Should digital files be considered a commons? Copyright infringement in the eyes of lawyers

Abstract In this article, we draw on a survey conducted with elite upcoming lawyers from all around the world to shed new light on the ethical acceptability of file sharing practices. Although file sharing is typically illegal, our findings show that lawyers overwhelmingly perceive it as an acceptable social practice. The main criterion used by lawyers to decide on the ethical acceptability of file sharing is whether or not the infringer derives any monetary benefits from it. Further, our findings show that lawyers in the public sector (including judiciary and academia) are even more tolerant of online copyright infringement than those in the private sector. Interestingly, our data suggests that this is largely the result of self-selection: lawyers who lean more on the side of broad disclosure and social sharing tend to orient themselves toward the public sector. Implications for the current state of the debate on the reform of copyright law are discussed.


Introduction
In the early 2000s, the advent of peer-to-peer file sharing technologies a la Napster and The Pirate Bay dramatically increased the public's consumption of digital media-concurrently causing a decrease in industry revenue, as well as an increase in social welfare (De Vany and Walls 2007;Rob and Waldfogel 2006). As a result of the potential threat posed by this technological shock to the long-term economic viability of creative works, copyright protection for digital goods emerged as a high stake legal issue (Li and Nergadze 2009;Nowak 2016).
The past decade, however, has seen the rise of new market entrants (e.g., Spotify and Netflix), which leverage technology to engage in bundle pricing. This evolution in the business model largely restored industry revenue and profits (Aguiar and Waldfogel 2015;Bai and Waldfogel 2012;Danaher and Waldfogel 2012;Dejean 2009;Evens and Donders 2018;Vonderau 2019;Waldfogel 2010), while maintaining consumer surplus (Beekhuyzen, von Hellens, and Nielsen 2015;Waldfogel 2012aWaldfogel , 2012b. Music and video as service have become de facto standards in the industry. Also, media creators and distributors have evolved and adapted to digital piracy (Aversa, Hervas-Drane, and Evenou 2019). Spotify in particular-created in Sweden, the hotbed of piracy-drew significantly from the pirate content distribution model, and built a sustainable competitive advantage based on intuitive usability and an excellent understanding of customers' needs (Sun 2019), as well as a clever combination of freemium and paid services (Yan and Wakefield 2018). The Spotify model is largely similar to that of Netflix (Aguiar and Waldfogel 2018;Lobato 2019). More traditional media corporations also try to adopt this model, even though they tend to be less successful than the new entrants, and face additional competition from tech giants, such as Amazon and Google (Wayne 2018).
Following the widespread diffusion of legal services for downloading and streaming high-quality content, the European Commission commissioned a study to inform the formulation of its copyright policy. Consistent with the academic literature cited above, the 2015 report concluded there was no robust statistical evidence of sales displacement by online copyright infringement. In effect, the practice of file sharing is both revenue neutral for the industry as well as globally welfare enhancing (van der Ende et al. 2014). Nevertheless, the tension between expanding the public domain and strengthening copyright protections for digital media continues to grow (Jetha, Berente, and King 2017). The recent debates in the EU Parliament about Article 11 and 13 of the copyright reform, and the US advocating for both increased intellectual property protection and a free and open Internet (Haggart and Jablonski 2017), are just two out of many examples.
In the meantime, the very notions of ownership and fairness when sharing digital goods (i.e., goods with zero reproduction cost) underwent significant changes. The understanding of what carries value, what ownership means, and what is fair, have all been largely redefined by consumers-with wide social acceptance of copying and sharing digital files, be it among the youngest generations (Altschuller and Benbunan-Fich 2009;Feldman and Nadler 2006;Robertson et al. 2012;Wang and McClung 2012) or the broader population (Mandel 2014;Mandel, Fast, and Olson 2015;Reyman 2009;Svensson and Larsson 2012).
In spite of the changes in business models and social perceptions and norms, file sharing is typically prohibited (and severely repressed) across legal jurisdictions, and exceptions to this rule (e.g., the fair use doctrine in the US) are both rare and heavily restricted. 1 In general lawmakers' approach to copyright protection of digital media has mostly remained relatively stable over the past decade (Drahos 2016). Legal scholars have been calling for a rethink of the intellectual property regime (Kapczynski 2011).
Unlike previous studies based on perceptions of populations such as students and laypersons, we focus on how lawyers themselves perceive of file sharing. We elicit the views of an international panel of lawyers on the ethics of file sharing, and analyze their perceptions of fairness as a function of: (i) the rationale provided for infringing copyright, and (ii) lawyers' personal characteristics. 2 Two lawyer-specific characteristics are of particular interest to us in this respect: (i) their region of origin, and (ii) their intended sector of professional practice (i.e., public or private). These characteristics are especially relevant to file sharing ethics. First, developed and developing economies face very different challenges in terms of copyright protection, and lawyers coming from different cultural backgrounds could very well take different ethical stances with respect to file sharing practices. Second, lawyers operating in the public and private sector might strike an entirely different balance in the tradeoff between letting society benefit widely from creative works on the one hand, and properly incentivizing authors on the other. 3 To this end, we studied the perceptions of international LL.M. (Masters of Law) students at Harvard University based on 109 completed quantitative survey questionnaires and 50 qualitative interviews. 4 These students already had a law degree from their home countries, providing a unique opportunity to compare the perceptions of top-notch lawyers from around the world in a single setting.
We decide to conduct this study from the social studies of the information society perspective-not ethics. Broadly, in information society related research domains, ethics-focused decision-making has been quite well researched over the last two decades (e.g., Chen, Bharadwaj, and Goh 2017;Moores and Chang 2006;Walsham 1996). Specifically, in the case of unauthorized digital file sharing, the existing research can be broadly divided into: moral reasoning focused or solutions focused (see literature analysis in Siponen and Vartiainen 2004). We are taking a somewhat different approach, following the call to shift from computer ethics to the practicalities of technology governance (Stahl et al. 2014). We believe that it is important to zero in on the actors' own conceptualizations and views, and avoid framing the conversation in terms of ethical models, while accepting that these views need not be theoretically coherent. Our findings do have clear practical implications-our purpose is mainly to understand and uncover something new about an understudied, yet highly relevant professional group for issues related to file sharing, rather than make an incremental contribution that refines the existing literature on proposed legal solutions (see, e.g., Aigrain 2012).
Ethnographic research on digital file sharers shows that they are often aware of the illegal nature of their practices. They go through with file sharing anyway as their needs are not satisfied by the market, and eventually perceive file sharing as acceptable (Beekhuyzen, von Hellens, and Nielsen 2015). The fear of punishment typically affects less habitual and less skilled downloaders (LaRose et al. 2005). These insights suggest that self-policing and personal ethics-effective in restricting Internet uses in some contexts (e.g., Li et al. 2014)-may not be sufficient for curbing unauthorized file sharing practices. For a nuanced understanding, identification of the contexts in which file sharing practices are perceived to be fair by different populations is critical (Jemielniak and Przegalinska 2020). Such understanding may also be a prerequisite for crafting effective policies.
It is rather surprising that no research to date has focused on lawyers' perceptions of file sharing practices, especially since their expertise is regularly invoked in lay discussions on the nature of copyright and its regulation (Fiesler, Feuston, and Bruckman 2014). While the attitudes of lay people on the ethics of file sharing are probably influenced by their misunderstanding of copyright law and its underlying objectives (Mandel 2014), it is less likely to be the case with the lawyers (Mandel 2017). Also, lawyers are at the forefront of the change in social norms brought about by digitization. 5 Further, lawyers have political influence and are likely to hold extreme views with respect to file sharing (Eisenhardt 1989;Pettigrew 1990). Therefore, an analysis of the attitudes of international elite lawyers toward copyright infringement would provide us with a good window into the gaps between emerging socio-ethical attitudes and practices and legal norms (Posner 2009). Understanding how lawyers deal with and make sense of the dissonance between their own attitudes on ethics of filing sharing and legal norms would offer unique insights into the emerging social understanding of fairness in file sharing, which could open up new thinking for legal reform. (Ellickson 1998;Hart 2012;Jemielniak 2014b;Ullmann-Margalit 2015). To that end, this article presents the findings of a study addressing the key question: How do lawyers perceive file sharing practices from the standpoint of ethics?

Hypotheses
It is abundantly clear that the perceptions of fairness in copyright as well as practice are increasingly diverting from the letter of the law (Haber 2018) and the views of the lawyers (Litman 1996;Samuelson and Glushko 1989). This is especially true of people actively engaged in online content creation (Fiesler and Bruckman 2014), who also often engage in discussions about the ethical and legal aspects of copyright protection (Fiesler, Feuston, and Bruckman 2015). In these discussions, the point of view of lawyers as a group is often referred to by nonprofessionals (Fiesler, Feuston, and Bruckman 2014). Even though there is no systematic research of what lawyers think about copyright, their views on regulation are generally more conservative than the average (Luban 1988;Rotunda 2011;Tyler and Darley 1999). One of the reasons for this phenomenon is that the lawyers have highly regulated codes of ethics, requiring it to follow the letter of law in its daily conduct and to be generally law-abiding (Simon 2009;Zacharias 2007). In fact, for lawyers, breaking the law can lead to social exclusion and disbarment (Ogando 1991;Shaffer and Shaffer 1991). From a cross-cultural perspective, lawyers have also been found to be more conservative in their views on regulation (Abel and Lewis 1988). We therefore hypothesize that part of the reason why copyright protection has proven so difficult to reform, even in the face of growing popular support, is because legal practitioners themselves adopt relatively conservative positions with respect to the existing legal regime: Further, there is a widely shared assumption among judges, lawyers, and other legal professionals that public sector lawyers should be held to especially high ethical standards, precisely because they represent the public interest (Berenson 1999). In fact, these lawyers have a duty to acknowledge the public interest in their views on regulation (Dodek 2010). It is therefore likely that lawyers who orient themselves toward the public sector, or, alternatively, are socialized within the public sector will take a more progressive stance on copyright than their private counterparts.
This line of reasoning resonates strongly with the attraction-selection-attrition framework in management (Schneider, Goldstiein, and Smith 1995), which postulates that individuals join and leave organizations based on a match on enduring characteristics, such as preferences and values. Applied to the management of public organizations, this framework has yielded the concept of "public service motivation" (PSM). PSM has been defined by Rainey and Steinbauer (1999) as a "general altruistic motivation to serve the interests of a community of people, a state, a nation or humanity," and by Vandenabeele (2007) as "the belief, values and attitudes that go beyond self-interest and organizational interest, that concern the interest of a larger political entity and that motivate individuals to act accordingly whenever appropriate." It follows from this conceptual framework that "the greater an individual's PSM, the more likely the individual will seek membership in a public organization" (Perry and Wise 1990). Similarly in economics, Besley and Ghatak (2005) and Auriol and Brilon (2014) stress the difference between "profit-driven" and "mission-driven" organizations. The latter attract mission-oriented individuals who readily give-up high-powered monetary incentives schemes to improve the match between their own goals and values and that of the organization for which they work. In the remainder of the article, we loosely use the terms "public" and "private" sectors not to recoup the legal definition of the term, but to capture this important distinction between "profit-driven" and "mission-driven" organizations. 6 As explained above, at the very core of the concept of copyright lies a clear-cut contradiction between private interests (of the right holders) and public interests (of society at large), which requires both balancing and mediating (Joyce et al. 2016;Klein, Moss, and Edwards 2015). It is therefore likely that lawyers' ethical stance on copyright infringementprecisely because it is both highly loaded and polarized-has the potential to capture respondents' "public service ethos" and should predict their career choices. One advantage of our data in this respect is that not only can we test whether ethical attitudes correlate with respondents' sector of occupation, we can also say something about whether this occurs as a result of selection (i.e., pro-sharing individuals orient themselves toward the public sector) as opposed to learning (i.e., individuals develop pro-sharing attitudes as a result of their exposure to public sector work)-a significant challenge in the existing literature (Wright 2008;Wright and Christensen 2010). We therefore hypothesize that lawyers who orient themselves toward the public sector should differ with respect to how much they lean more on the side of broad disclosure, social sharing and the ethical acceptability of copyright infringement: H2: Lawyers who engage in public sector practice take relatively less conservative ethical positions on the issue of online copyright infringement.
Finally, legal ethics is also under the influence of local socio-economic standards and practices (Parker and Evans 2018). Attitudes toward file sharing have been argued to be significantly influenced both by culture (Sang et al. 2015) and income levels (Karaganis 2011). According to those early studies, it is possible that lawyers who differ with respect to their culture and/or the level of economic development of their home country will also differ in terms of their attitudes toward copyright infringement. Given the exploratory nature of this conjecture, we loosely formulate our third hypothesis: H3: Non-Western lawyers and/or lawyers coming from developing countries take different ethical positions on the issue of online copyright infringement.
We see this last hypothesis as expanding a descriptive line of work on individuals' perceptions of copyright law started by Mandel (2014), who argues that regardless of IP law's policy objectives, how people understand the law is crucial to the success of the IP system. This is true both for creators who obtain IP rights and IP consumers who may be regulated by those rights (Mandel 2017).
Following this reasoning, Mandel (2014) leverages a representative sample of the US population to explore laymen's intuitive understanding of copyright, also as a function of income, age, education, political ideology, and gender. The present article extends this approach to a sample of top international legal practitioners. 7

Research design
One of the authors took advantage of a visiting position at Harvard University from August 2015 to September 2016 to develop informal links with the Harvard LL.M. (Legum Magister, a master's degree in law) students from two consecutive cohorts. He did not teach any courses related to copyright or copyright law, or any courses taken by LL.M. students. He started the project by dedicating a whole semester to blending in informally: participating in the discussions of the closed Harvard LL.M. students' Facebook group, attending private parties in dorms by invitation from the group members, giving advice on living as a foreigner in Cambridge, organizing a running club, and, in one case, helping a student with furniture assembly. He followed a strategy of entering the field, that is, typical for ethnography (Agar 1980;Clifford and Marcus 1986), with some elements of digital ethnography (Jemielniak 2014a), focusing on gaining access and developing a social network (Feldman, Bell, and Berger 2003). As a result, he was perceived as a valuable external member of the group, even though he clearly communicated the fact that he was not a student, but a visiting research fellow.
Social connections and acceptance as a useful contributor to the group proved priceless when he reached out to all students participating in the Facebook group in the spring and fall of 2016 for interviews directed at exploring their ethical perceptions of digital files sharing practices. The response was very positive and nearly all contacted students agreed to participate. But then, logistics of in-person data collection-time constraints and scheduling problems both on the side of the students and the author-reduced the overall number of filled in survey questionnaires.
About 200 students enroll in the Harvard LL.M. program each year, out of which 50 and 59 were interviewed in the spring and fall, respectively, through one-to-one meetings. 8 All data were collected in one-on-one meetings in the vicinity of the Harvard campus, in a consistent setting. This approach was chosen to ensure that the respondents were limited to LL.M. students, and also to increase the level of engagement with the study (online questionnaires are notorious for obtaining negligent and un-reflexive responses). All 109 respondents were first reminded of the purpose of the research, given all relevant details regarding the measures taken to ensure their anonymity and privacy, and provided informed consent forms. They then filled a questionnaire presented in an anonymous window on the researcher's computer.
The survey consisted of 3 sections. The first section elicited respondents' age, gender, region of origin 9 , and current or envisioned occupation (for those who had no work experience). The last item was used to determine respondents' (envisioned) sector of professional practice (public or private): 1. Works in following places were classified as public sector work: local governments, judiciary, NGOs (non-governmental organizations), international agencies, international intergovernmental organizations, and academia. 2. Work in law firms and corporations and also work as independent attorney/barrister was classified as private sector work.
The second section listed 17 specific copyright infringement scenarios, designed so as to distinguish between the many possible real-world motivations for infringing IP (see Appendix A for the full list of questions.) Respondents had to rate the ethical acceptability of each infringement scenario on a Likert scale ranging from "very unacceptable" to "very acceptable." As a first approximation, the first step of our analysis does not distinguish between those various scenarios and motivations, and takes the average of respondents' ratings over all 17 cases as the dependent variable of interest. In a second step, we disaggregate respondents' answers and group the relevant scenarios in 5 infringement categories (see Appendix B for the list of questions falling under each category): 1. Infringement of copyright with a commercial purpose. 2. Infringement of copyright because of lack of legal access to the material. 3. Infringement of copyright because legal access to the material was not affordable. 4. Infringement of copyright to avoid payment. 5. Infringement of copyright for educational purposes.
It is important to note that the subjects were not primed to think about ethics through any particular philosophical angle, be it deontological (i.e., rights or duty-based), consequential (i.e., outcome-based) or virtue based (i.e., based on desirable characteristics of the agent performing the action). 10 We did not do such priming to avoid framing/demand effects (Zizzo 2010) and invite respondents' intuitive ethical judgment. 11 It is also important to note that this design rests on the assumption that respondents do not distinguish between different types of creative works when it comes to assessing the ethical acceptability of copyright infringement. It is consistent with the idea that all copyrightable works share important features. This should especially hold in our population of lawyers because in law all copyrightable works, irrespective of the medium, fall under a common legal regime. As this assumption is important to our design (i.e., we aggregate our survey questions according to the justification provided for infringing copyright, not the type of work considered), we put it to test, as described in Appendix D.
Finally, the third section of the survey relied on Likert scales to elicit potentially relevant control variables: 1. Respondents' personal values (Redding and Reppucci 1999): (i) left or rightwing in economic matters (e.g., state interventions, health care, education, pension system), (ii) left or rightwing in social matters (e.g., abortion, legalization of drugs, LGBTQ rights, gender roles), and (iii) level of religiosity. 2. Respondents' computer knowhow and expertise in copyright law.
The third section of the questionnaire was added to the second data collection batch, so that we only collected the corresponding variables for 59 respondents.

Results
The relevant criterion for ethical acceptability: No commercial purpose One of the main legal challenges associated with file sharing practices is the law doesn't distinguish thatwith the important exception of the "fair use" doctrine-between the varieties of circumstances in which copyright infringement occurs. Table 1 presents some summary statistics on lawyers' ratings of the overall acceptability of our 5 copyright infringement categories. We get a clear-cut picture: file sharing ranks relatively high in terms of ethical acceptability, irrespective of the stated reason for engaging in copyright infringement (on average 3.23 out of a 5 points scale, with a standard deviation (SD) of 0.80). The only notable exception is infringement of copyright with a commercial purpose, which ranks significantly lower in terms of acceptability (1.76 out of 5), with the difference being statistically significant at the 1% level with each and every other infringement category. 12 In contrast, the categories "infringe copyright because no legal access," "infringe copyright because cannot afford" and "infringe copyright for educational purposes" do not significantly differ from one another, be it practically or statistically. It is interesting to note that this finding suggests that our respondents did not judge the ethical acceptability of the presented scenarios according to the degree to which they fit the American "fair use" doctrine (or its equivalent in other countries), that is, according to which the presented scenarios are a more or less clear-cut instances of copyright infringement. Had this been the case, the category "infringe copyright because cannot afford" should have ranked significantly lower in terms of ethical acceptability. 13 Finally, while the category "infringe copyright to avoid payment" ranks lower than the above three categories by about 0.5 point out of a 5 points scale (statistically significant at the 1% level), this difference is strikingly low (less than half a SD on average) given the nature of the infringement scenario considered. It therefore appears that lawyers perceive file sharing practices as generally ethically acceptable-as long as individuals do not derive monetary benefits from them.
Not only is it surprising that lawyers' ethical attitudes vis-a-vis copyright infringement is in line with that of students and laypeople, we find it striking that the relevant criterion they identify for ethical acceptability-no commercial purpose-is consistent with that of piracy communities (Shore et al. 2001), where there was strong early evidence that charging for access elicited negative ethical judgments: "The adherence to a nonprofit mindset allowed pirates to portray themselves as non-criminals and, thus, morally acceptable. Those who made money were not, however, acceptable participants in the piracy subculture" (Holt and Copes 2010, 641). Similarly, "an act of file sharing may be classified into one of two broad categories: the activity of professionals who seek profit through copyright infringement or the selfless behavior of activists willing to make culture available to the community. The former will tend to evoke negative affect and ethical judgments, while the latter will typically be evaluated positively and induce feelings of gratitude. The key distinction between these two types is whether the individuals involved in piracy derive any direct material benefit from it" (Krawczyk et al. 2017, 652).
A clash of cultures? How ethical attitudes determine the selection of lawyers into private versus public sector practice Our most striking finding is the significant divide in attitudes on ethics of file sharing that appears between lawyers who operate in the "public" and "private" sectors. 14 Figure 1 presents the distribution of lawyers' attitudes toward copyright infringement on averagedistinguishing between those who work (or intend to work) in the public sector and those who work (or intend to work) in the private sector. We see that the distribution of lawyers' average attitudes toward copyright infringement differs significantly between the two groups. Public sector lawyers consider file sharing to be more acceptable than private sector ones (diff ¼ 0.35; p ¼ 0.02; N ¼ 100). Consistent with the above finding that lawyers overwhelmingly reject file sharing as ethically unacceptable if it is done with a commercial purpose, this difference in attitudes between public and private sector lawyers vanishes when one focuses on the acceptability of for profit copyright infringement (diff ¼ -0.06; p ¼ 0.97; N ¼ 109). Importantly, the data suggests that this difference in overall attitudes is mainly due to selection as opposed to learning. This means that lawyers with stronger preexisting pro-sharing attitudes tend to orient themselves toward the public sector-as opposed to developing those attitudes as a result of their exposure to public sector practice. This can be seen by focusing the analysis on the youngest students in our sample, that is, those who were less than 27 years old at the time of the study. These students are very unlikely to have had significant work experience prior to their Harvard LLM studies, be it in the public or in the private sector. Since preferences typically take years to evolve (Carlsson, Johansson-Stenman, and Nam 2014), finding a significant difference in attitudes on ethics of file sharing for young lawyers who only intend to work for "mission oriented" organizations constitutes evidence of value-based work selection.
We see that the observed difference in overall attitudes toward copyright infringement does replicate in this restricted sample (diff ¼ 0.60; p ¼ 0.04; N ¼ 37). Since Harvard lawyers typically do not choose to work in the public sector because of other more lucrative alternatives (Weisbrod 1983), we interpret this finding that the public sector oriented lawyers lean on the side of broad disclosure and social sharing as reflecting of a public goods oriented ethos (Loughlin and Tschorne 2016). This, in turn, incentivizes them to engage in public sector practice.

Regression analysis
Finally, we confirm our results by exploring the factors that drive the ethical acceptability of file sharing practices in a regression framework. Table 2 presents OLS estimates of the effect of our individual level variables on the ethical acceptability of copyright infringement, averaging the answers over the 17 scenarios that were presented to respondents. The overall fit of the model is relatively good, with an adjusted R 2 ranging from 0.07 in column (1) to 0.23 in column (3). We disaggregate our dependent variable according to the categories presented in Table 1 in Appendix B, without any major impact on our conclusions.
The fact that our respondents come from many different parts of the world (column (1) of Table 2)-a unique feature of our data-enables us to test for the impact of their region of origin on attitudes on ethics of file sharing. 15 Surprisingly, we find that the region of origin does not influence evaluations of ethical acceptability of file sharing, despite the fact that it is found to be an important factor in studies on piracy, albeit not focused on lawyers (Bender and Wang 2009;Rybina 2011). While interesting, this finding should certainly not be taken as the final word on this issue. First, the measure of geographic location that we could derive from our data was relatively coarse: we did not ask for the country of origin to protect anonymity of the respondents (some countries have only one representative in the LL.M. program in a given year). Second, our population of lawyers was, by definition, a highly internationalized and relatively affluent one, so that their views might not adequately reflect the socio-economic condition of the populations in their respective countries.
In the same vein, column (2) replaces our 7 regional dummies with a single dummy variable indicating whether the respondent comes from an economically developed region. 16 Here again, we observe that respondents who come from economically developing regions do not differ in terms of how much they think copyright infringement is acceptable.
In column (3), we add some control variables indicating whether the respondent identifies himself as rightwing in economic matters, rightwing in social matters, religious, or technically knowledgeable about computers and technology. 17 Lawyers who self-identify as "leftists" in social matters are a lot more flexible in terms of copyright infringements, with each point increase in the leftwing social orientation scale (out of 5) being associated with half a SD increase in acceptability. 18 Finally, column (4) controls for respondents' selfreported level of expertise in copyright law. There is a great deal of variation along this dimension within our sample. 19 Perhaps surprisingly, we find that a higher level of expertise in copyright law is associated with a significant increase in the ethical acceptability of file sharing-a result consistent with that of Mandel (2014), who leverages a broader sample of 1700 U.S. adults and finds that individuals who report having more experience with copyright law (e.g., working as an attorney, paralegal, creator of copyrighted work) tend to desire weaker copyright protection. 20 As we have seen from Table 1, the ethical acceptability of file sharing is relatively high on average within our population. Even then, looking across all four columns in Table 2, our regression analysis confirms that lawyers who work (or intend to work) in the public sector consider file sharing to be significantly more acceptable-roughly amounting to half a SD increase in overall acceptability across all specifications. Interestingly, we see that gender plays no role in driving respondents' ethical assessment of copyright infringement practices. The same goes for age, as the associated coefficient loses statistical significance and drops to a value close to zero when more control variables are included in the model (see columns (3) and (4)).
We end this section with a brief discussion of Tables C1-C5, where we disaggregate our main  dependent variable from Table 2 according to the categories presented in Table 1: (i) "infringe copyright with a commercial purpose," (ii) "infringe copyright because no legal access," (iii) "infringe copyright because cannot afford," (iv) "infringe copyright to avoid payment," and (v) "infringe copyright for educational purposes." Disaggregating our dependent variable in such a way does not significantly alter our empirical conclusions, so we do not report these tables in the main text. A few additional insights are nonetheless worth considering: 1. Consistent with our results presented in Table 2, respondents' region of origin never impacts their assessment of the ethical acceptability of file sharing, except when it is done with a commercial purpose. In this case respondents from economically developed countries consider file sharing to be significantly more acceptable than respondents from developing ones, which makes for a  (3) and (4): rightwing_econ, rightwing_social, religious, computer_xp, and copyright_xp are all reported on a 5 points scale.
counterintuitive result (see columns (2) and (3) from Table B1). 2. The difference in ethical attitudes between practitioners from the public and the private sector vanishes in two instances: (i) when copyright is infringed with a commercial purpose, in which case acceptability is relatively low in both groups (Table B1), and (ii) when copyright is infringed because there is no legal alternative available, in which case acceptability is relatively high in both groups (Table B2). 3. Compared with the estimates reported in Table 2, the magnitude of the difference in attitudes on ethics of file sharing between public and private lawyers increases when copyright is infringed specifically (i) to avoid payment (Table B3), (ii) for educational purposes (Table B5), and most strongly (iii) because one cannot afford to pay, where operating in the public sector is associated with a one SD increase in the assessed ethical acceptability of the practice (Table B4).

Summary of results
Because of its social function, legal practice is often viewed as a relatively conservative occupation (Freidson 2001;Jemielniak 2005). Therefore, we expected lawyers to take highly conservative ethical positions on the issue of online copyright infringement (H1). This hypothesis is strongly rejected by our data. We find that digital file sharing ranks relatively high in terms of ethical acceptability among our population of lawyers-with the only notable exception being infringing copyright with a commercial purpose. Moving away from the previous focus on laymen and undergraduate students, it therefore appears that the lawyers themselves perceive file sharing practices as ethically acceptable, as long as individuals do not derive monetary benefits from them. Those relatively liberal views are unlikely to be affected by income effects, as may be the case in many previous research settings ( Rob and Waldfogel 2006;Smallridge and Roberts 2013). As a matter of fact, most lawyers in our sample are in the top echelons of income in their home countries (Wilkins, Fong, and Dinovitzer 2015). Attitudes toward file sharing have also been argued to be significantly influenced by culture (Sang et al. 2015) and income levels (Karaganis 2011). As a result, we expected non-Western lawyers and/or lawyers coming from developing countries to take different ethical positions on the issue of online copyright infringement (H3)-although given the exploratory nature of this test, we did not have a clear prediction as to the direction of the effect. This hypothesis is also not supported by our data. Instead, we find that lawyers' region of origin (defined both geographically and in terms of level of economic development) has no consistent impact on their ethical views vis-a-vis file sharing. One obvious limitation of our study design in this respect is that most of the lawyers in our sample come from highly privileged socio-economic environments, so that their views might not accurately reflect those of the populations in their respective countries. At the very least, our results call for a more systematic investigation of the geographical distribution of attitudes toward copyright protection.
The only hypothesis which was strongly supported by our data was H2. To be sure, the current ideological struggle over the reform of copyright is largely the reflection of the highly loaded and polarized question of how to strike an efficient balance between the contradicting goals of incentivizing creative works on the one hand, and letting society benefit from their wide diffusion while preserving user privacy (Bakioǧlu 2016) on the other. This polarization is likely to embed itself in the organizational cultures of "profitdriven" and "mission-driven" organizations, respectively (Besley and Ghatak 2005). As a result, we hypothesized that lawyers who work (or intend to work) for the latter (which we loosely refer to as the "public sector") would be significantly more tolerant of online copyright infringement. Beyond providing unambiguous support for this hypothesis, our data also suggests that this phenomenon is largely the result of self-selection: young lawyers who lean more on the side of broad disclosure and social sharing tend to orient themselves toward the public sector, while the reverse holds for those who are more defensive of copyright. Such an attitudinal divide might further hinder copyright reform in the face of increasing economic and social evidence of its inadequacy.

Study limitations
Three limitations of our research protocol need to be kept in mind while interpreting our results. The first one relates to our research sample. While our focus on Harvard LL.M. students provides a unique opportunity to study the ethical attitudes of a top-notch sample of (otherwise geographically dispersed) legal practitioners in a consistent way, only 37% of our respondents rated their expertise in copyright law as greater of equal to 3, out of a 5 points scale. Therefore, while our all of our respondents can be described as highly qualified lawyers, not all of them are copyright experts. Note, however, that our results hold when we control for expertise in copyright law in our regressions. Together with the fact that expertise in copyright law is actually associated with more favorable ethical assessments of file sharing practices, this suggests that our results are not driven by a lack of expertise in copyright law.
The second limitation relates to our finding that respondents' region or origin and its associated level of economic development do not seem to influence their ethical perceptions of copyright infringement. Given the coarse nature of our geographical measures, this result should be considered as a first step toward a more conclusive test of our hypothesis, which would warrant some additional research efforts.
Third, the explicit focus of this article is on the factors that affect respondents' perceptions of the ethical acceptability of file sharing on the "demand side" (i.e., their demographics, or the intended use of the good). As a result, our design largely abstracts from the "supply side" of the ethics equation. For instance, no scenario presented to subjects provided details on whether and how the content creator relied on sales to sustain his activity. Our results are therefore best thought of as "maintaining the supply side of the equation fixed." In this respect, we assume that, all else equal, the more content creators depend on sales to sustain their activity, the more infringing their copyright rights will be considered unethical by respondents (provided that this fact is made salient to them).

Concluding remarks
Almost a decade after the Sony BMG Entertainment vs. Tenenbaum case in which student Joel Tenenbaum was judged liable for $67,500 in damages for downloading 30 songs via peer-to-peer networks-and following a significant restructuration of the creative works industry-evidence has accumulated that digital file sharing practices do not significantly displace revenue from copyright holders. Also, research has shown that younger generations as well as the general population increasingly consider file sharing practices as socially and ethically acceptable. Yet, the legal regime governing copyright protection has not seen any major evolution over the same period.
Given the prominent position that lawyers hold in politics and policy making (Manning 2017;Posner 2009;Winterhalter 2013), it may be surprising that such a wide gap could survive between the letter of the law, on the one hand, and generally held ethical views and also economic evidence that file sharing does not significantly displace copyright holders' revenue, on the other. Moreover, the deterrence strategies rely on raising enforcement to levels that may have further undermined support for the existing copyright regime (Depoorter and Van Hiel 2015;Depoorter, Van Hiel, and Vanneste 2010;Sinha and Mandel 2008). We suggest that part of the answer to this puzzle might be found in the significant attitudinal divide between lawyers who operate in private and public sector organizations-it might be a factor hampering negotiate a new balance between public and private interests in copyright law.
On a more general level, the fact that even the international elite lawyers perceive digital file sharing as generally acceptable signals that policies are increasingly misaligned with social practices. In this respect, it is striking that lawyers in our sample overwhelmingly used a non-commercial-purpose criterion to determine whether a file sharing practice was ethical. The same criterion had been put forward by the earliest proponents of the reform of copyright, such as the Pirate Party, to define the legal boundaries of a sphere of "non-market exchange" that would not be subject to copyright (Aigrain 2012). When lawyers and pirates concur in terms of their ethical assessment of file sharing practices, the legal status quo appears to be more of a tool for advancing informational capitalism (Zukerfeld 2017) than reflecting everyday practices of common sense and fairness perception. These findings support the calls for further de-criminalization of copyright legislation (Haber 2018).

The economic inefficiencies and the social injustice
brought about by the current institutional approach to cultural and scientific production are also discussed in areas of intellectual property law other than copyright. This is notably the case for patent law, which, if it does incentivize costly investments in research and innovation, does not ensure equality of access to medicine for all citizens, both in developing (Kapczynski 2009) and developed countries (Brennan et al. 2016). This raises a set of ethical challenges that is arguably distinct from that of copyright, in the sense that patent protected scientific innovations typically require upfront investments an order of magnitude higher than any copyrightable creative works (even the most expensive Hollywood blockbuster). At the same time, the stakes are significantly higher for society. In the case of drugs, the lives of many individuals rest on our collective ability to generate and diffuse significant product innovations.
2. By doing so, we explicitly focus on the factors that affect respondents' perceptions of the ethical use of file sharing on the "demand side" (i.e., their demographics, or the intended use of the good), as opposed to the "supply side" (i.e., does the content creator rely on sales to sustain content production). In this respect, we assume that, all else being equal, the more content creators depend on sales to sustain their activity, the more unethical infringement of their copy rights will be considered by the respondents (provided that this fact is made salient to them).
3. In the case of the U.S. constitution, this need for a balanced approached led Thomas Jefferson and James Madison to agree on the formulation that Congress could "promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries" (Article 1, Section 8).
4. Since classifications differ across legal cultures, for the purpose of this article we assume that the term "lawyer" covers anyone in the Harvard LL.M. program, and includes judges, attorneys, solicitors, inhouse lawyers, etc. Many students already have some professional experience, many are on sabbaticals from a position they are going to return to, and many have been admitted to the bar.
6. As Besley and Ghatak (2005, 616) note,"not all activities within the public sector are mission-oriented. For example, in some countries, governments own car plants.
[ … ] Not all private sector activity is profitoriented. Universities, whether public or private, have many goals at variance with profitmaximization." In this article, when we write that a lawyer working for an NGO or a university operates in the public sector, we do not mean it in a technical sense, but, following Besley and Ghatak (2005), organizations with a strong mission-orientation.
7. See also Redding and Repucci (1999) for a study on how lawyers' socio-political attitudes affect their judgments and perception of policies.
8. Two-sided t-tests provide no statistical evidence of difference in respondents' ethical perceptions of copyright infringement depending on whether they were studied in the spring or in the fall.
9. The research design intentionally did not include questions about the country of origin, since this information would reveal the identity of some respondents.
10. For a review of the theories of ethics, see, for example, Thiroux and Krasemann (2012).
11. Of course, the drawback of this data collection strategy is that it does not allow us to unpack respondents' thinking process in addressing the ethics question, which we leave open for future research.
12. The significance levels we report in this section are based on two-sided t-tests.
13. Four factors need to be jointly considered in order to determine whether the use made of a work falls under the "fair use doctrine": (i) the purpose and character of the use, including whether such use is of a commercial nature or is for nonprofit educational purposes, (ii) the nature of the copyrighted work, (iii) the amount and substantiality of the portion used in relation to the copyrighted work as a whole, and (iv) the effect of the use upon the potential market for or value of the copyrighted work.
14. Remember that, in this article, we use the term "public sector work" quite loosely to refer to work in "mission-oriented" organizations.
15. Our sample contains 25 law students from Europe, 5 from North America, 20 from South America, 7 from the Middle East, 38 from Asia, 5 from Australia and New Zealand, 2 from the Caribbean, and 7 from Africa.
16. We classify the following regions as economically developed: Europe, North America, Australia-New Zealand.
17. We only thought of eliciting responses for these control variables in the middle of our questionnaire administration procedure, so that our sample size drops significantly when we include them in columns (3) and (4).
18. This economically large effect could be predicted from our interviews, as many "leftwing" respondents used the image of "big rich corporations" in the context of justifying some form of copyright infringement.
19. Specifically, 22% of respondents report a level of expertise in copyright law of 1 out of 5, 33% report a level of 2, 16% a level of 3, and 8% a level of 4 and 5, respectively. No respondent reports a level of copyright expertise of 0.
20. In informal conversations about these results, a number of copyright experts described this body of law as typically "complex", "inefficient" and sometimes "extractive." They were therefore only slightly surprised-if at all-by the fact that lawyers who are more knowledgeable about copyright law also felt that is was more acceptable to break it on ethical grounds.

ORCID
Dariusz Jemielniak http://orcid.org/0000-0002-3745-7931 X frequently downloads current movie premieres from the Internet (torrents). X does not watch these movies later in the cinema. X frequently downloads music from the Internet (torrents). X could have legal access to it, but X would have to pay.
X frequently downloads fiction ebooks from the Internet to avoid payment. 5. Infringement of copyright for educational purposes.
Average rating for the following questionnaire item: X frequently downloads academic books from the Internet to avoid payment.
Appendix C. Regression results by copyright infringement categories in Table 1   Table C1. Ethical acceptability of file sharing-for profit.
Appendix D. Do ethics of file sharing depend on the type of digital file shared?
The main body of the article focuses on understanding lawyers' attitudes on ethics of file sharing as per the rationale presented for infringement of copyright (i.e., the demand side) and leaving aside the potential impact of infringement on the dynamics of content creation (i.e., the supply side). While conducting this exercise, we assume that, when it comes to the ethical acceptability of copyright infringement, respondents treat different types of works similarly, so what really matters for assessing a particular case of infringement is not the kind of creative work considered, but its intended use. We make this assumption because copyrighted works share some important characteristics that justify their unification under a common legal regime, which our population of lawyers should be able to readily recognize (Landes and Posner, 1989;Leaffer 2010). First, in the case of (non-scientific) creative works of the mind (stories, songs, drawings), once the initial cost of production is made, copies can be distributed at relatively much lower costs and near zero cost for digital ones. Second, the purpose of copyright is to protect the expression of the author's creativity-a unique arrangement of words, musical notes, etc.-as opposed to the creation itself (i.e., the "new idea," which can only be protected in the case of scientific innovations by securing a patent). We tested this implicit assumption by examining whether attitudes on ethics of file sharing differ based on the type of creative works considered. To do so, we leverage the responses provided to questions 3, 5, 7, 9, and 13 in our survey (see Appendix A for the full list of presented scenarios). Table D1 provides the data on justification "avoid payment"-and reports averages of ratings based on the kind of creative works considered.
We can see from the table that our assumption is largely holds: on a 5 points scale, there are no practically significant differences in the reported ethical acceptability of copyright infringement across different types of media-TV shows, movies, music, and fiction books. (We did not ask this question with respect to software, unfortunately.) Consistent with the approach we take in the article, it is interesting to note that significant differences in ethical assessments arise again when one adds the "educational purpose" rationale to the last copyright infringement scenario presented in Table D1. This is done in survey question 14, which is different from question 13 only in the sense that it considers the case of copyright infringement for academic books (as opposed to fiction books). The average ethical acceptability in this case is 3.28, compared to the value of 2.71 in the case of regular fiction books (diff ¼ 0.53; p ¼ 0.001). Again, we take this as suggestive evidence that "intended use" dwarfs the "type of file considered" criterion when it comes to assessing the ethical acceptability of copyright infringement.  (3) and (4): rightwing_econ, rightwing_social, religious, com-puter_xp and copyright_xp are all reported on a 5 points scale. Table D1. Averages of ratings for questionnaire items related to the justification "avoid payment" based on the kind of creative works considered.