called deterritorialisation and reterritorialisation. The concept has been utilised in a number of studies on medical sociology and anthropology as well as science and technology studies (STS) to comprehend the amorphous and ephemeral aggregations involved in contemporary technoscientific projects that define public life. Inspired by Keane's (2003) depiction of the practices and technologies of harm reduction policy as an assemblage, I further suggest that the office making this policy be seen as an assemblage as well. Keane's suggestion and mine echo the concept of ethno-epistemic assemblages described by sociologists Alan Irwin and Mike Michael (2003, p. 119) , which refers to aggregates of people and truth claims situated in the contemporary decision making of public affairs. The "ethno" part also addresses the locality and indexicality of the aggregates. Criticizing a citadel version of science-society relations where boundaries are demarcated, they address the rhizomic involvement of knowledge and publics in many, if not all, governmental decisions. Therefore, they suggest that the relationship of science, society, and governance be reconceptualised and even refashioned (Irwin & Michael, 2003) . Although it has never been fully theorised and substantiated by empirical studies (except, perhaps, DeLanda's (2006) efforts), the concept of assemblages, with its variants and derivatives, can be seen in the study on the "global assemblages" of contemporary technologies and ethics (Ong & Collier, 2005; Ong, 2010) and also Greenhalgh's ethnographic work (2008, and also Greenhalgh & Winckler, 2005 ) that traces China's one-child policy assemblage. Greenhalgh's work can be seen as most exemplary of the complexity, dynamism, and contingency in policy-making. It demonstrates how these cybernetics-assisted population projections were constructed by various actors and entangled in ideologies, national desires for modernity, and trust in scientific truth claims. When two concepts from different backgrounds are juxtaposed, certain theoretical risks are posed (for another example, please see McCann, 2011) . The two concepts, biopower and assemblages, do not appear as commensurate as they are heuristic, because they are both slippery and situated ideas. A preliminary attempt of combining governmentality and assemblages can be seen in the work of Irwin and Michael (2003) . They comment on the ways in which biopower subjects and subjectivizes modern "scientific" citizens, who are concerned with questions of governmentality such as how "the way we think about exercising authority draws upon the theories, ideas, philosophies, and forms of knowledge that are part of our social and cultural products" (Dean, 1999 , p. 16, cited in Irwin & Michael, 2003 . However, by treating the ethnoepistemic assemblage as an instrument of biopower through which a "scientific citizenship" is constituted, they fail to show how the phenomena of assemblage making diversify the very notion of biopower. Given the fact that the existing attempts of combining Deleuzean assemblages and Foucauldian biopower have been patchy and incomplete, this current paper endeavours to conceptualise both by introducing guanxi. The manoeuvers of guanxi, as will be shown, reflect actors' connection-building efforts that establish and sustain, however temporarily, the policy assemblage. It enables the assemblage not on a foundation of rationality but on a thread of shared sentiment. Therefore, biopower does not only work through sentiments; it is also bred out of sentiments. The research, designed as a multi-sited qualitative study on harm reduction, aims to examine the processes of policy making and implementation in Taiwan (Chen, 2009 ). The fieldwork was undertaken mostly in Taipei and Taoyuan, two major residential areas in northern Taiwan, as well as in Kaohsiung and Tainan, cities of southern Taiwan. The materials of this multi-sited study were collected using three methods: archival research, in-depth interviews, and direct field observation. They focus on the ways in which harm reduction was re-/presented by actors and these actors re-/presented themselves. Archival research refers to a detailed textual analysis of available literature in Taiwan, such as medical journals and daily newspapers. It aims to disclose and supplement the historical and discursive background from which harm reduction emerged. Interviews with people involved in policy making and implementation constituted the second part. Collected by snowball sampling, these in-depth interviews were the main source of information. For a qualitative research that focuses on interactions within and about policy-making, it is very difficult to collect real-time facts, because they are often invisible to the public (and also to the researcher). In relation to this, Wedel, Shore, Feldman, and Lathrop (2005, p. 41) contend that intensive interviewing is often "the only means of gathering first-hand information." In the present study, 32 people were interviewed, including bureaucrats, psychiatrists, HIV professionals (workers and researchers), pharmacists, and the like. Most interviews lasted one to two hours long, starting with questions that engaged the interviewees to tell their stories of participation in harm reduction. The interviews were transcribed verbatim, and the contents were taken as interpreted phenomena of the responders rather than as outright social facts. In addition to supplementing the interviews, the third part of the data came from field notes based on direct field observation, which was extended to places such as methadone clinics, needle-distributing pharmacies, and academic conferences. In these places, the author observed at close range both verbal and non-verbal performance between bureaucrats and specialists, or between drug users and their service providers. The analysis of collected data from the above three methods was carried out on the basis of the constructivist version of grounded theory's principles and methods, including its postmodern revision, situational analysis (Charmaz, 2006; Clarke, 2005; Strauss, 1987) . These collected data were, as Strauss (1987) suggests, serially coded (i.e., open, axial, and selective coding), analysed, and compared with existing data and frameworks. Unlike the objectivist version of grounded theory that tends to identify social facts or processes "out there," this constructivist approach aims to include as many voices as possible and to see "both data and analysis as created from shared experiences and relationships with participants and other sources of data" (Charmaz, 2006, p. 130) . Therefore, the resulting data and analysis were treated as situated and relational. As Clarke (2005, p. 37) argues, "In simplified form, situational or relational ecology is closest to policy arena analysis." This pluralistic and interactional approach seems to capture without reification the various stakeholders, objects, practices, and discourses involved in the policy-making process. A central concept of this current article, the office, describes the aggregates of various actors in the creation of a policy such as harm reduction. They are usually, but not always, located in the government. This term is adopted for good reasons. Previous literature on drug use concentrated mostly on the street life of drug users (Page & Singer, 2010) , but this street-based approach failed to "study up" and expose how the policy was created and made drug users' lives marginalized and disenfranchised. The office is a conceptual tool that depicts the dynamism and changeability of policy-making. It refers to neither a circumscribed locality nor a well-defined group of human and nonhuman elements. As Nader (1972) contends, studying up and down would lead researchers to think anew by reversing many commonsensical questions, and not doing so may seriously impair further development in social theory and anthropological insight. Anthropological research on organisations and especially developmental projects has demonstrated the importance of "studying up," whose significance lies not only in the manners that development unfolds vis-à-vis local bureaucracies, institutions, and people (Mosse, 2005) , but also in the ways that policies are rendered anti-political (Ferguson, 1990; Mitchell, 2002) . These achievements have shown that a policy and its stakeholders can make good subjects for further ethnographic scrutiny. Thus, if researchers want to view drug problems in full light, it is important to look at what drug users are doing on the street (the "studying down" method) and at how policymakers are picturing them whilst proposing administrative measures (the "studying up" approach). A little more space should be devoted to the discussion of positionality. As Mosse (2006) rightly argues, a researcher does not always enjoy leniency from research subjects in terms of collected data or analytical interpretations. Instead, he/she needs to know that the interpretations are themselves interventional because these give meanings to what the research subjects believe and practice in terms of policy projects. On this note, the futility of the insider/outsider dichotomy is all too obvious. Every policy ethnographer may find him/herself in a part-insider/part-outsider position (Mosse, 2005, p. 13) . I knew this from the outset. It was not too difficult for me, a psychiatrist with some guanxi, to have access to addiction specialists and government officials. For example, right after the commencement of my field work, I was invited to contribute to a governmental policy plan on drug use. However, I sensed the unease as I went on, especially when I interviewed my previous colleagues and esteemed scholars from a standpoint that was supposed to be critical. I suffered similar, although much less severe, criticism from policy participants as Mosse (2006) did. However, as he pointed out, the researcher's representation of what happened in the field reflects his/her own perspective that can be criticized but always needs to be respected. The making of the office is a dynamic process whose heterogeneity, variability, and indeterminacy can be best demonstrated by examining its inception. When the escalating HIV infection amongst IDUs was discovered in early 2005 (Fig. 1) , the Taiwan CDC determined that previous anti-HIV measures, including youth rallies and health slogans, were inadequate. In addition to the discontent with past anti-drug crusades, the recently controlled epidemic of Severe Acute Respiratory Syndrome (SARS) in 2003 also led to CDC's rapid action. SARS was a scary illness that CDC fought not long ago, and now it wanted the public to believe that HIV is another dreadful challenge. Retrospectively, the SARS outbreak might have prevented the CDC from reacting on early signs of the HIV epidemic amongst imprisoned IDUs. However, at this moment, by sounding the alarm, CDC helped promote the image and urgency of "a house on fire" that necessitated immediate responses. As per recommendations from the World Health Organisation (WHO) and the Joint United Nations Programme on HIV/AIDS (UNAIDS), a pilot harm reduction programme launched in August 2005. The programme had three major components: enhanced education and screening, drug substitution treatment, and needle syringe programmes. However, most Taiwanese people knew only the latter two because of the media's attention on them. The combination of two key controversial issues, HIV and addiction, meant a greater likelihood of vehement public opposition. A more powerful and persuasive tool was therefore needed to advance a new anti-HIV/anti-drug policy against the long-held suppressive ideology. Statistics played this role, as Fig. 1 shows. Numbers speak, and they now speak loudly. However, the statistics on which this policy was founded was not unchallenged. When someone was tested HIV-positive, public health workers would investigate into his/her risk factors such as injection drug use and sexual history. When the person was known to have habits of unprotected sex and drug injection, only the latter would be counted. A local paper, Taipei Times, criticized this "one person, one risk factor" approach: "Other critics of the CDC's pilot programme said blaming drug use for HIV also aggravates the situation and makes it difficult for infected persons to get help. Bombarding the public with statistics that finger point needle syringe is misleading, said Tony Lee ( ). . ..The figures do not discriminate between unsafe and drug injection practices, suggesting transmission between IDUs could be a result of unprotected intercourse. The CDC viewed these statistics in a mutually exclusive way. If we put all our money and energy into providing clean needles and ignore the risks of unsafe sex, then we give the wrong impression on the public. . .. It is more difficult to get IDUs to use a condom than a clean needle" (Freundl, 2005) . Not only does the "one person, one risk factor" approach imply a hierarchy of risk factors which can be outright misleading, but the ways by which these numbers were generated determined the "fact" of an IDU-propagated HIV epidemic. The cited statistics compelled the Taiwan CDC to pursue harm reduction programmes and work with new allies other than HIV professionals, so it found addiction specialists that were long marginalized in medical or social arenas. Disputes such as this one exemplify the ways in which Taiwan's harm reduction policy was made. Although statistics can be a misleading and even fallible element in the policy assemblage, the numerical governance illustrated in this case actually underwrites many policies in Mandarin-speaking countries such as Taiwan and China (for example, see Greenhalgh, 2008) . Numbers seem to conjure a sense of truth and trust, but the problem with numbers in policy-making lies not just in their potential fallibility but also their inability to frame drug use as an embedded socio-behavioural problem. Frontline workers and drug users knew best, but it was mind-boggling that drug users were often silenced in the central government. This ignorance led to waste of money on unfit needles and syringes, oversized care packages (boxes or envelopes with health information and clean paraphernalia), and so forth. The dependence on quantitative data seemed to have reduced drug users to numbers instead of human beings who feel, think, and act (see also Bourgois, 2004) . Along with numbers, Taiwan's harm reduction policy depended on experts' input. Notably, HIV professionals (researchers and workers) and addiction specialists (mainly psychiatrists) were included in the policy-making process, not to mention the seasoned technocrats in CDC. The level of their participation varied as the harm reduction policy unfolded. Although the usual rule was to consult local experts before making a policy, the question at this time was qualification and legitimacy. Considering the multiple sources of expertise recruited and the urgency of policy implementation, a unified version of action plan was desperately needed. Thus a high-rank CDC official explicated the reason by saying that, "To unify domestic opinions, [the visit of internationally recognised experts] was pivotal. After all, we venerate foreign experts. It is simply not easy for us to persuade [local experts]." Certainly, this strategy aroused mixed feelings amongst domestic experts. Some experts felt marginalized, whereas some grabbed the opportunity to jump onto the bandwagon. Carefully crafted and articulated, the finalized policy proposal for the pilot programme was an amalgam of governmental planning, expert knowledge, and institutional interaction. However, clashes and conflicts abounded. Two controversies stood out as the pilot programme was advanced in the summer of 2005: Which medicine should be used in the substitution programme? Moreover, how did central government departments coordinate? Methadone or Suboxone? The first controversy was both technical and political, and it involved medicine, a technological object, that distinguished harm reduction from other policies (Gomart, 2002b) . When the drug substitution programme was first organised, both methadone and Suboxone were proposed as suitable candidates. Methadone had a history of use for more than four decades and it was offered in syrup in Taiwan, whilst Suboxone, a sublingual tablet of two off-patent medications (buprenorphine and naloxone mixed in a specific ratio), was new in the market. Which medicine would be better for substitution treatment seemed to be a question on their properties, but "better" herein did not simply refer to their physiological, economic and normative properties. Instead, it referred to how these properties were seen by the participants as the performances of the two medications. In charge of all controlled substances, the National Bureau of Controlled Drugs favoured Suboxone because methadone, a Schedule Two Controlled Drug, was considered more addictive and thus more prohibitive than buprenorphine, a Schedule Three Drug. Furthermore, the pharmaceutical design of Suboxone made it even safer. Taken orally as indicated, buprenorphine may soothe craving without causing adverse effects of overdose. If drug users attempt to crush, dissolve and inject the medicine, the naloxone therein, an opioid antagonist absorbable only through the blood stream, can lead to severe withdrawal symptoms and discomfort, thus preventing further injection attempts. This combination did not only win Suboxone a patent but also inscribed a way of disciplining its users. Advocates of Suboxone believe that the risk of diversion-related overdose can be prevented. However, this was not CDC's concern. Since this policy was enacted by crisis and the funding was temporary, the CDC had to achieve success in the shortest time and at the lowest cost possible. As a result, it preferred methadone to Suboxone for two simple reasons. First, it costs less for both the user and the government. Suboxone could cost twenty times or more than methadone. Second, if any trouble took place during policy implementation, CDC would have some foreign experience for reference because it had been used elsewhere for decades. Besides, the half-life of syrup methadone would compel its users to return to distribution sites about every 24 h, making them visible for the sake of surveillance. Comparing Suboxone and methadone, an addiction specialist argued, "People often spend lots of time and money every day getting to the hospital just for a sip [of methadone]. I think some people will want to shift to Suboxone even if it costs more, because time is money and they do not have to come to the hospital every day." His depiction implies a potential stratification phenomenon in which more affluent drug users may find Suboxone affordable but those who have limited means can only use the cheaper methadone. Whether this opinion will be confirmed or not, it is noteworthy that Suboxone and methadone would actualise different target populations even if these people were all too often categorised simply as "injection drug users" (also see Fraser & Valentine, 2008 and Gomart, 2002a) . In this case, newer and more convenient Suboxone was not better, but cheaper and older methadone was. Although CDC allowed both medications to be used in the pilot programme, it apparently preferred methadone. In the long run, most if not all substitution programmes offered only methadone. This section is not to be read as a story of participants' differences in preference, but one of medications' performances that resulted in different potential policy assemblages. Similar to what Gomart (2002a) illustrates in her observation of methadone clinics, the ontology of methadone determines and is determined by service providers as well as target populations (i.e., drug users). In this case harm reduction was seen by CDC as a biomedically sound and historically reliable strategy. CDC did not really select methadone as their better alternative. Instead, methadone got itself selected because it proved to be agentic, or "making a difference," in this urgent situation. (On making a difference as a manifestation of object agency, see Latour, 1999) . The lack of consensus also manifested itself in the second controversy, which involved the definitional ambiguity of harm reduction amongst different sectors of the government as they imagined and practised what harm reduction was. From the outset, astute frontline public health workers as well as pharmacists who helped distribute and recycle needles expressed their grave concerns about being arrested by the police as drug solicitors. Seen as sanctioned medical treatments, drug substitution treatments were less likely to be subject to police actions; however, hospitals that offered these treatments were still worried about increased drug-related activities and police arrests around their sites. These issues involved authorities other than those involved in public health, so the Department of Health (DOH) that supervised CDC had to work out a plan with the Ministry of Justice (MOJ) and the National Police Agency (NPA). These units of central government had to coordinate their actions urgently so as not to incriminate frontline workers for doing harm reduction. This coordination issue was deemed by many interviewees as indispensably vital in the early stage of harm reduction policy implementation. Notably, it was not so much handled through formal channels as facilitated remarkably by the use of guanxi ( ), which entailed informal negotiations, favour exchanges, and interpersonal sentiments. A long-term HIV researcher insightfully commented on the key to harm reduction policy: "Guanxi is important. It is important in the whole world, but especially important in Taiwan." Why was guanxi especially important? Formally, the interdepartmental coordination had to go through a joint committee within the Executive Yuan, the highest administrative agency in Taiwan that superintends all the above units. However, according to some participants, the committee was often hindered by sectionalism. A discontented high-rank CDC official said, "People are all talks, but no one has ever pointed out the inherent conflicts of these measures." Meanwhile, the MOJ demanded legal deliberation on the matter of harm reduction. Without MOJ's affirmation, public health workers hesitated and police actions varied. However, epidemics like this could not wait. Hou Sheng-mao, then Minister of Health, knew this would be a tough task because "the key was a conceptual U-turn." The previous ideology of suppression had to be overturned, at least temporarily. He therefore tried to orchestrate the whole project by building social bonds with other department leaders, especially with the Minister of Justice Shi Mao-lin. Hou summarised it as a tricky combination of luck and efforts: "I am very, very lucky, but I worked so much on the Minister [of Justice]." To better promote the concept of harm reduction, Hou reckoned in public his relationship with Minister Shi as old schoolmates. Hou recalled, "Actually we did not know each other during high school even though we graduated in the same year. . ..but when I found out [about this], I would hang around with him, invite him to dinner. . ..or, share good stuff that I had." This gift economy within the bureaucracy constituted a prominent feature of policy-making because harm reduction was controversial on many levels. "But you have to act like a salesman who repeats [your idea] seven times. That is the spirit of a missionary," he said. The web of guanxi did not stop at the level of ministers. Various business trips were organised for other stakeholders, including prosecutors and psychiatrists. Many of them had not known one another before the trip, but their relationship would "naturally" arise due to it. Many professionals mobilised their overseas guanxi by making phone calls, writing e-mails, or attending international conferences to collect as much information as possible. Guanxi became part of their newly acquired expertise on harm reduction. All these endeavours helped weave an interpersonal web of guanxi, which is of vital importance to our understanding of the office in this paper. Guanxi can be roughly translated into English as "connection" or "relationship." According to Yang (1994) , it is widely, although often implicitly, recognised in Chinese societies, including China, Taiwan, and even Singapore; it usually refers to social bonds established through the exchange of gifts, favours, and banquets. A more defining feature of guanxi is the social relationship with both commitment and sentiment that goes beyond organisational barriers and, sometimes, even implies some kind of "backdoor politics" (Langenberg, 2007) . Noting that the art of guanxi, or guanxixue ( ), arose in the middle of the Cultural Revolution, Yang insightfully points out its root in ancient Chinese thoughts and conflicting roles in China's modernisation projects. Guanxi, she argues, is not an extension or manifestation of expanding state or official power. Instead, it constitutes a rhizomic form of power that "exerts a subversive effect on the microtechniques of administrative power" (Yang, 1994, p. 190) . In her depiction (p. 295), guanxi was so culturally laden that it overshadowed "the two forms of full-fledged civil society: the individual and the association or group." It enacted a civil sphere, or minjian in Chinese, which was distinct from its counterpart in Euro-American contexts. However, I would like to push Yang's argument further. This is because I believe that guanxi does not only exist outside the political sphere represented by the CDC bureaucracy or the Executive Yuan; it is actually present inside as well and plays an enlivening part within such context. Minister Hou and some other main advocates of this policy thought that appealing to sentiments was necessary for a controversial policy like this. This is called bua ganjing ( ) in Taiwanese, meaning the use of all efforts possible to win the approval and affection of someone else. Ganjing, or ganqing in Mandarin, literally means "feelings and affections." A major advocate of this policy, Yen Chun-Zuo, then Deputy Mayor of Tainan County, recounted his reaction when the proposal of harm reduction was softly put off by the Secretary-General of the Executive Yuan: "I told him this would surely be a risky business. Even if a plan was made, it was never certain which local government would want to do it. . .. But I told him I would definitely make this work. He was touched by me, so he told the Premiere." In the case of the office, it was ganjing, not bureaucratic rationality, which acted as a hinge upon which comradeship was built and agreement was reached. This constituted a unique way of making the office: Schoolmateship, sending gifts, and exchanges of various good-will contacts and promises all contributed to the making of a guanxi web that shaped the assemblage of the office. Although the building up of strong ties through this channel may "breed local cohesion" at first but "lead to overall fragmentation" in the long run (Granovetter, 1973 (Granovetter, , p. 1378 , the significance of guanxi in circumventing the obstacles of inter-departmental communication in this case is beyond doubt. A detailed comparison of guanxi and a similar concept, social capital, is beyond the scope of this current paper. Suffice it to say that they are similar because they both address the value of social relationship as some kind of resource. Nonetheless, they differ in terms of the focus of analysis. Guanxi portrays more closely the processes of establishing and utilising social ties by means of sentiments (ganqing), whereas social capital offers a rationalist explanation for individual actions within a given social structure (Field, 2003; Lin, 2001) . In the end, the two departments reached a consensus that they would work out some plans for the treatment of IDUs. The measure of postponed prosecution was an example, in which IDUs were asked to receive substitution treatments instead of going directly to correctional facilities. Prosecution was applied only when arrested IDUs did not comply with the treatment. As Chen (2011) has illustrated, the substitution treatment was not so much based on human rights claims as on public health demands. Drug users were provided with new entitlements but at the same time subject to greater surveillance. On the other hand, the web of guanxi was not complete without the NPA. The above-mentioned high-rank CDC officer explained his strategy: "I knew I needed to hold onto the international trend, so I spoke of WHO all the time and I found someone internationally renowned to manage the opposition. . .. If I had to tell the NPA people what to do, they would think, 'Why should I listen to you, the DOH folks?' However, if I found a famous and experienced Australian police officer, they would probably listen." Eventually, this strategy worked. The NPA sent an announcement to patrolling policemen that discouraged "aimless searching" around methadone clinics and needle distribution sites. The announcement was even printed in posters and affixed on the walls of many places (Fig. 2) . The poster in Fig. 2 , similar to the document in Riles' booklet (2000) or the medical record in Berg's analysis (1996) , is not just a descriptive representation of guanxi but also a prescriptive presentation of the office. In other words, guanxi was inscribed onto materials that, in turn, directed people's practices. In a nutshell, the office thus formed was a web of social ties that connected people, knowledge, and technology. Contrary to the common belief that organised actions bring forth policy success, the office as an assemblage proved to be functioning in a state of marked heterogeneity and occasional in-coordination. Distinct from a bureaucracy rationality that Weber (1946) describes, Taiwan's ways of assembling illustrate a unique pattern of policy negotiation, which is deeply embedded in the local political culture of guanxi. That the office can be analytically treated as an assemblage is not something distinctly Taiwanese. In fact, the notion of the office is applicable elsewhere when the black box of policy-making is to be studied and opened up. As demonstrated by the case study in the current work, the office is characterised by its aptness to address multiple yet heterogeneous human and non-human actors (statistics and methadone in this case), especially the rhizomic and dynamic relationships amongst them. In this case, the constitutive elements have been illustrated to be linked by guanxi, which exerted instrumental functions but depended on emotional bonds. Although the overall effects might be abbreviated as biopower, the idea of the office as an assemblage compels readers to study up the messy aggregate of the power apparatus. The studying up approach does not usually agree with perspectives that treat policymaking as a staged process, in which multiple social organisations and political interest groups compete as policy windows open, nor does it agree with the perspectives that see a policy as marked by punctuation and equilibrium (Sabatier, 2007) . Most notably, the approach does not limit itself to analysing domestic factors only; it is always transnational. The transnationality of this assemblage is illustrated on multiple levels. Not only is the pathogen that initiated the policy transnational--the HIV strain is proven to originate in Southwestern China (Chen & Kuo, 2007) , but the expertise and policy know-how are transnational as well, as shown above. Moreover, it shows from my interviews that immanent in the harm reduction policy is a desire for global recognition, because Taiwan has been excluded from international organisations such as WHO and the UNAIDS for decades. This approach, exemplified by this Taiwan case, echoes Ong and Collier's (2005) notion of global assemblages. It presents a picture distinct from Fraser and Valentine (2008) not just because of the fields researched (Taiwanese policy contributors vs. Australian methadone-distributing sites), but also due to the dimension of transnationality that is present in the former but comparatively absent in the latter. In short, this transnational dimension of the office argues against an implicit premise in most policy studies that situate policy processes within national borders. This understanding is important because it is typical for a late advanced country such as Taiwan to make a policy this way when officials in charge feel compelled to pick up precedents, select and introduce expertise, and transplant relevant know-how due to the relative paucity of existing local research that argues for or against any given position. Policies thus made are invariably entangled with elements from other countries. The office, accommodating interpersonal guanxi as it constitutive skeleton, does not dismiss the intricacies of bureaucracy and knowledge or limit itself to domestic terrains. In summary, this current work demonstrates the potential of the office as an analytical concept. It addresses both local uniqueness and global connectedness. Studying up the office as an assemblage helps to theorise and propose a novel perspective on how policies are assembled and how biopower is manoeuvered. This analytical approach and the new focus are expected to help future researchers examine drug issues in a more comprehensive manner. Harm Reduction in Taiwan: Showcasing Asian Leadership on Harm Reduction Practices of reading and writing: The constitutive role of the patient record in medical work Anthropology and epidemiology on drugs: The challenges of cross-methodological and theoretical dialogue HIV/AIDS plus Illegal Drugs Constitute a New Challenge for Prevention and Control Posters for HIV Prevention Constructing grounded theory: A practical guide through qualitative analysis Assembling harm reduction policy in Taiwan Beyond human rights and public health: Citizenship issues in harm reduction HIV-1 in Taiwan Situational analysis: Grounded theory after the postmodern turn Governmentality: Power and rule in modern society A new philosophy of society: Assemblage theory and social complexity Anti-oedipus: Capitalism and schizophrenia. Minneapolis A thousand plateaus The anti-politics machine: "Development", depoliticization, and bureaucratic power in Lesotho The History of Sexuality Substance and substitution The point of needle exchange. Taipei Times Methadone: Six effects in a search of substance Toward generous constraint: Freedom and coercion in a French addiction treatment. Sociology of Health and Illness The strength of weak ties Just one child: Science and policy in Deng's China Governing China's population: From Leninist to neoliberal biopolitics Science, social theory, and public knowledge Critiques of harm reduction, morality and the promise of human rights Guanxi and business strategy: Theory and implications for multinational companies in China For David Bloor. . .and Beyond: A Reply to David Bloor's 'Anti-Latour'. Studies on History and Philosophy of Science Social capital: A theory of social structure and action Current Status of Harm Reduction Policy Implementation in Taiwan (PowerPoint file in Chinese Veritable inventions: Cities, policies, and assemblage Rule of experts: Egypt, techno-politics, modernity Cultivating development: An ethnography of aid policy and practice Anti-social anthropology? Objectivity, objection, and the ethnography of public policy and professional communities Up the anthropologist: Perspectives gained from studying up Introduction: An analytics of biotechnology and ethics as multiple scales Global assemblages: Technology, politics, and ethics as anthropological problems Comprehending drug use: Ethnographic research at the social margins The network inside out Theories of the policy process Qualitative analysis for social scientists Toward an anthropology of public policy The changing epidemiology of prevalent diagnosed HIV infections in Taiwan Gifts, favors and banquets: The art of social relationship in China This paper is largely founded on my previous Ph.D. research in the University of California at San Francisco (UCSF). I would like to thank my committee members for their precious comments on my work. They are Professors Adele E. Clarke, Philippe Bourgois, Howard Pinderhughes, and Vincanne Adams. Also, I thank my informants for their time, experience, and opinion. This paper has been presented in several conferences, from whose participants I received a lot of useful advice. I am especially grateful to the anonymous reviewers who helped me improve this piece of writing.