CC BY
0
Эпистемология и философия науки 2024. Т. 61. № 3. С. 51-61 УДК 167.7
Epistemology & Philosophy of Science 2024, vol. 61, no. 3, pp. 51-61 DOI: https://doi.org/10.5840/eps202461339
Response to turner:
is it useful and/or accurate to think of algorithmic curation as epistemic coercion?
Gil Eyal - Professor of Sociology. Columbia University. 116th and Broadway, New York, NY 10027, USA;
e-mail: [email protected]
Elizaveta Sheremet -
PhD Student and
Paul F. Lazarsfeld Fellow.
Department of Sociology,
Columbia University.
116th and Broadway, New York,
NY 10027, USA;
e-mail: [email protected]
We argue that the concept of epistemic coercion is neither accurate nor useful for describing and thinking about the significance of the new practices of algorithmic curation, and that Foucault's concept of rarefaction is better suited for this purpose. After establishing what Turner means by epistemic coercion, we show that it differs from how the concept of coercion is typically defined and used by philosophers and sociologists, especially because Turner does not identify a threat that causes the coerced people to act under duress. We then detail our reasons for why the concept of coercion, to our minds, flattens and to some extent distorts our understanding of the practice of curation. Among these reasons are the blurry lines and interdependence between curation and self-curation, thus between "coercion" and "resistance"; the absence of a plausible "threat" that could justify conceptualizing the operation as coercion; the inescapability of curation in order to navigate the "information glut"; as well the question of whether users of social media are aware that their information environment is curated. Finally, and directly following from these reasons, we show that Foucault's concept of discursive "rarefaction" offers a lot more insight into the novelty and nature of contemporary curation practices. Indeed, we argue that viewed from this perspective, social media appears to represent a new rearrangement and ordering of discourse, the formation of an interface between "ordinary utterances" and "disciplines," between everyday talk and expert discourse. This intermediate realm, where discursive events are neither ephemeral nor preserved "in the true," depends on rarefaction-qua-curation for its existence and functioning.
Keywords: curation, algorithmic curation, discourse, Foucault, discourse analysis, rarefaction, coercion
Ответ тернеру: полезно/корректно ли
рассматривать алгоритмическое курирование как эпистемическое принуждение?
Хиль Эяль - профессор социологии.
Колумбийский университет. 116th and Broadway, Нью-Йорк 10027, США; e-mail: [email protected]
Мы утверждаем, что концепция эпистемического принуждения не является ни точной, ни полезной для описания и осмысления значимости новых практик алгоритмического кураторства и что концепция разреженности Фуко лучше подходит для этой цели. После прояснения того, что Тернер подразумевает под эпистемическим принуждением, мы показываем, что трактовка Тернера отличается от обычного определения и использования концепции принуждения философами
© Gil Eyal, 2024 © Elizaveta Sheremet, 2024
Елизавета Шеремет - и социологами. Дело в том, что Тернер не определяет угрозу,
аспирантка. которая заставляет людей действовать по принуждению. Да-
Кафедра социологии лее мы подробно объясняем, почему концепция принужде-
Колумбийского ния упрощает и в некоторой степени искажает наше понима-
университета. ние практики кураторства. Среди причин этого - размытые
116th and Broadway, границы и взаимозависимость между кураторством и самоку-
Нью-Йорк 10027, США; раторством, а значит, между «принуждением» и «сопротив-
e-mail: [email protected] лением»; отсутствие правдоподобной «угрозы», которая могла
бы оправдать концептуализацию действия как принуждения; неизбежность кураторства для навигации в «информационном переизбытке»; вопрос о том, осознают ли пользователи социальных сетей, что их информационная среда курируется. Исходя из этих соображений, мы показываем, что концепция «разрежения» дискурса, выдвинутая Фуко, позволяет гораздо лучше понять новизну и природу современных практик кураторства. Мы показываем, что с этой точки зрения социальные медиа представляют собой новую перестройку и упорядочение дискурса; они формируют интерфейс между «обыденными высказываниями» и «дисциплинами», между повседневной речью и экспертным дискурсом. Эта промежуточная область, в которой дискурсивные события не являются ни эфемерными, ни неизменно «истинными», в своем существовании и функционировании зависит от разреженности-как-кураторства.
Ключевые слова: кураторство, алгоритмическое кураторство, дискурс, Фуко, дискурс-анализ, разреженность, принуждение
While Turner dedicates a large part of his article to science, it is evident that the impetus for formulating the concept of epistemic coercion comes from his reaction to what he calls "curation." The advent of digitalization and social media, especially as the latter is increasingly algorithmically managed, argues Turner, creates new affordances to control the cognitive environment of users. Curation, according to Turner, is a type of technical intervention that enables "unobtrusive" coercion and makes us particularly vulnerable precisely because it operates "in the course of [us] doing something else, such as browsing social media or searching for information where we are unaware of what is being withheld, promoted, or presented in a context designed to make it more plausible" [Turner, 2024, p. 36]. Put differently, algorithmic curation grafts itself onto the free and seemingly autonomous practices of the users of social media, creating the "illusion of freedom" for them [Ibid.]. Its manipulation of their attention and information environment constitutes a new form of control exercised by an "unknown and unseen" manipulator, making it especially hard to detect and resist. As a key feature of the current "epistemic situation of pervasive digitalization and social media" [Ibid., p. 24], Turner presents curation as exhibit one for the relevance and utility of the concept of epistemic coercion, and for the need to come up with a new understanding of the power that pervades the relations of information, knowledge, and discourse.
Turner deserves credit for calling attention to the role that curation plays in shaping a "new epistemic situation," and for insisting on the need to develop concepts that can illuminate its workings and effects. To our minds, however, the concept of coercion is neither accurate nor useful for this purpose. We will develop this argument as follows: first, we will show that Turner does not provide a useful working definition of coercion, and that there is textual evidence that the primary purpose of the concept is polemical. Second, we will provide a brief survey of what is typically meant by "coercion" (including an independent usage of the very same term - "epistemic coercion"), noting the points of difference from Turner's usage. Third, we will detail our reasons for why the concept of coercion flattens and to some extent distorts our understanding of the practice of curation. Among these reasons are the blurry lines and interdependence between curation and self-curation, thus between "coercion" and "resistance"; the absence of a plausible "threat" that could justify conceptualizing the operation as coercion; the inescapability of curation in order to navigate the "information glut"; as well the question of whether users of social media are aware that their information environment is cu-rated. Finally, and directly following from these reasons, we will suggest that the concept of discursive "rarefaction," introduced by Foucault (1970) in his inaugural lecture at the College de France, as well as in The Archeology of Knowledge (1972), offers a lot more insight into the novelty and nature of contemporary curation practices.
What Does Turner Mean by "Epistemic Coercion"?
The word "coercion" appears 77 times in Turner's article, but search as we may, we couldn't find a definition. This is not, by itself, a fault. We do not fetishize definitions. Per Wittgensteinian "family resemblance," the meaning of a term can be clarified through a set of strategic contrasts that do not sum up to necessary and sufficient conditions. This seems to be Turner's approach. Epistemic coercion is contrasted with plain coercion that involves "commands and enforcement," with rational persuasion, and with hegemonic power [Turner, 2024, p. 25]. The upshot seems to be that by epistemic coercion Turner means "limits on thought and behavior that are not even recognized as such," and that result from the fact that certain speakers and certain contents are excluded from discourse. [Ibid.]
The difficulty, which Turner recognizes, is that all discourse, and certainly scientific discourse, rests on a set of exclusions. The distinction between epistemic coercion and rational persuasion, he says, is "meaningful only at the margins. There is an element of power in persuasion, and of persuasion in power." [Ibid.] If discourse was a free-for-all, it
would be a cacophony and nothing would be accomplished. He cites Polanyi to the effect that "excluding unreasonable objections and incompetent persons may be a requirement of progress," and that there is simply a need to "calibrate the inclusions and exclusions for effectiveness." [Turner, 2024, p. 27] If so, why single out particular practices -such as curation or peer review - as instances of epistemic coercion? Turner does not give a clear answer. He seems to imply that there are some instances when this coercion is "appropriate" because it serves the goals of scientific discovery, but others when it is not, because it serves some ulterior motives. [Ibid.] But he provides no way of deciding when this happens.
Most importantly, his characterization of this form of power as epistemic coercion seems to rest on the idea that the exclusion involved is unrecognized. If it is recognized, it might still be coercion, but no longer an epistemic one. When "pervasive conditions of constraint... are unconsciously internalized as normal and then serve as self-imposed limits on thought and behavior that are not even recognized as such" [Ibid., p. 25], then this has "epistemic consequences. For our purposes, then, these are forms of epistemic coercion." [Ibid., p. 27] This is far too vague since he doesn't tell the reader whether "unrecognized" means that it is impossible for ordinary people to recognize these limits (but somehow possible for the critical scholar), or that they are only temporarily deceived, or that they are unrecognized only as a condition of successful practice (in the same way that riding a bicycle requires backgrounding the explicit knowledge of how to ride a bicycle), or maybe it is a form of motivated misrecognition, namely actors know that their information environment is curated, but go along with the curation because they believe that it serves their interests?
In short, Turner's concept of "epistemic coercion" seems to rest on an unclarified normative criterion of the "appropriateness" of exclusion, and on an unclarified empirical question about whether and in what sense are people unaware of what is being excluded. Given these difficulties, we suspect that Schmitt's epigraph applies also to this concept. It is "incomprehensible" unless as a polemical concept directed against others who are "to be affected, combatted, refuted, or negated by such a term." [Ibid., p. 22] From this perspective, using the term "coercion" has the primary function of labeling certain practices illegitimate and incompatible with the principles of a liberal society.
What do Philosophers and Sociologists Typically Mean by "Coercion" and by "Epistemic Coercion"?
In contrast to Turner, most discussions of "coercion" in philosophy and social science link its occurrence to the idea that a certain threat is issued in order to compel behavior. This is true both for authors who seek a "precising" definition, limiting the application of the concept, and for those who, like Turner, seek to broaden its scope. It should be self-evident that if coercion depends on a threat being perceived, it cannot operate if "unrecognized."
According the Stanford Encyclopedia of Philosophy [Anderson, 2023], coercion is (a) distinct from mere disapproval or/and emotional manipulation, (b) exercised by an application of force or a threat of it, and (c) includes the intentional attempt of a coercer to alter actions of others. This approach establishes clear boundaries around the concept, thereby concentrating analytic attention on specific instances and contexts where coercion is most salient, such as law enforcement, governmental control, sexual and domestic abuse, etc. A related approach is Weber's [Weber, 1978, pp. 212-215] concept of domination as a form of power that operates through commands. Commands are obeyed, says Weber, for a variety of reasons, including the threat of force, ulterior motives or self-interest, as well as genuine belief that the command is legitimate. Commands are, of course, explicit and need to be recognized by the subordinate in order to have effect. Thus, most standard approaches to coercion would exclude Turner's usage and would consider the idea of epistemic coercion - a coercion that operates without recognition or awareness by those subject to it - self-contradictory.
We have found at least one instance of an author who attempts to broaden the definition of coercion and expand its application also to instances characterized as "epistemic coercion." Dandelet [2021] proposes a concept of epistemic coercion building on the ideas of J.S. Mill about the coercive potential of public opinion. In Dandelet's framework, coercion still involves issuing an implicit or explicit threat, but it becomes "epistemic" if the threat modifies how the threatened individual structures their epistemic inquiries. This conceptualization captures phenomena such as self-gaslighting, where a victim of sexual abuse may alter their perception of their own experiences due to the threat of skepticism from others. The threat of being labeled a liar or not mentally stable creates a social pressure that ultimately influences individuals' own perception and memory. Dandelet's framework thus extends the concept of coercion beyond its traditional boundaries, illuminating the subtle yet impactful ways in which coercion operates within social contexts. We think it probably captures also the example that Turner gives of professions of faith that are coerced by forced repetition until they become
internalized [Dandelet, 2021, p. 3], but it is also clear that this meaning of epistemic coercion is not of central interest to Turner and is not relevant to the key test case of curation.
Reasons Why Epistemic Coercion Is an Inaccurate and Unproductive Way to Characterize Curation Practices
To evaluate Turner's assertion that algorithmic curation constitutes a form of epistemic coercion, it is necessary to examine, however briefly, what is involved in curation. We draw on Davis [2017], who offers a theoretical framework for analyzing social media curation. Several points made by Davis are especially pertinent to this question.
First, self-curation is a massive and inescapable phenomenon. "Through digital media, people curate both who they are and what they consume." By the same token, they are constantly subject to the "curatorial efforts" of other individuals. [Ibid., pp. 771-774] Second, this massive interplay between productive and consumptive curation takes place within boundaries set by "curatorial code," namely "platform architecture and algorithms... through which users are encouraged or alternatively prevented from producing and consuming in particular ways." [Ibid., p. 776] In short, curation refers to a vast set of activities, of which the operation of algorithms that make certain posts more or less discoverable for certain individuals is a relatively small portion. Third, while some mechanisms underlying algorithmic curation are known, such as prioritizing image-based posts over text-based ones, tailoring content based on user preferences and past behavior, or restricting the reach of posts violating the platform's policies, much of the process remains "black-boxed." As a result, "neither producer nor consumer can fully predict which pieces of content will stand out or alternatively, pass in quiet obscurity," or what audience will they reach. [Ibid., pp. 777-778]
This brief description of social media curation suggests to us that it is inaccurate to describe the operation of curation neither as epistemic coercion of consumers of online information, nor as coercion of producers of online information.
The crux of Turner's argument is about consumption. Being prevented from reading certain categories of posts, he argues, should count as an instance of epistemic coercion. The problem with this argument is that given the overwhelming volume of information generated and shared on social media platforms, coupled with our limited cognitive capacities, the mechanism of curation is ubiquitous, inescapable, and absolutely necessary in order to navigate what has been termed the "information glut." In this context, the absence of curation could potentially be just as
coercive, completely overwhelming consumers' abilities to absorb and evaluate any information. As we learned from Davis, self-curation is the primary mechanism by which this reality is handled. While the curatorial code sets limits to the primary mechanism of self-curation, there is no reason to assume that consumers are unaware of algorithmic curation. This means that one cannot disregard curatorial agency when assessing the operation and impact of the curatorial code. Users may choose to go along with the choices made by the code, adopting them as continuous with their project of self-curation. They may also seek to "teach" or "game" the algorithm, since much of algorithmic curation relies on analyzing users' past behaviors and engagements with content. Or they may consciously resist the limits imposed by the curatorial code by diversifying their sources, or by devising strategies to circumvent existing restrictions. The main point is that the curatorial code does not confront individual curatorial strategies as an external limit, but the two are intertwined in complex ways. Self-curation strategies rely on and are continuous with the affordances of the curatorial code, while the code depends for its operation on the actors' interest in self-curation. In short, the boundary between coercion and resistance is blurry, as becomes clearer later in Turner's article when resistance is presented as essentially a project of self-cura-tion, and the "means of resistance are kin to the means of coercion." [Turner, 2024, p. 34]
By seemingly treating all curation of social media as potentially epis-temic coercion, Turner, just as he criticizes recent discussions on epis-temic injustice for having a hidden "ideal theory" of communicative situation, seems to have a hidden ideal theory of the circulation of discourse suggesting a free, unrestricted, unstructured information environment. The corollary of this Feyerabendian "anarchism" is a tendency to treat consumers as "information dopes" suggesting a lack of agency and autonomy while overlooking their ability to navigate curated content, and indeed the primacy and necessity of self-curation.
While the crux of Turner's argument is about the conditions surrounding the consumption of information, the plausibility of characterizing curation as epistemic curation relies, as we saw earlier, on the idea that what is being internalized is an inappropriate constraint on speakers and content, i.e. it is an argument about curation as coercion exercised over the production of information. This argument too is unconvincing. The concept of coercion, as we saw, requires the idea of a threat being communicated. But awareness by users of algorithmic curation does not rise to the level of perceived "threat" that could justify conceptualizing the operation of the curatorial code as coercion. As we noted earlier, the black box nature of the algorithm means that individuals cannot fully predict how their posts will be treated by the curatorial code. Moreover, even when they can predict, namely when curation rules are explicit, such as those aimed at restricting harmful content, the effect of the rules is not
to sanction users but to modify the likelihood of their posts reaching particular audiences. The ensuing dynamic is better understood not as coercion but as "strategic interaction" and the sort of "covering" and "uncovering moves" analyzed by Goffman as "expression games" [1969]. Arguably, the fear of one's account being suspended, which is the most severe sanction in the toolkit of algorithmic curation, can be thought of as a threat and thus a form of coercion. Yet, not only is it rare and hard to enforce, this sanction too merely modifies the half-life and reach of posts. What is involved is not a restriction on one's freedom of expression, but merely a limit on how widely and how long one's statements circulate in the public sphere. We emphasize this not only because it is questionable whether the concept of coercion should be stretched to encompass such outcomes, but also because it will be central to our argument below about curation as rarefaction.
Curation as Rarefaction
When Michel Foucault gave his inaugural address at the College de France [1970/1972], nothing could have been further from his mind than algorithmic curation, a practice that did not yet exist, not even as a twinkle in the eye of a young programmer. And yet, we believe that Foucault's concept of rarefaction, elaborated in this lecture, following the development of it in The Archeology of Knowledge [1969/1972], and his account of the different mechanisms and strategies of discursive rarefaction, can be especially illuminating when considering algorithmic curation.
Early into the inaugural lecture, Foucault famously advanced an "hypothesis... to fix the terrain. in which I shall be working," namely that "in every society the production of discourse is at once controlled, selected, organized, and redistributed according to a certain number of procedures whose role is to avert its powers and dangers, to cope with chance events, to evade its ponderous, awesome materiality." [Foucault, 1972, p. 216] Some of these procedures are "rules of exclusion. [concerning] what is prohibited." They operate "on the exterior" of discourse, so to speak. The bulk of the lecture, however, was dedicated to "internal rules, where discourse exercises its own control." [Ibid., p. 220] It is with respect to the different systems of internal rules that Foucault introduces the concept of "rarefaction."
While systems of rarefaction can take different forms - Foucault talks of commentary, the author, and discipline as three distinct systems -they share several crucial characteristics that we discuss below, pointing out their relevance for understanding algorithmic curation.
First, rarefaction is "involved in the mastery of. events and chance." [Ibid.] An event is something that happens once and will not be repeated,
unless work and energy are expended in preservation, dissemination, and repetition. Foucault suggests that we treat statements as discursive events, and that "the analysis of statements. is a historical analysis" tasked with explaining "what it means for them to have appeared when and where they did - they and no others." [Foucault, 1972, p. 109] In this respect, he suggests a "gradation between different types of discourse in most societies." On the one hand there are "ordinary" discourses that exist momentarily and vanish quickly: "uttered in the course of the day and in casual meetings. [they] disappear with the very act that gave rise to it." [Ibid., p. 220] On the other hand, there are discourses that are "spoken and remain spoken, indefinitely, beyond its formulation, and which remain to be spoken." [Ibid.] In these discourses - typically religious, literary, juridical, and scientific discourses - operate certain mechanisms of rarefaction that select statements to be preserved, reproduced, regulated, reactivated, and disseminated such that they have a lasting presence and influence over time on future discourse and knowledge production. This should sensitize us to realize that social media platforms are sites of a novel form of discourse, a sort of middle ground between those momentary "ordinary" utterances that Foucault thought were destined to disappear immediately and thus require no organized system of rarefaction, and those statements carefully selected - shall we say "curated" - to become lasting discursive events. So new is this type of discourse, it has given rise to a new demand and a new "right to be forgotten." It should sensitize us that what is involved in algorithmic curation is not a restriction on expression, but on the preservation and dissemination of expression, namely rarefaction, the activation of a set of internal rules limiting the half-life and circulation of statements that do not conform to certain conditions.
Second, rarefaction is not an external constraint on discourse, but a constitutive affordance of the very possibility of discourse formation. The "rarity of statements" is a necessary condition for them to become "things that are transmitted and preserved, that have value, and which one tries to appropriate." [Ibid., pp. 119-120] The procedures of rarefaction, including rules of exclusion, classification, and application of claims, are not only omnipresent while being historically contingent, numerous, and imposing, but they are absolutely necessary for the appearance of dis -course. As Foucault points out, "it is not easy to say something new" [Ibid., p. 44], and we are enabled to do so exactly by preexisting rules, both limiting and allowing to formulate heterogeneous novel statements." [Ibid., p. 224] Foucault notes that the aforementioned "gradation" or "gap" between the plethora of ordinary utterances and the rarity of preserved discursive statements "is neither stable, nor constant, nor absolute. but while the details of application may well change, the function remains the same, and the principle of hierarchy remains at work." [Ibid., p. 220] This seems almost prescient, though we are sure he did not have
in mind algorithmic curation. Nonetheless, the rise of social media does indicate that the gap between ephemeral utterances and statements "that are transmitted and preserved, that have value" has shifted once again. The boundary between the two is blurred by the electronic media that transform ephemeral utterances into widely circulated and enduring statements. It is unsurprising, therefore, that mechanisms of rarefaction are activated to introduce at least a modicum of rarity to this new type of discourse.
Finally, the principle of rarefaction that Foucault describes as "discipline," provides a particularly illuminating lens through which to understand why curation is better analyzed as rarefaction, rather than coercion. Foucault describes disciplines as "anonymous system[s] at the disposal of anyone who wants to or is able to use [them]," [Foucault, 1972, p. 221] providing a framework for the construction of statements, allowing new propositions to be produced, ad infinitum, within specific restrictions. Each discipline, as a particular discursive field, consists of "onerous and complete conditions" that a statement must fulfill "before it can be admitted within a discipline; before it can be pronounced true or false it must be... within the true." [Ibid., p. 224] Being within the true, even if the statement proves to be an error, is what gives the statement "value. [that] one tries to appropriate." [Ibid., p. 120] Thus, disciplines do not coerce speakers to speak in a particular way. They tempt or incentivize them by offering them the power of being "in the true," of enunciating what is taken to be true discourse. Individuals are not told what to say or not to say, but they are invited to step into a specific discursive position or "enunciative modality" [Ibid., pp. 50-55] and speak from it. Social media curation operates analogously. It doesn't offer speakers the power of being "in the true," but it does tempt and incentivize them with the power of "visibility." By definition, not all statements can be visible, nor to the same extent. Visibility, like "being in the true," relies on a principle of rarity. Algorithms favoring certain types of content, explicit restrictions regarding content deemed harmful, or regarding modes of expression deemed offensive, a curatorial code that favors and promotes certain formats over others - these are operators of rarefaction that together constitute an enunciative modality that users are encouraged, tempted, and in-centivized to inhabit if they want to be "visible."
At its core, rarefaction in the context of social media involves the controlled, selective, and organized production and dissemination of curated content, content producers, and modes of expression, which might be considered analogous to the principle of discipline. Importantly, Fou-cault's analysis allows us to distinguish rarefaction from coercion. While rarefaction involves the imposition of constraints and regulations on discourse production, individuals are not forcibly coerced into compliance. Curation-qua-rarefaction does not operate on the "exterior of discourse" prohibiting expression. It is a set of "internal rules" that is constitutive
of the very possibility of discourse. Curation invites users to step into a particular discursive subject position and speak from within it, in order for their statements to be preserved, disseminated, and have the value of "visibility" that "one would like to appropriate." Turner has drawn our attention to the increasing significance of algorithmic curation, but it does not indicate an intensification of epistemic coercion. Quite differently, it indicates a new rearrangement and ordering of discourse, the formation of an interface between "ordinary utterances" and "disciplines," between everyday talk and expert discourse, an intermediate realm where discursive events are neither ephemeral nor preserved "in the true." Instead of denouncing it as coercion, it would be better to try to work out what should be the new rules of rarefaction and how the two realms of discourse can be adjusted to one another.
References
Dandelet, 2021 - Dandelet, S. "Epistemic Coercion," Ethics, 2021, vol. 131, no. 3, pp. 489-510.
Davis, 2017 - Davis, J.L. "Curation: A Theoretical Treatment," Information, Communication & Society, 2017, vol. 20 (5), pp. 770-783.
Anderson, 2023 - Anderson, S. "Coercion," in: Zalta, E.N. & Nodelman, U. (eds.) The Stanford Encyclopedia of Philosophy. [https://plato.stanford.edu/archives/ spr2023/entries/coercion/, accessed on: 15.04.2024].
Foucault, 1972 - Foucault, M. The Archaeology of Knowledge and the Discourse on Language, trans. A.M. Sheridan Smith. New York: Pantheon Books, 1972.
Goffman, 1969 - Goffman, E. Strategic Interaction. Philadelphia: University of Pennsylvania Press, 1969.
Turner, 2024 - Turner, S. "Epistemic Coercion," Epistemology & Philosophy of Science, 2024, vol. 61, no. 3, pp. 21-38.
Weber, 1978 - Weber, M. Economy and Society. Berkeley, Los Angeles, London: University of California Press, 1978.
