Cybernetic Proletarianization

For most developed economies, digitalization is seen as a hope for escaping the long-term stagnation of productivity growth. This article examines the process by which digitalization promotes proletarianization in manual labor, and attempts to combine two perspectives—empirical analysis of the labor process and theoretical class analysis—to argue that algorithmic control has a significant impact on class composition. Algorithmic control creates a contradictory process in which living labor is both expelled from and reintegrated into production. To substantiate this argument, this article draws on empirical research on algorithmically controlled manual labor in German manufacturing and delivery logistics.

对大多数发达经济体而言，数字化是摆脱生产增长率长期停滞的希望所在。本文探讨了数字化促进手工生产劳动中的无产阶级化过程，并尝试把关于劳动过程的实证分析与理论层面的阶级分析这两种视角结合起来，提出算法控制对阶级构成具有重要影响。算法控制制造了一个矛盾的过程，即在生产中活劳动既遭受驱逐又被重新融入。为论证这一观点，本文将通过对德国制造业和配送物流业中由算法控制的手工劳动的实证研究来说明。

Germany is one of the major economies implementing production digitalization. Under the label “Industry 4.0,” the German government has heavily promoted and subsidized production digitalization, focusing primarily on the implementation of algorithmic management systems rather than automation. Particularly in manufacturing sectors such as mechanical engineering and chemical production, algorithmic management is becoming increasingly important: the labor processes of 64% of chemical industry workers and 69% of mechanical engineering workers are controlled by software. In the years before the “Industry 4.0” label was introduced, the delivery logistics sector had already become a pioneer in implementing algorithmic work control, coinciding with a general rise in the economic status of this sector. According to calculations by the German Federal Statistical Office, in 2017 the delivery logistics sector employed over 2.4 million people, with a total turnover of approximately 330.7 billion euros. Despite increasing automation, employment in the German delivery logistics sector grew by 18.8% between 2012 and 2016, far exceeding the overall employment growth rate. The development of the logistics sector has benefited from the construction of digital platforms: on the one hand, the growth of online retail platforms has significantly increased the number of jobs in German warehouses and freight transport; on the other hand, algorithmic control has also greatly helped to solve problems such as “last-mile delivery.”

德国是实现生产数字化的主要经济体之一。在“工业4.0”的标签下，德国政府大力宣传和补贴生产数字化，主要聚焦在算法管理系统而非自动化的实施上。特别是在机械工程、化工生产等制造业部门，算法管理日益重要，64%的化工行业工人和69%的机械工程工人的劳动过程都是由软件控制的。在“工业4.0”标签提出的前几年，配送物流部门就已经成为实施算法工作控制的先行者，这与该部门经济地位的普遍提高相吻合。根据德国联邦统计局的计算，2017年配送物流部门的员工总数超过240万，总营业额约为3307亿欧元。尽管自动化程度不断提高，但2012年至2016年间，德国配送物流部门的就业人数却增长了18.8%，远高于整体就业增长率。物流部门的发展得益于数字平台的建设：一方面，在线零售平台的发展大大增加了德国仓库和货物运输的就业数量；另一方面，算法控制在很大程度上也有助于解决“最后一英里配送”等问题。

In light of this, the case studies in this article focus on the digitalized manufacturing sector and the delivery logistics sector, comprising four cases representing two companies from each sector. From 2017 to 2019, the author conducted participatory observation in all cases: working on the assembly line at the mechanical engineering company “Smart Electric,” working as a bicycle courier for the food delivery company “Smart Delivery,” and participating in a six-day series of meetings organized by the works councils of the chemical manufacturer “Smart Solutions” and the online retailer “Smart Shopping” (all company names are pseudonyms). These participatory observations involved a total of 55 comprehensive interviews with workers, engineers, and managers. Half of these interviews were conducted at other digitalized companies outside the selected cases to ensure the external validity of the observations. On this basis, the author analyzed the data according to standards of qualitative content analysis.

有鉴于此，本文的案例研究聚焦数字化制造部门和配送物流部门，包括四个案例，分别代表两个部门的两家公司。从2017年到2019年，笔者对所有案例进行了参与式观察：参与机械工程公司“智能电气”的装配线工作，在食品配送公司“智能配送”担任自行车快递员，以及参加化学品制造商“智能方案”和在线零售商“智能购物”劳资委员会组织的为期六天的系列会议（以上公司均为化名）。这些参与式观察涉及对工人、工程师和经理的55次综合访谈。其中一半访谈是在所选案例之外的其他数字化公司进行的，以确保观察结果的外部有效性。在此基础上，笔者根据定性内容分析标准对数据进行了分析。

This article combines empirical findings with Marxist class analysis to propose the concept of “cybernetic proletarianization.” This new form of proletarianization can be understood as cybernetic because the affected workers are drawn into digital feedback loops that not only control their work but are also the fundamental reason for their future expulsion from the production process. Thus, part of these workers’ labor consists in making themselves redundant. At the same time, however, this “cybernetic proletarianization” also gives rise to a new wave of labor conflicts, which this article describes as “technopolitics from below.” These conflicts take the form not only of individual and collective “technological disobedience” but also of collective self-organization that extends beyond the traditional institutional framework of industrial relations.

本文将实证结果与马克思主义的阶级分析相结合，提出了“控制论无产阶级化”的概念。这种新形式的无产阶级化可以被理解为控制论，因为受影响的工人被卷入数字反馈回路中，这些回路不仅控制着他们的工作，也是他们未来被驱逐出生产过程的根本性原因。因此，这些工人的一部分工作就是让自己变成多余之人。但同时，这种“控制论无产阶级化”也会带来新一波劳资冲突，本文将其描述为“来自底层的技术政治”。冲突不仅采取个人和集体“技术不服从”（technological disobedience）的形式，也会采取集体自组织的形式，并超出了工业关系的传统制度范围。

The second and third parts of this article will reconstruct the Marxist theory of proletarianization, drawing in particular on relevant content from *Theories of Surplus Value* concerning technological proletarianization. The following sections then address the empirical phenomena of “cybernetic proletarianization.” The fourth to sixth parts will elucidate the various forms in which living labor is expelled from production, both quantitatively and qualitatively. The seventh part describes the contradictory process of the reintegration of devalued labor into production, accompanied by algorithmic control. The eighth part shows that “cybernetic proletarianization” leads to various labor conflicts that have the potential to break the spiral of devaluation.

本文第二部分和第三部分将重构马克思主义的无产阶级化理论，特别是借鉴《剩余价值理论》中关于技术无产阶级化的相关内容。接下来的部分则涉及“控制论无产阶级化”的经验现象。第四部分至第六部分将阐明活劳动在数量和质量上被从生产中驱逐出去的各种形式。第七部分描述了贬值劳动力又重新融入生产的矛盾过程，其间还伴随着算法控制。第八部分的研究表明，“控制论无产阶级化”导致了各种劳资冲突，这些冲突具有打破螺旋式贬值的潜力。

II. Does a Digital Proletariat Exist?

二、存在数字无产阶级吗？

The core criterion of Marx’s concept of the “proletariat” is wage dependence, which distinguishes it from the concept of “worker.” The latter refers to the actual position within wage labor, while the term “proletariat” emphasizes wage dependence, including the unemployed or the (falsely) self-employed in the so-called gig economy. In Marx’s view, “proletarianization” comprises two steps: the “formal” subsumption and the “real” subsumption of labor under capital. “Formal” subsumption is the elimination of forms of subsistence other than wage labor; “real” subsumption refers to the replacement of skills and the subsequent automation of living labor in the process of industrialization.

马克思关于“无产阶级”概念的核心标准是工资依赖性，因而不同于“工人”概念。后者指的是在雇佣劳动中的实际地位，而“无产阶级”这个术语则侧重于工资依赖性，包括失业者或所谓的零工经济的（虚假的）自雇者。在马克思看来，“无产阶级化”包括两个步骤：劳动对资本的“形式”从属和“实际”从属。“形式”从属是消灭雇佣劳动以外的其他生存形式，“实际”从属则是指工业化过程中技能的替代以及随后的活劳动的自动化。

Modern debates on proletarianization emphasize the tendency toward devaluation within existing wage relations. This debate can be divided into two approaches: inequality research and the sociology of labor. The former mainly refers to income distribution and lifestyle. The latter focuses on the labor process as the core site of proletarianization, emphasizing the trend toward management control through production rationalization and the deskilling of workers. Both approaches indicate that the proletariat is not a purely industrial phenomenon. On the contrary, in post-industrial economies since the 1980s, a “service proletariat” has emerged within a new cycle of subjective and objective devaluation. This is consistent with Marx’s earlier description: “Through the use of machinery (and generally through the development of the productive forces of labor), net income (profit and rent) increases, and the bourgeois will need more domestic servants than before.”

关于无产阶级化的现代辩论强调现存工资关系中的贬值倾向。这场辩论可分为两种方式：不平等研究和劳动社会学。前者主要指收入分配和生活方式。后者侧重于作为无产阶级化核心场所的劳动过程，强调通过生产合理化的管理控制和工人去技能化的趋势。这两种方式都表明，无产阶级并非纯粹的工业现象。相反，在1980年代以来的后工业经济体中，一种“服务性无产阶级”出现在主客观贬值的新循环之中。这与马克思曾经的描述是一致的：“由于使用机器（一般由于劳动生产力的发展），纯收入（利润和地租）会增加，资产者会比以前需要更多的家仆。”

Digitalization has not reduced the total demand for low-skilled workers. Research has found that since the late 1990s, routine activities in manual labor have increased. Of course, the digitalization process also creates new, highly skilled, high-wage jobs. However, the emergence of these jobs depends primarily on sustained economic growth and is difficult to achieve in the manual labor fields discussed in this article. Overall, compared with productivity, wages in OECD member countries shrank between the 1980s and 2010. In Germany and the United States, this gap is particularly pronounced. In the context of digitalization, wage inequality will intensify across the board: on the one hand, demand for highly skilled workers drives up wages in the high-wage sector; on the other hand, work intensification and automation are increasing wage pressure on medium- and low-skilled workers. Increasing wage inequality leads to the following situation: companies become more willing to replace high-wage labor with low-skilled, low-wage labor. This also applies to private households: the widening gap between high-wage and low-wage jobs increases the relative cost advantage for wealthy families in outsourcing work to others. Therefore, employment polarization is the most likely long-term impact of digitalization on the labor market.

数字化并没有降低对低技能工人的总需求。研究发现，自1990年代末以来，手工劳动中的日常活动有所增加。当然，数字化过程也会创造出新的高素质、高薪资的工作。但这些工作的出现主要取决于经济的持续增长，并且很难出现在本文所讨论的手工劳动领域之中。总体而言，与生产率相比，经济合作与发展组织成员国的工资在1980年代至2010年间有所缩水。在德国和美国，这一比例的差距尤为明显。在数字化背景下，工资不平等将全面加剧：一方面，对高素质工人的需求导致高薪部门的工资上涨；另一方面，工作强化和自动化正在增加中低技能工人的工资压力。日益增加的工资不平等导致了如下情形：公司越来越愿意用低技能、低薪资的劳动力取代高薪劳动力。这同样适用于私人家庭：高薪工作与低薪工作之间日益扩大的差距增加了富裕家庭将工作外包给他人而获得的相对成本优势。因此，就业的两极分化是数字化最有可能对劳动力市场产生的长期影响。

What emerges in the digitalization process is not technological unemployment, but new labor-intensive production processes. For example, algorithmic work control is an important technical foundation for online mail order, which also relies on a large and continuously growing use of human labor. Algorithmic work control is also the foundation of the delivery logistics sector. Another labor-intensive sector of the digital economy is platform-mediated domestic services—especially cleaning and maintenance in private households. The platform care.com alone has over 14.6 million “care workers.” Furthermore, algorithmic work control also generates another form of labor-intensive production, namely “click work.” In most cases, this involves data-related classification tasks, such as identifying offensive images and text on social media platforms, maintaining databases, inputting documents, or training artificial intelligence. This work is usually done at home and therefore remains invisible—it is also called “ghost work.” Nevertheless, millions of people engage in these poorly paid and burdensome jobs.

在数字化过程中出现的不是技术性失业，而是新的劳动密集型生产过程。例如，算法工作控制是网上邮购的重要技术基础，网上邮购也依赖于大量且持续增长的人力使用。算法工作控制也是配送物流行业的基础。数字经济的另一个劳动密集型部门是以平台为媒介的家政服务——尤其是私人家庭的清洁和维护。仅care.com平台就有1460多万“护理人员”。此外，算法工作控制还产生了另一种劳动密集型的生产形式，即“点击工作”（click work）。在大多数情况下，这涉及到与数据相关的分类工作，如识别社交媒体平台上的攻击性图片和文字、维护数据库、输入文档或训练人工智能等。这项工作通常在家里完成，因此不为人所见——它也被称为“幽灵工作”。尽管如此，仍有数百万人从事这些收入微薄且负担沉重的工作。

III. Technological Proletarianization

三、技术无产阶级化

Nick Dyer-Witheford describes how the global digital economy creates a new “cyber-proletariat” subject to a cycle of human labor displacement and reintegration, the “cybernetic vortex.” The cyber-proletariat includes workers in all areas of the digital economy, from resource extraction to the creation of digital technologies to their use. The concept of the cyber-proletariat is well suited to analyzing global processes occurring between different regions and economic sectors of the world. However, the subjective dimension of proletarianization, i.e., the experience of the labor process, is necessarily excluded from this global perspective. This raises the following question: how much do a Silicon Valley programmer and a slave in a Congolese coltan mine have in common? Is it sufficient to regard them as both affected by the “cybernetic vortex” and to treat the cyber-proletariat as a social group, or even a class?

尼克·戴尔-维瑟福特（Nick Dyer-Witheford）描述了全球数字经济是如何创造一个新的“赛博无产阶级”的，它受制于人类劳动转移和重新融入的循环，即“控制论漩涡”。赛博无产阶级包括数字经济各个领域的工人，从资源开采到数字技术的创造再到数字技术的使用。赛博无产阶级的概念非常适合分析世界不同地区与经济部门之间发生的全球进程。然而，无产阶级化的主体维度，即劳动过程的经验，必然被排除在这种全球视野之外。这就提出了如下问题：硅谷的程序员和刚果钶钽铁矿的奴隶有多少共同之处？认为他们都受到“控制论漩涡”的影响，把赛博无产阶级视为一个社会群体甚至一个阶级，这就足够了吗？

Based on the labor process, proletarianization can be understood in four aspects: First, the expulsion of living labor from production, both in qualitative and quantitative terms. In most cases, quantitative expulsion is achieved through automation or work intensification, i.e., extending the working day or accelerating the labor process. Qualitative expulsion usually takes the form of deskilling. This means that production knowledge is objectified in machinery or centralized in management, thereby devaluing labor. Second, this expulsion typically does not lead to technological unemployment but is supplemented by the reintegration of devalued labor. For example, through the use of precarious labor forms such as bogus self-employment, restructuring can be implemented at the enterprise level. This is often facilitated by labor market policies. In terms of proletarianization, these two processes cannot take merely the form of occupational change, but also that of labor devaluation, with a decline in the wages of affected workers as its main manifestation. Third, the intensification of relations of domination in production. This manifests as technical, bureaucratic, or personal managerial control, but it also typically contains elements of worker consent. Fourth, attention to the capacity of the proletariat to take political action. In a Marxist context, this is usually understood as “class consciousness.” The traditional Marxist concept of consciousness requires insight into the objective conditions of exploitation and an ideological stance toward class struggle. However, other concepts of proletarianization grasp this political dimension in a more practical way. For example, Claude Lefort argues that class conflict is a core dimension of proletarianization, but there is no direct causal relationship between ideology and actual conflict. Therefore, the “proletarian experience” in the labor process and the accompanying actual conflicts, especially those in the workplace, are the core issue. The following sections will draw on these four aspects of proletarianization to analyze the impact of algorithmic work control on manual and logistics work.

以劳动过程为基础，可以从四个方面来理解无产阶级化：第一，从质量和数量上将活劳动从生产中驱逐出去。在大多数情况下，数量上的驱逐是通过自动化或工作强化来实现的，即延长工作日或加快劳动过程。质量上的驱逐通常采取去技能化的形式。这意味着生产知识在机器中被对象化或在管理中被集中化，从而使劳动贬值。第二，这种驱逐通常不会导致技术性失业，而是通过贬值劳动力的重新融入来补充。例如，通过使用虚假的自雇等不稳定的劳动形式，可以在企业层面采取重组这一方式。这通常是由劳动力市场的政策促成的。就无产阶级化而言，这两个过程不能仅仅采取职业转换的形式，也要采取劳动力贬值的形式，受影响工人的工资下降是其主要表征。第三，强化生产中的支配关系。这表现为技术、官僚或个人管理控制的形式，但它通常也包含工人同意的要素。第四，关注无产者采取政治行动的能力。这在马克思主义的语境中通常被理解为“阶级意识”。传统马克思主义的意识概念要求洞察剥削的客观状况，并要求对阶级斗争持意识形态的态度。然而，无产阶级化的其他概念以一种更实际的方式把握了这一政治维度。例如，克劳德·勒福特（Claude Lefort）认为，阶级冲突是无产阶级化的一个核心维度，但意识形态与实际冲突之间没有直接的因果关系。因此，劳动过程中的“无产阶级经验”以及与之相伴的实际冲突，尤其是在工作场所的冲突，是核心问题之所在。以下各部分将借鉴无产阶级化的这四个方面，分析算法工作控制在手工和物流工作中的影响。

IV. Deskilled Flexibilization

四、去技能的灵活化

In all cases studied in this article, algorithmic management is employed to control the labor process. At Smart Electrical Engineering Company, algorithmic management includes installing a screen in the workplace that displays assembly instructions in pictures and text. At Smart Delivery Company, algorithmic management is implemented through an application on the courier’s smartphone, guiding them through the city. At Smart Shopping Company, workers are controlled by a handheld scanner with a display screen. All these systems issue detailed instructions to workers, making the labor process relatively easy to execute. Therefore, a manager at Smart Electrical explained that the goal is “either to make things go faster, or to enable people with lower qualifications to do the work.” Thus, management hopes “to transfer work to… workers with lower qualifications and lower costs.” At Smart Delivery, algorithmic work control made it possible to replace individual courier training with online group video conferences. A team leader explained that because the labor process is simple, “(Smart Delivery’s) principle” is “to hire anyone.” A middle manager explained that at Smart Shopping, algorithmic work control shortened on-the-job training to one and a half days. In all these cases, the company’s goal appears to be reducing labor costs through deskilling. Digitization also increases the complexity of certain jobs, thereby polarizing the workforce. However, in the manual labor cases studied here, these two processes are unbalanced, because deskilling has the potential to further reduce labor costs.

在本文研究的所有案例中，它们都采用了算法管理来控制劳动过程。在智能电气机械工程公司，算法管理包括在工作场所安装一个屏幕，以图片和文字显示装配说明。在智能配送公司，通过快递员智能手机上的应用程序实现算法管理，引导他们穿过城市。在智能购物公司，工人由一个带显示屏的手持扫描仪控制。所有这些系统都向工人发出详细的指令，使劳动过程相对容易执行。因此，智能电气公司的经理解释说，这样做的目标是“要么让事情进展得更快，要么让资质较低的人能干”。因此，管理层希望“将工作转移给……资历较低、成本较低的工人”。在智能配送公司，算法工作控制使通过在线小组视频会议取代快递员的个人培训成为可能。一位团队负责人解释说，由于劳动过程简单，“（智能配送公司的）原则”是“雇用任何人”。一位中层管理人员解释说，在智能购物公司，算法工作控制使在职培训缩短到一天半。在所有这些案例中，公司的目标似乎都是通过去技能化来降低劳动成本。数字化还增加了某些工作的复杂性，从而使劳动力两极分化。然而，在本文研究的手工劳动案例中，这两个过程并不平衡，因为去技能化有可能进一步降低劳动成本。

Now, a core element of algorithmic work control lies precisely in the automation of middle management. This also leads to a new form of digital deskilling. On the one hand, the digital machine itself is capable of absorbing workers’ knowledge through digital tracking. On the other hand, the formalization of machine knowledge becomes a core work task for the workers themselves. A typical example is that at Smart Electrical, skilled workers in the assembly department are responsible for encoding their knowledge into the digital work control system. Therefore, they are instructed to create digital manuals that “anyone off the street” can understand. In this way, workers input production knowledge that only they possess into the digital system. This means that this knowledge is now objectified as the company’s direct property, independent of its human carriers, and can be used to absorb low-skilled workers. This is explicitly seen as a shift from expensive automation to cheap manual labor. One manager explained: “Why should we automate everything? Because our employees are too expensive.” Now, he has cheaper workers and can “have them do it manually.”

现在，算法工作控制的一个核心要素恰恰在于中层管理的自动化。这也导致了一种新的数字去技能化形式。一方面，数字机器本身能够通过数字跟踪来吸收工人的知识。另一方面，机器知识的形式化成为工人本身的核心工作内容。一个典型的例子是，智能电气公司装配部门的技术工人负责将他们的知识编入数字工作控制系统。因此，他们被告知要制作“大街上随便一个人”都能看懂的数字手册。这样一来，工人们把只有他们才拥有的生产知识输入到数字系统中。这意味着，这些知识现在被对象化为公司的直接财产，独立于人类载体，并可以用来吸纳低技能工人。这被明确视为从昂贵的自动化转向廉价的手工劳动。一位经理解释说：“为什么我们要让一切都自动化？因为我们的员工太贵了。”现在，他有了更便宜的工人，可以“让他们用手工来做”。

Algorithmic work control also enables the flexibilization of labor use. At Smart Shopping and Smart Delivery, due to the widespread use of digital instructions, algorithmic work control reduces the costs associated with high labor turnover. Furthermore, at Smart Delivery, the application not only tells workers how to work but also when to work. This more directly links labor input to fluctuations in demand. At Smart Solutions, a chemical manufacturer, algorithmic work control is used to outsource certain tasks while maintaining control over the labor process, thereby ensuring uniform standards. At Smart Electrical, algorithmic work control is used for the scheduling of workers across different workstations within the factory. This makes it possible to offer a wider variety of products without needing dedicated workers for each variation. A scientist involved in implementing these algorithmic management systems acknowledged that the resulting flexibility is one-sided: the company can adjust work input according to order fluctuations, while workers must adjust their lives according to changes in working hours.

算法工作控制还实现了劳动力使用的灵活化。在智能购物公司和智能配送公司，由于数字指令的广泛应用，算法工作控制降低了劳动力高流动率所带来的成本。此外，在智能配送公司，应用程序不仅告诉员工如何工作，还告诉他们何时工作。这使得劳动力的投入更直接地与需求的波动挂钩。在化学品制造商智能方案公司，算法工作控制用于外包某些任务，同时保持对劳动过程的控制，从而确保标准统一。在智能电气公司，算法工作控制用于工厂内不同工作站之间工人的调度。这样就有可能提供更多种类的产品，同时又不需要为每一种变动配备专门的工人。一位参与实施这些算法管理系统的科学家承认，由此产生的灵活性是单方面的：公司可以根据订单的波动调整工作投入，而工人则必须根据工作时间的变化调整自己的生活。

In summary, this process can be described as “deskilled flexibilization.” The process of digital deskilling involves encoding human knowledge into the machine, a process in which workers sometimes must directly participate. On this basis, companies can arrange workers more flexibly; this is what is meant by deskilled flexibilization. Overall, deskilled flexibilization amounts to the expulsion of living labor from the production process, both qualitatively and quantitatively. Workers’ production knowledge is absorbed by the “system.” Subsequently, this knowledge is objectified as the company’s direct property, independent of its human carriers, and can be used at will to “equip” other low-skilled workers.

总之，这个过程可以被描述为“去技能的灵活化”。数字去技能化过程是将人类知识编进机器，工人有时必须直接参与其中。在此基础上，公司可以更有弹性地安排工人，这就是所谓的去技能的灵活化。总的说来，去技能的灵活化相当于从质量和数量上将活劳动从生产过程中驱逐出去。工人的生产知识被“转向系统”所吸收。随后，这些知识被对象化为公司的直接财产，独立于其人类载体，并可被随意用于“装备”其他低技能工人。

This correspondingly reduces the number of workers required for flexible production. Because, relatively speaking, if the existing workforce can be transferred, temporary workers can more easily join the labor process, and fewer workers are needed to maintain flexible production. However, algorithmic work control not only issues instructions but also tracks workers’ performance. As the next section will elaborate, this generates a new system of time management.

这也相应地减少了灵活生产所需的工人数量。因为相对而言，如果现有劳动力可以转移，临时工就更容易加入劳动过程，而维持灵活生产的工人就可以减少。然而，算法工作控制不仅能发出指令，还能跟踪工人的工作表现。正如下一部分将阐述的，这产生了一种新的时间管理制度。

V. Cybernetic Intensification

五、控制论的强化

Assessing the labor process is the core function of algorithmic work control. At Smart Electric Company, the system measures the time required for each work step in units of one-tenth of a second. At Smart Delivery Company, the application monitors couriers’ movements via GPS, then uses this data to automatically create employee performance profiles, including speed, punctuality, time spent with customers, and other factors. Couriers believe these profiles determine whether their employment contracts will be extended. At Smart Shopping Company, every move workers make in the warehouse is monitored by handheld scanners. One worker said: “They can even see how many steps you take. So they can see where you are, how much you work, how much you do, how long you are idle, or when you went to the bathroom.”

评估劳动过程是算法工作控制的核心功能。在智能电气公司，系统以1/10秒为单位测量每个工作步骤所需的时间。在智能配送公司，应用程序通过全球定位系统监视快递员的动向，然后利用这些数据自动创建员工的绩效档案，包括速度、准时性、与客户相处的时间和其他因素。快递员认为，这些档案是决定其雇佣合同能否延长的依据。在智能购物公司，工人在仓库中的一举一动都受到手持扫描仪的监控。一名工人说：“他们甚至能看到你走的步数。因此，他们可以看到你在哪里，你工作了多少，你做了多少，你有多长时间无所事事，或者你什么时候去了趟洗手间。

In all cases, the data collected by the program is automatically fed back to the workers. At Smart Electric Company, the control system displays the worker’s current speed compared to the average speed. The purpose is to allow workers to “self-adjust” and continuously increase their speed. At Smart Delivery Company, when a courier’s performance is poor, the application plays a prompt tone or automatically calls them. At Smart Solutions Company, the machines used for production automatically measure human utilization; if it falls below 80%, they automatically assign additional tasks to the worker.

在所有的案例中，程序收集到的数据都会自动反馈给工人。在智能电气公司，控制系统显示工人当前的速度与平均速度的对比。这样做的目的是让工人“自我调整”，不断提高速度。在智能配送公司，当快递员的表现不佳时，应用程序会播放提示音或自动拨打电话给他们。在智能方案公司，用于生产的机器自动测量人力的利用率，如果低于80%，它就会自动给工人分配额外的任务。

These automatic feedback technologies can be understood as a shift from the Taylorist control model to a cybernetic model and work intensification. The middle management created by Taylorism has been automated through algorithms. The general standards that every worker must comply with are replaced by an open-ended optimization process, which draws on the cybernetic idea of feedback-based self-organization. These cybernetic ideas have a long history. However, only the application of digital sensor technology can meet the technical requirements for truly realizing feedback-based self-organization. This means that the data collected by the program is no longer handed over to superiors for punishment, but is fed back to employees so they can “self-adjust.”

这些自动反馈技术可以被理解为从泰勒主义的控制模式向控制论模式和工作强化的转变。泰勒主义创建的中层管理通过算法实现了自动化。每个工人都应遵守的一般标准被一种开放式的优化过程所取代，它借鉴了控制论中基于反馈的自组织思想。这些控制论思想由来已久。然而，只有数字传感器技术的应用才能满足真正实现基于反馈的自组织的技术要求。这意味着，程序收集到的数据不再交给上级用于惩戒，而是反馈给员工，让他们“自我调整”。

This cybernetics has the basic expected effect of intensifying work. Surplus time is automatically detected and eliminated in order to produce more labor power in the same amount of time. This control model is by no means as novel as it first appears. It is a new wave of digitally supported systematic rationalization. This means that the labor process is analyzed and optimized across its entire operation and between companies (e.g., in supply chains). In this process, the principles of “lean production” are essentially applied. Thus, “waste” in the form of unnecessary work delays or poor resource utilization can be identified and eliminated. Through seamless tracking of the supply chain, companies can further achieve just-in-time logistics or sequence logistics, thereby reducing work interruptions caused by operations. However, most importantly, digital feedback encourages workers themselves to discover and eliminate “time waste” in the work process. The purpose is to initiate a permanent self-optimization, extending the collective “continuous improvement process” of lean production to the individual and pushing it to its extreme. Therefore, the novelty here lies in the central role of automatic feedback-based self-organization. This concept originates from classical cybernetics, which is why this systematic rationalization can be called “cybernetization.”

这种控制论具有强化工作的基本预期效果。多余的时间将被自动检测和消除，以便在相同的时间内产生更多的劳动力。这种控制模式绝非初看起来那么新颖。它是数字化支持系统合理化的一股新的浪潮。这意味着，劳动过程在其整个运营和公司之间（如供应链）得到分析和优化。在这一过程中，“精益生产”的原则基本上得到了应用。因此，不必要的工作延误或资源利用不善等形式的“浪费”能够被识别并消除。通过对供应链的无缝跟踪，公司可以进一步实现按时物流或按序物流，从而减少因操作引起的工作中断。然而，最重要的是，数字反馈鼓励工人自己发现和消除工作流程中的“时间浪费”。这样做的目的是启动一种永久性的自我优化，将精益生产的集体“持续改进过程”扩展到个人，并使之达到极致。因此，这里的新颖之处是基于自动反馈的自组织的核心地位。这一理念源于经典的控制论，这就是为什么这种系统性的合理化可以被称为“控制论化”。

Subjectively, most of the workers interviewed considered this cybernetic model an additional source of stress. An employee at Smart Shopping Company commented: “You have to rely entirely on this damn device. I no longer feel like myself; as long as I enter Smart Shopping, I am Smart Shopping. I cannot make any decisions myself. The scanner tells me: right or left, down, up.” A member of the works council at Smart Solutions Company explained: “Of course, it’s a psychological burden. People’s fears are triggered… it creates work pressure. Because they say the machine feeds back data… saying you’re not efficient enough; you have to improve more.” At Smart Delivery Company, 63% of the couriers surveyed believed they were often at the mercy of digital technology. Overall, the cybernetic intensification of work seems to produce a strong sense of lack of autonomy in workers, which is also another factor in quantitatively expelling living labor from the production process, because if employed workers work faster, fewer workers are needed. However, this indirect expulsion sometimes also turns into a direct substitution of living labor.

从主观上看，接受访谈的大多数工人都认为这种控制论模式是一个额外的压力来源。智能购物公司的一名员工评论道：“你得完全依赖这个破设备，我不再觉得自己是我，但只要我进入智能购物，我就是智能购物。我自己不能做任何决定。扫描仪告诉我：向右或向左、向下、向上。”智能方案公司的一名劳资委员会成员解释道：“当然，这是一种心理负担。人们的恐惧被激发出来……产生了工作压力。因为他们说，机器会反馈数据……说你效率不高；你必须要有更多的进步。”在智能配送公司，有63%的受访快递员认为自己经常受到数字技术的摆布。总体而言，控制论的工作强化似乎会让工人产生强烈的不自主感，这也是在数量上将活劳动驱逐出生产过程的另一个因素，因为如果受雇人员的工作速度更快，所需的工人就更少。然而，这种间接的驱逐有时也会变成对活劳动的直接替代。

VI. Data-Based Automation

六、基于数据的自动化

This article focuses on the specific form of “data-based automation,” which refers to automation processes directly based on human labor data collected in advance within the cybernetic model of work control. At Smart Shopping Company, data monitoring warehouse workers’ movements is used for automated sorting by transport robots. In another case, workers in internal logistics were digitally tracked while driving transport vehicles through the factory. This data was not only used to control their labor process but also served as the basis for developing artificial intelligence for automated transport vehicles, in order to completely replace human drivers.

本文重点关注“基于数据的自动化”这一特定形式，这指的是在工作控制的控制论模式中，直接以事先收集的人类劳动数据为基础的自动化过程。在智能购物公司，监控仓库工人行动的数据被用于运输机器人实现自动分拣。在另一个案例中，内部物流的工人在驾驶运输车辆穿过工厂时受到了数字跟踪。这些数据不仅被用来控制他们的劳动过程，还被用作开发自动运输车辆人工智能的基础，以便完全取代人工司机。

Thus, digital tracking can not only control the labor process but also realize its automation. Another example of this process is Uber’s statement that the company’s business purpose is to collect data to program self-driving cars. Therefore, data generation often plays a role equivalent to the direct production of commodities. This data can be sold as an additional commodity or become the basis for automation processes.

因此，数字跟踪不仅可以控制劳动过程，还可以实现劳动过程的自动化。这个过程的另一个例子是优步的一个声明，即该公司的业务目的是收集数据，为自动驾驶汽车编程。因此，数据生成往往与商品的直接生产有着同等作用。这些数据可以作为附加商品出售，也可以成为自动化流程的基础。

In Marx’s view, workers engaged in wage labor expend themselves because the capitalist appropriates the products of their labor. Production increases both the capitalist’s wealth and power, while also intensifying the workers’ poverty and powerlessness. Under algorithmic work control, this relationship is further intensified: workers expend not only the products of their labor but also their production knowledge, and thereby the basis for selling their own labor power. Part of the work of the cybernetic proletariat is to make themselves redundant by producing data that is used to completely replace them in the production process. Thus, for the cybernetic proletariat, this data is even more “alien and hostile” than any previous product of wage labor. Data collection is used not only for an additional level of value but also to replace its source—human labor.

在马克思看来，从事雇佣劳动的工人之所以耗费自己，是因为资本家占有他们的劳动产品。生产既增加了资本家的财富和权力，也加剧了工人的贫困和无能为力。在算法工作控制下，这种关系得到进一步强化：工人不仅耗费自己的劳动产品，而且耗费自己的生产知识，进而耗费出卖自己劳动力的基础。控制论无产阶级的部分工作就是通过生产被用来在生产过程中完全取代自己的数据，从而让自身变得多余。因此，对控制论无产阶级来说，这些数据比以往的任何雇佣劳动产品都更加“异己和敌对”。数据收集不仅用于额外的价值层面，还用于取代其来源——人类劳动。

In summary, trends such as deskilling flexibilization, cybernetic (work) intensification, and data-based automation mean the expulsion of human labor from the production process. Fundamentally, this is an inherent tendency of capitalist production. However, in the cases studied in this article, it takes a specific form: unlike the indirect substitution of labor power by machinery described by Marx, current workers are directly “making themselves redundant.” The data workers themselves produce becomes the basis for their expulsion from the production process. For example, workers at Smart Electric Company are required to encode their production knowledge into auxiliary systems. Or digital tracking is used, such as internal logistics workers being replaced by AI-driven transport systems. This new form of expulsion can be understood as cybernetic, because workers are drawn into digital feedback loops that not only rationalize their work but also lay the foundation for their future automation. However, this expulsion does not lead to technological unemployment; rather, it is accompanied by a contradictory process in which human labor is reintegrated into the production process.

总之，去技能的灵活化、控制论的（工作）强化和基于数据的自动化等趋势意味着将人类劳动从生产过程中驱逐出去。从根本上说，这是资本主义生产的固有趋势。不过，在本文研究的案例中，它采取了一种特定的形式：与马克思所描述的用机器间接替代劳动力不同，当前的工人们是直接“使自己变得多余”。工人自己生产的数据成为他们被驱逐出生产过程的基础。例如，智能电气公司的工人被要求将他们的生产知识编入辅助系统。或者采用数字跟踪，例如内部物流工人被人工智能驱动的运输系统所取代。这种新形式的驱逐可以被理解为控制论的，因为工人被卷入数字反馈回路中，这不仅使他们的工作合理化，也为他们未来的自动化奠定了基础。然而，这种驱逐并不会导致技术性失业，而是伴随着一个矛盾的过程，即人类劳动重新融入生产过程。

VII. The Reintegration of Labor Power

七、劳动力的重新融入

“Cybernetic proletarianization” includes not only expulsion but also the reintegration of human labor into the production process. In countries such as the United States, the United Kingdom, or Germany, only a small portion of laid-off workers remain long-term unemployed, while the majority are forced to compete with newly employed young people for precarious jobs. This situation gives rise to new forms of digital production with extremely high labor intensity. The labor processes of the two logistics companies studied in this article are examples of this algorithmic impact on the reintegration of labor power. As mentioned above, the interviewed managers explicitly regarded cheap, digitally deskilled labor power as a substitute for automation.

“控制论无产阶级化”不仅包括驱逐，还包括将人类劳动重新融入生产过程。在美国、英国或德国等国家，被裁员的人中只有小部分长期失业，而大部分不得不被迫与新就业的年轻人竞争不稳定的工作。这种情况出现了新的劳动密集程度极高的数字化生产形式。本文研究的两家物流公司的劳动过程就是这种算法对劳动力重新融入产生影响的例子。如上文所述，受访的经理们明确地将廉价的、数字化的去技能化劳动力视为自动化的替代品。

An important reason for the emergence of this cheap labor force is that a large part of the labor force consists of migrants. At Smart Shopping Company, workers explained that in some warehouses, up to 70% of the labor force is made up of migrants, many of whom are refugees with unstable residency status. In some cases, the company directly transports refugees from shelters by chartered bus. Most of the couriers at Smart Delivery Company are non-European migrants. Managers interviewed explicitly stated that after implementing algorithmic work control, they hoped to make greater use of migrant workers. In most cases, the qualifications of these migrants far exceed the work they perform. Most of them left their countries of origin and their professions for various political and economic reasons, and they are now forced to engage in low-skilled, algorithmically controlled work.

这种廉价劳动力产生的一个重要原因是，很大一部分劳动力是由移民组成的。在智能购物公司，工人们解释说，在一些仓库里，高达70%的劳动力由移民构成，其中许多人是居住身份不稳定的难民。在某些情况下，公司包车直接从庇护中心运来难民。智能配送公司的大多数快递员是非欧洲移民。受访经理明确表示，在实施算法工作控制后，希望更多利用移民工人。在大多数情况下，这些移民的资质远远超过了他们所从事的工作。他们中的大多数人因各种政治经济原因而离开原籍国和自己的职业，现在他们被迫从事低技能、受算法控制的工作。

It is not difficult to see that this complements a strong tendency toward the material devaluation of workers. At Smart Shopping Company, workers’ wages are so low that many warehouse workers rely on additional social welfare benefits. One worker explained:

不难看出，这与工人们物质贬值的强烈倾向是相辅相成的。在智能购物公司，工人的工资非常低，以至于许多仓库工人依赖额外的社会福利款。其中一名工人解释说：

I am a full-time employee, no longer living on temporary contracts, but the money I earn on the Smart Shopping platform is nowhere near enough. I live in Z, where rent is very high, and the fuel cost from Z to Y already exceeds 200 euros. As I said, the money is not enough. I work very hard, but I still have to apply for state assistance as a supplement. Unfortunately, that is still not enough; it’s terrible.

我是全职员工，不再依靠临时合同讨生活，但我在智能购物平台挣的钱无论如何都不够用。我住在Z地，房租很高，从Z地到Y地的燃料费已经超过200欧元了。我说过，钱不够。我工作很努力，但还是要向国家申请救济作为补充。不幸的是，这还不够，太糟糕了。

At Smart Delivery Company, workers are paid the hourly minimum wage. However, they still need to use this money to purchase and maintain work equipment. In addition, there are always delays in wage payments. This can lead to survival difficulties, especially for migrant workers who cannot access social assistance or family security. For example, the interviewed delivery workers sometimes cannot afford to buy food due to lack of money, and one of them was even homeless.

在智能配送公司，工人的工资为每小时最低工资。不过，他们仍需用这笔钱购买和维护工作装备。此外，在支付工资时总会出现拖延。这可能会引发生存困难，尤其是对那些无法获得社会援助或家庭安全保障的移民工人。例如，接受访谈的送餐员们有时也会因为囊中羞涩而买不起食物，其中一人甚至无家可归。

We must assume that the transfer and reintegration of human labor in the process of “cybernetic proletarianization” is by no means accidental. George Caffentzis argues that this connection can even be formulated as a general law: every introduction of a new technology in one industrial sector leads to an increase in labor-intensive production in another sector. However, “cybernetic proletarianization” is not a mechanical process determined by the logic of capital accumulation. On the contrary, it generates various conflicts that can influence or even break the trend of devaluation.

我们必须假定，在“控制论无产阶级化”的过程中，人类劳动的转移和重新融入绝非偶然。乔治·卡芬齐斯（George Caffentzis）认为，这种联系甚至可以被表述为一般规律：一个工业部门每引进一项新技术，就会导致另一个部门劳动密集型生产的增加。然而，“控制论无产阶级化”并不是一个由资本积累逻辑决定的机械过程。相反，它会产生各种冲突，可以影响甚至打破贬值趋势。

VIII. Technology Politics from Below

八、来自底层的技术政治

Many critical approaches to algorithmic management assert that algorithmic management leads to complete control over the labor process by management and the “total atomization” of workers, thereby eliminating all possibilities of opposition. This is not the reality. On the contrary, in the cases studied in this article, workers have developed various individual and collective strategies to deal with digitalization, which can be termed “technology politics from below.” It consists of three elements: first, technological disobedience, i.e., the use of technology contrary to its intended purpose. Second, an adversarial proletarian technological culture, i.e., the establishment of solidaristic adversarial relations among workers, thereby forming a critical culture toward certain technologies or technological uses. Third, organized technology politics, meaning the representation of interests through institutions or organizations such as works councils, trade unions, or political parties to influence the concrete implementation of projects and regulations. All these forms of technology politics played a role in the cases studied in this article.

许多关于算法管理的批判性方法断言，算法管理会导致管理者对劳动过程的完全控制和工人的“彻底原子化”，从而消除所有对立的可能性。这并非真实情况。相反，在本文研究的案例中，工人们制定了各种个人和集体策略来应对数字化，这些策略可被称为“来自底层的技术政治”。它由三个要素组成：一是技术不服从，即技术的使用与实施意图相反。二是对抗式的无产阶级技术文化，即在工人之间建立团结的对抗关系，从而形成一种对某些技术或技术使用的批判性文化。三是有组织的技术政治，这意味着通过劳资委员会、工会或政党等机构或组织代表来影响项目和法规的具体实施。所有这些形式的技术政治都在本文研究的案例中发挥了作用。

Many workers perceive algorithmic work control as an infringement on their personal dignity. One worker said: “I feel like a robot, constantly being monitored.” When interviewed workers mentioned algorithmic work control, similar phrases recurred; they felt they were being mistreated. For most workers, this infringement on dignity is a more important motive for resistance than material devaluation. The intensity of this resistance is astonishing, but it almost always takes informal forms.

许多工人认为算法工作控制侵犯了其人格尊严。一名工人说：“我感觉自己像个机器人，时刻受到监控。”当受访工人提到算法工作控制时，类似的话反复出现，他们觉得自己受到了虐待。对于大多数工人来说，这种对尊严的侵犯是比物质贬值更重要的对抗动机。这种对抗的激烈程度令人吃惊，但几乎都是非正式的形式。

On the other hand, in the cases of the two platforms, material working conditions are at the core of the conflict. Delivery services such as Smart Delivery are characterized by highly dangerous work processes that repeatedly cause injuries and deaths. At the same time, wages are so low that working in this industry often leads to poverty, and in some cases even homelessness. Similar situations are reported in the warehouses of the global online retailer. Workers respond to these conditions through different forms of struggle, including manipulating algorithms and going on strike. Couriers at Smart Delivery Company carried out informal logout actions. In Germany, warehouse workers at Smart Shopping Company hold regular strikes. Thus, they are part of a global movement of platform workers against precarious working conditions in the industry. These protests are particularly frequent in the platform-mediated delivery sector. In the UK, for example, according to Callum Cant’s estimates, the number of working days lost to strikes in this industry is 42% higher than the national average. Similar protest movements have occurred in other countries. The core demand of couriers is usually fair wages, rather than contracts based on false self-employment. Following protests, this demand has been implemented in many places, such as at NoTime delivery service in Switzerland. Similarly, around the world, Uber and its competitors are experiencing strikes and other forms of protest.

另一方面，在两个平台的案例中，物质工作条件是冲突的核心。如智能配送等送餐服务的特点是工作过程高度危险，屡屡造成人员伤亡。同时，工资非常低，以至于在这个行业工作往往会导致贫困，在某些情况下甚至会无家可归。全球在线零售商的仓库也报告了类似的情况。工人们通过不同形式的斗争来应对这些状况，包括操纵算法和罢工。智能配送公司的快递员开展了非正式的注销行动。在德国，智能购物公司的仓库工人定期举行罢工。因此，他们是全球平台工人反对行业不稳定工作条件运动的一部分。这些抗议活动在以平台为中介的快递领域尤为频繁。以英国为例，根据卡勒姆·坎特（Callum Cant）的估计，该行业因罢工而损失的工作日数量比全国平均水平高出42%。类似的抗议运动也发生在其他国家。快递员的核心要求通常是获得公平的工资，而不是在虚假的自雇基础上签订合同。在抗议活动之后，许多地方已经落实了这一诉求，如瑞士NoTime快递服务公司。同样，在世界各地，优步及其竞争对手也正在经历罢工和其他形式的抗议。

Overall, the digital economy seems particularly prone to generating conflict. The two typical devaluation trends—precarious employment and infringement on worker dignity—are at the core of this conflict. However, large-scale labor conflicts occur not only in typical digital economy sectors but are also embedded in all fields. A typical example is the successful “wildcat” strikes by teaching staff in multiple U.S. states. The trigger for these strikes was that teachers were forced to use digital fitness trackers so that their health insurance premiums would not be significantly increased. Not only in application, but also in the development of digital technology is another field of conflict. For example, the Tech Workers Coalition listed 100 different protest actions by tech industry workers in 2019, most of which occurred in the United States. In 35 of these protests, ethical concerns about the nature and use of the technology being developed were the trigger for conflict.

总体而言，数字经济引发冲突的可能性似乎特别大。就业不稳定和侵犯工人尊严这两个典型的贬值趋势是这一冲突的核心。然而，大规模的劳资冲突不仅发生在典型的数字经济领域，而且蕴含在所有领域之中。典型例子是美国多个州的教职员工成功举行的“野猫式”罢工。这次罢工的诱因在于教师被迫使用数字健身追踪器，这样他们的医疗保险费就不会被大幅提高。不仅是在应用方面，数字技术的发展也是另一个冲突领域。例如，技术工人联盟列出了2019年技术行业工人的100种不同的抗议行动，其中大部分发生在美国。在其中的35次抗议中，对所开发技术的性质和使用的伦理关切是引发冲突的导火索。

In the cases studied in this article, despite the trend toward technological and organizational atomization, workers managed to establish a culture of solidarity, which is the basis for collective action. Furthermore, the critical use of technology also played an important role. Workers did not accept the self-optimization demands of feedback technology; instead, they developed a critical organizational culture. In all cases, these critical organizational cultures also led to technological disobedience, i.e., the use of technology contrary to its intended purpose. In some cases, they also led to various forms of collective self-organization by workers.

在本文研究的案例中，尽管存在技术和组织原子化的趋势，但工人们仍设法建立了团结文化，这是集体行动的基础。此外，基于对技术的批判性利用也发挥了重要作用。工人们不接受反馈技术的自我优化要求，而是发展了批判性的组织文化。在所有的案例中，这些批判性的组织文化也导致了技术不服从，即技术使用与其实施目的相反。在某些情况下，它们还会导致工人以各种形式进行集体自组织。

The role of trade unions in these conflicts is contradictory. In the companies studied in this article, union strength is negatively correlated with “cybernetic proletarianization.” The more pronounced the latter, the weaker the union organization. However, none of the companies studied in this article are union strongholds. At Smart Electrical Company, the degree of unionization corresponds to the typically high level in the metalworking industry; Smart Electrical Company has a works council supported by an industry union; however, the previously applicable collective bargaining agreement was terminated before the digitalization process began. Smart Solutions Company has a strong works council, but its connection to the union is quite loose, relying more on informal worker self-organization. The degree of unionization at Smart Shopping Company is quite low; due to the low level of organization and the company’s use of advanced digital resource planning, the union, while relatively proactive, is too weak to influence company operations through strikes. At Smart Delivery Company, the level of organization is even lower. However, as we have seen above, informal resistance forces have accumulated strongly. Overall, the digital economy seems to dislike the social partnership institutions that are highly developed in Germany. Instead, a more adversarial model of negotiation may emerge.

工会在这些冲突中的作用是矛盾的。在本文研究的公司中，工会的力量与“控制论无产阶级化”呈负相关。后者越明显，工会组织就越薄弱。然而，本文研究的公司都不是工会的大本营。在智能电气公司，工会化程度与金属加工业典型的高平均水平相符，智能电气公司有一个由行业工会支持的劳资委员会；然而，之前适用的集体谈判协议在数字化进程开始之前就被终止了。智能方案公司有一个强大的劳资委员会，但它与工会的联系十分松散，更多地依靠非正式的工人自组织。智能购物公司的工会化程度相当低，由于组织水平低下以及公司采用先进的数字资源规划，工会虽然相对积极主动，但力量太弱，无法通过罢工来影响公司的运营。在智能配送公司，组织程度甚至更低。然而，正如我们在上文所看到的那样，非正式的对抗力量积聚得很强。总体而言，数字经济似乎不太喜欢社会合作机构，而这些机构在德国却非常发达。相反，一种更加对立的谈判模式或将出现。

IX. Conclusion

九、结 论

It is taken for granted that the polarization of work is the main impact of digitalization on the labor market. However, in the manual labor fields of manufacturing and distribution logistics studied in this article, a continuous cycle of devaluation can be observed, which is described here as “cybernetic proletarianization.” In all cases studied in this article, the following core features can be found to varying degrees: (1) issuing commands via computers. (2) Digital evaluation of work activities. (3) Using the collected data to replace human labor in the production process, specifically through the realization of deskilled flexibilization, work intensification, or data-based automation. In either case, part of the labor process of the cybernetic proletariat is to make itself superfluous. (4) Precarious and low-paid employment relations.

人们理所当然地认为，工作两极分化是数字化对劳动力市场的主要影响。然而，在本文研究的制造和配送物流的手工劳动领域，可以观察到连续的贬值循环，本文将其描述为“控制论无产阶级化”。在本文研究的所有案例中，都可以不同程度地发现如下核心特征：（1）通过计算机发出指令。（2）对工作活动进行数字化评估。（3）利用收集到的数据取代生产过程中的人类劳动，具体做法是实现去技能的灵活化、工作强化或基于数据的自动化。无论是哪种情况，控制论无产阶级的部分劳动过程就是使自身成为多余。（4）不稳定和低报酬的雇佣关系。

“Cybernetic proletarianization” is defined not only by a common objective economic situation, such as “cyber-proletariat,” “high-tech proletariat,” or “precariat,” but objective and subjective factors play an equal role. In other words, the cybernetic proletariat actually creates a common subjective experience. On the one hand, these are feelings of involuntariness and disrespect evoked by digital commands. On the other hand, the pressure from continuous digital feedback and the existential insecurity of precarious employment relations are also crucial. Therefore, the concept of “cybernetic proletarianization” describes a dynamic process, not a static social group.

“控制论无产阶级化”不仅由共同的客观经济形势来定义，如“赛博无产阶级”、“高科技无产阶级”或“岌岌可危阶级”，而且客观因素和主观因素发挥着同等的作用。换言之，控制论无产阶级实际上创造了共同的主观经验。一方面，这些都是由数字指令所唤起的不自主体验感和不被尊重感。另一方面，持续的数字反馈带来的压力感以及不稳定的雇佣关系带来的生存不安全感也至关重要。因此，“控制论无产阶级化”概念描述的是一个动态的过程，而不是一个静态的社会群体。

This process can be seen as successive stages of a spiral devaluation (Figure 1). In the companies studied in this article, the first stage begins with the implementation of algorithmic work control systems. These systems facilitate deskilled flexibilization, thereby driving the expulsion of living labor in the initial stage. The second stage that follows is cybernetic (work) intensification. This is achieved through the digital evaluation of the labor process. The data collected in this way enables the third stage of this cycle: data-based automation, which heralds a further drive to expel living labor. However, the displacement in these three stages leads not only to technological unemployment, but also to the fourth stage of “cybernetic proletarianization”: the reintegration of living labor. This occurs when new labor-intensive production processes emerge in the course of digitalization, as in the case of the distribution logistics company studied in this article. Yet this reintegration is also a devaluation process, because these labor-intensive areas in the digital economy are almost always accompanied by extremely poor working conditions. At this point, the cycle of “cybernetic proletarianization” begins anew, with deskilled flexibilization, work intensification, and automation once again playing their roles in a new round, and so on.

这一过程可以被视为螺旋式贬值的连续阶段（图1）。在本文研究的公司中，第一阶段始于实施算法工作控制系统。这些系统有助于去技能的灵活化，从而在最初阶段推动了对活劳动的驱逐。紧随其后的第二阶段是控制论的（工作）强化。这是通过对劳动过程的数字化评估来实现的。以这种方式收集的数据促成了这一循环的第三阶段：基于数据的自动化，这预示着进一步推动对活劳动的驱逐。然而，这三个阶段的位移不仅仅导致技术性失业，而且导致“控制论无产阶级化”的第四阶段：活劳动的重新融入。当数字化过程中出现新的劳动密集型生产过程时，就会发生这种情况，例如本文研究的配送物流公司。然而，这种重新融入也是一个贬值过程，因为数字经济中的这些劳动密集型领域几乎总是伴随着极其恶劣的工作条件。此刻，“控制论无产阶级化”的循环又开始了，去技能的灵活化、工作强化和自动化又开始发挥新一轮作用，循环往复。

Figure 1: Stages of “Cybernetic Proletarianization”

图1：“控制论无产阶级化”的阶段

(Drawn by Hetti Frank)

（赫蒂·弗兰克绘制）

If “cybernetic proletarianization” is imagined as a constantly repeating spiral cycle of devaluation, then the labor processes of the companies studied in this article can be located in different cycles. Smart Electric Company is in the first cycle. This means that with the implementation of algorithmic work control systems, “cybernetic proletarianization” has only just begun. Therefore, the labor process still largely consists of high-level assembly work and relatively secure employment. However, the devaluation trends of deskilled flexibilization and work intensification are already in effect. Smart Solutions Company is also in the first cycle, but has progressed further. Thus, all stages of “cybernetic proletarianization” are occurring, including the digital automation of internal logistics and the planned reintegration of devalued workers through “crowdsourcing.”

如果把“控制论无产阶级化”想象成一个不断重复的螺旋式的贬值循环，那么本文所研究的公司的劳动过程就可以定位在不同的循环之中。智能电气公司处于第一个循环。这意味着，随着算法工作控制系统的实施，“控制论无产阶级化”才刚刚开始。因此，劳动过程在很大程度上仍然由高水平的装配工作和有较好保障的就业构成。然而，去技能的灵活化和工作强化的贬值趋势已经生效。智能方案公司也处于第一个循环，但已经更进一步。因此，“控制论无产阶级化”的所有阶段都在发生，包括内部物流的数字自动化以及通过“众包”方式有计划地融入被贬值的工人。

The work at Smart Delivery Company has already entered the second cycle. Therefore, workers face a labor process based on the completed first cycle. This means that platform-mediated delivery work, as a new digital business model, is replacing previous non-digitalized forms of food delivery. Here, the cycle begins anew, in the form of deskilling through digital control and work intensification through permanent evaluation. Smart Shopping Company has developed even faster, with some warehouses having completed the second cycle or begun the third cycle. Here, non-digital distribution processes (non-platform-organized retail or mail-order business) are also being replaced. Large numbers of workers are being reintegrated under highly precarious conditions. However, this generation of the cybernetic proletariat is also already being replaced by the large-scale automation of internal logistics work in some warehouses. This is the third cycle of cybernetic devaluation.

智能配送公司的工作已经进入第二个循环。因此，工人面临着以完成的第一个循环为基础的劳动过程。这意味着，以平台为中介的快递工作作为一种新的数字商业模式，正在取代以往的非数字化的食品配送形式。在这里，循环又重新开始了，其形式是通过数字控制进行去技能化，以及通过永久性评估来强化工作。智能购物公司则发展得更快，有些仓库已经完成了第二个循环或开始了第三个循环。在这里，非数字分配过程（非平台组织的零售或邮购业务）也被取代了。大批工人在十分不稳定的条件下正在被重新融入。然而，这一代控制论无产阶级也已经被一些仓库里的内部物流工作的大规模自动化所取代。这是控制论贬值的第三个循环。

Since “cybernetic proletarianization” must be thought of within a dynamic category, this concept cannot simply be limited to certain occupational groups. Therefore, if the cybernetic proletariat can be described as a social group, it is one with constantly blurring boundaries. Generally speaking, however, the more cycles of “cybernetic proletarianization” a production process undergoes, the closer its workers come to the core group of the cybernetic proletariat. Due to the qualitative design of this study, it is not possible here to assess the full extent of “cybernetic proletarianization.” Whether it develops into a general trend depends on the techno-political actions of the workers themselves. As we have seen, “techno-politics from below” takes multiple forms, some of which have already halted or even reversed the devaluation process.

既然必须在动态的范畴内对“控制论无产阶级化”进行思考，这个概念就不能简单地局限于某些职业群体。因此，如果控制论无产阶级可以被描述为一个社会群体，那么它就是一个边界不断模糊的社会群体。不过，一般而言，一个生产过程经历的“控制论无产阶级化”循环越多，工人就越接近控制论无产阶级的核心群体。由于本研究的定性设计，在此无法评估“控制论无产阶级化”的全部程度。它是否会发展成为一种普遍趋势，这取决于工人自身的技术政治行动。正如我们所看到的，“来自底层的技术政治”有多种形式，其中一些已经阻止甚至逆转了贬值过程。

(Simon Schaupp: Department of Social Sciences, University of Basel; Jiang Lin: Center for Foreign Languages and Literature Studies, Sichuan International Studies University)

（西蒙·朔普（Simon Schaupp）：瑞士巴塞尔大学社会科学系；蒋林：四川外国语大学外国语文研究中心）

Online Editor: Tongxin

网络编辑：同心