On what the system permits you to oppose.
Absorption
In 1944, Theodor Adorno and Max Horkheimer published Dialectic of Enlightenment, whose chapter on the culture industry argued that mass culture under monopoly capitalism does not emerge from the public but is manufactured for it. The culture industry does not repress. It standardizes. It absorbs political differences into a grammar of consumption so complete that the absorption is experienced as choice. "All mass culture under monopoly is identical," they wrote. The observation was not about television. It was about a system in which even the categories of critique are produced by the apparatus they claim to critique (Adorno and Horkheimer, 1944, pp. 94-136).
Twenty-three years later, Guy Debord extended the argument from culture to the totality of social life. The Society of the Spectacle (1967) opens with the proposition that "all that once was directly lived has become mere representation." Thesis 4 provides the definition that governs this analysis:
"The spectacle is not a collection of images; it is a social relation among people, mediated by images."
The spectacle is not the screen. It is the condition in which the screen becomes the primary medium of social organization. When that condition obtains, every activity, dissent included, is metabolized into the logic of the system. The Situationists gave this metabolism a precise name: recuperation, the process by which radical critique is absorbed, stripped of structural content, and repackaged as commodity (Debord, 1967, thesis 59). The counterculture becomes a brand. The revolution becomes a poster. The graffiti becomes a museum exhibit. And the protest becomes content.
The proof arrived promptly. In May 1968, the largest mass upheaval in France since 1848 shut down factories, paralyzed the government, and covered the walls of Paris in Situationist slogans: "Down with the spectacular-commodity economy." "Be realistic, demand the impossible." Within months, the advertisements that followed had borrowed the typography. The films had borrowed the aesthetics. The academy had borrowed, then domesticated, the vocabulary. Scholarship on the aftermath records that the spectacle "subjugated its antagonists by manipulating their desires and satisfying false needs," while "a whole generation of university intellectuals" undertook the assimilation and distortion of the movement’s meaning (Amorós, 2008). The uprising’s "only major victory was survival in memory." The recuperation was total. The structure was untouched. The slogans survived as fashion.
This analysis examines a contemporary instance of the same process. It is not about whether the Palestinian cause is legitimate. It is. Over 72,000 Palestinians have been killed since October 2023. The suffering is documented by the International Court of Justice, by the United Nations, by the 248 journalists and media workers killed in Gaza, a rate the Committee to Protect Journalists calls unprecedented in the history of modern conflict. What this analysis examines is not the cause but its capture: what happens to solidarity when it enters the spectacular frame, when the march becomes ceremony, when the boycott becomes consumer preference, and when the infrastructure that sustains the conditions being protested remains untouched because it is, in Fredric Jameson’s formulation, beyond the threshold of cognitive mapping: the point at which the complexity of the global system exceeds the capacity of the individual to represent it (Jameson, 1991, pp. 51-54). You cannot boycott what you cannot see. And the architecture of late capitalism is organized so that the substrate is invisible.
How tolerance represses
Herbert Marcuse’s "Repressive Tolerance" (1965) remains the sharpest articulation of a counter-intuitive proposition: tolerance can function as domination. The system does not suppress protest. It permits it, because the permission functions as evidence that the system is open, responsive, and therefore legitimate. The tolerance is repressive not despite but because of its generosity. It produces the appearance of democratic health while the structural conditions that provoked the protest remain unchanged (Marcuse, 1965, pp. 81-123). The march is permitted. The demand is heard. The portfolio does not move.
The decades since have confirmed Marcuse with a specificity he could not have anticipated, because the infrastructure of permission has changed. Consider the anatomy of the contemporary solidarity march. The route is negotiated with police, its boundaries are temporal and spatial, and its output is not political pressure but content: photographs of signs, video of chants, posts that are liked, shared, commented upon, and algorithmically distributed to audiences constituted by prior agreement. The algorithmic distribution is not incidental. It is the mechanism by which solidarity becomes spectacular. The platform’s recommendation engine ensures that the solidarity post reaches people who already agree, generating engagement metrics that generate revenue for the platform, which holds contracts with the system being protested. The circuit is closed before the march begins.
In the spring of 2024, pro-Palestinian encampments spread across American university campuses. At Columbia University, the central demand was divestment. The university’s president, Minouche Shafik, explicitly refused, offering instead to "publish a process for students to access a list of Columbia’s direct investment holdings" (Reuters, April 29, 2024). The offer was procedural rather than substantive: transparency about the process, not change to the portfolio. Over 100 students were arrested. The encampment was cleared by NYPD. A second formed, occupied Hamilton Hall, and was cleared again. Shafik resigned as president. The divestment policy did not resign with her. Across the country, no major university endowment divested. The opacity of endowment portfolios, managed through thousands of pooled funds by third-party managers under confidentiality agreements, rendered even the identification of relevant holdings structurally intractable.
The arrests became content. The police response became content. The debate about whether the protests were legitimate or antisemitic became content. All of it circulated. None of it moved capital. The tolerance operated precisely as Marcuse described: the demand was acknowledged, the protest was absorbed, and the structure remained. Acknowledgment was the system’s way of digesting the demand. The digestion was the repression.
Action, labour, and who gets killed
Hannah Arendt, in The Human Condition (1958), distinguished three fundamental activities of the vita activa. Labour is cyclical, biological, bound to necessity, consumed in the moment of its production; the dish washed today must be washed again tomorrow. Work fabricates the durable world of objects and institutions; it persists beyond the act. Action is political in the deepest sense: it occurs between people, it discloses who someone is rather than merely what they do, and its consequences are irreversible and unpredictable (Arendt, 1958, pp. 7-17). What makes action distinct from both labour and work is that it initiates something genuinely new. It requires a public space in which to appear. And it cannot be undone.
The distinction clarifies what has happened to solidarity. The march, the social media post, the boycott of a consumer brand: these are, in Arendt’s terms, closer to labour than to action. They are cyclical. The march this Saturday, the next march next Saturday. They are consumed as they are produced: the post circulates, is liked, decays in the algorithmic feed, and is replaced by the next cycle’s content. They do not fabricate a durable object or institution. They do not initiate something new in the world. They reproduce a form. The reproduction, however sincere, is not the initiation of the new. It is the repetition of the known, within bounds the system has already established as acceptable.
What Arendt called action is visible elsewhere in this conflict. The legal teams at the International Court of Justice initiated something genuinely new when South Africa filed its application under the Genocide Convention in December 2023. The ICJ’s May 24, 2024 order, adopted 13-2, required Israel to immediately halt military operations in the Rafah governorate, keep the Rafah crossing open for humanitarian assistance, and ensure access for UN investigative bodies. This was action in Arendt’s sense: irreversible, consequential, operating in the public space of international law, initiating a legal process whose implications cannot be recalled. It did not trend. It produced an order.
The 179 UNRWA staff members killed in Gaza since October 2023, the highest staff death toll in the history of the United Nations, were engaged in what Arendt would classify as labour, the biological maintenance of life. But the conditions under which that labour was performed transformed it into something else. To deliver aid in a zone where aid delivery is a lethal act is to disclose who you are in the public space, with consequences that are irreversible. The labour became action. Secretary-General Guterres reported that they "were killed delivering life-saving aid; others alongside their families; others while shielding the vulnerable." The 248 journalists and media workers killed were not performing solidarity. They were performing documentation under conditions that made documentation lethal. Five Al Jazeera journalists were killed in a single strike in Gaza City in August 2025.
The system’s differential response reveals its classification of threat:
The gradient is the argument. The system does not suppress dissent uniformly. It sorts it by the level at which it operates, and the sorting reveals, with more precision than any theoretical framework, what the system considers a threat. Consumer-level opposition is tolerated because it does not reach the infrastructure. Institutional opposition is cleared. Infrastructure-level opposition is expelled. Material aid is defunded. Documentation is killed. The response escalates in direct proportion to the structural depth at which the opposition operates. Spectacular solidarity is safe, visible, and socially rewarded. Material solidarity is dangerous, invisible, and systematically eliminated. The sorting is the politics.
The system does not fear the boycott it tolerates.
It defunds the agency it cannot absorb.
Complicity of the medium
Marshall McLuhan argued in Understanding Media (1964) that the medium shapes the message: the form of a technology determines its social consequences independent of its content. The railway’s significance was not the freight it carried but the restructuring of space, time, and settlement it produced. Electric light’s significance was not what it illuminated but the activities it made possible. The medium is the message because the medium determines the structure of the interaction, regardless of what is communicated through it (McLuhan, 1964, pp. 7-21).
The corollary this analysis identifies has not, as far as the available literature indicates, been articulated in this form: when the medium is the infrastructure of the system being opposed, every act of opposition conducted through that medium is simultaneously an act of participation in the system. The solidarity and the complicity become the same gesture. They are performed through the same data centres, monetized by the same attention economy, and settled by the same infrastructure contracts.
When the medium is the infrastructure of the system being opposed, every act of opposition conducted through that medium is simultaneously an act of participation in the system.
Structural corollary to McLuhanThe evidence is architectural.
Project Nimbus is a $1.2 billion contract awarded in April 2021 to Google and Amazon Web Services to provide cloud computing and AI services to the Israeli government, including the defence establishment (Reuters, April 21, 2021). The contract stipulates that neither company can deny service to any Israeli government entity. Both committed to building local data centres in Israel. When Google employees protested in April 2024 under the banner "No Tech for Apartheid," staging sit-ins at offices in New York and Sunnyvale, Google fired them: initially 28, then over 50. The workers who attempted to oppose the infrastructure were expelled from it. The consumers who posted solidarity content on platforms served by that infrastructure were permitted to continue, and their engagement generated revenue.
The revenue is not marginal. Alphabet’s 2024 revenue was $350 billion, of which $264.6 billion came from advertising, which is to say, from the engagement metrics generated by every interaction on its platforms, including the solidarity posts (Statista, 2025). AWS generated $107.6 billion in 2024, with an operating income of $39.8 billion and a 37% margin. These are the financial substrates of the infrastructure. The solidarity post, circulating on a Google platform, generates a fractional contribution to the $264.6 billion in advertising revenue earned by the company that holds the contract. The contribution is infinitesimal per post and structural in aggregate.
Microsoft’s entanglement is not commercial but operational. A 2025 Guardian investigation revealed that Israel’s Unit 8200 uses Microsoft Azure to store recordings of approximately one million intercepted Palestinian phone calls per hour. The arrangement originated in a 2021 meeting between CEO Satya Nadella and the Unit 8200 commander. Intelligence gathered through this mass surveillance is transcribed, translated, and cross-checked with AI targeting systems to select bombing targets in Gaza. Microsoft provided the IDF with at least $10 million in technical support and access to GPT-4. In September 2025, after sustained investigative journalism by the Guardian and Associated Press, Microsoft partially restricted access, an outcome produced not by consumer boycotts but by the form of action the system classifies as a threat: documentation that makes the invisible infrastructure legible.
Infrastructure entanglement: the dependency chain
- Google/AWS: $1.2B Project Nimbus contract; cannot deny Israeli government service; data centres under construction in Israel
- Microsoft Azure: Stores ~1M intercepted Palestinian phone calls/hour for Unit 8200; provided GPT-4 and $10M+ technical support to IDF
- HP Enterprise: BASEL biometric ID system at West Bank checkpoints; ethnicity-stratified ID cards for Palestinian travel permits (Who Profits)
- AWS: 30% of global cloud market; hosts banking, e-commerce, social media, authentication infrastructure
- Azure: 20% of global cloud; hosts enterprise, government, healthcare systems
- Google Cloud: 12% of global cloud; combined with YouTube and Search: $350B/yr in total Alphabet revenue
Together, three American companies operate over 60% of the cloud infrastructure on which modern economic life depends. A single AWS outage in October 2025 disrupted airlines, payment processors, social media, and authentication services globally.
HP Enterprise developed and maintains the BASEL biometric identification system deployed at West Bank checkpoints and manufactures the ethnicity-stratified ID cards required for Palestinian travel permits (Who Profits).
The causal chain is complete. The solidarity post is composed on a device whose operating system is provided by one of these companies. It is uploaded to a platform owned by another. It is served from data centres operated by a third. It generates engagement that generates revenue that funds the infrastructure that includes the contracts. The post opposing the system circulates through the system, and its circulation is the system’s revenue. McLuhan’s proposition was that the medium shapes the message. The proposition here is stronger: when the medium is the infrastructure of the condition being opposed, the medium does not shape the message. It negates it. The solidarity and the complicity are architecturally identical. They are the same HTTP request.
Surface and substrate
Consumer boycotts have measurable effects at the retail level. Starbucks experienced a 7% revenue decline in mid-2024, its largest since the pandemic, attributed in part to pro-Palestinian boycotts. The company’s stock lost $11 billion in market value over nineteen days in late 2023 (TRT World, 2024). McDonald’s reported slower sales globally. These are not negligible figures. They demonstrate what consumer behaviour, aggregated, can accomplish at the retail surface.
But the surface is not the substrate. Pierre Bourdieu’s sociology clarifies why the boycott targets the one and not the other. The consumer-facing brand is the site of what Bourdieu called distinction: the accumulation of cultural capital through consumption choices that signal social position (Bourdieu, Distinction, 1979, pp. 169-225). The decision not to enter Starbucks is observable. It communicates values. It positions the boycotter within a social field. Catherine Liu, in Virtue Hoarders (University of Minnesota Press, 2021), extends this to the Professional Managerial Class, which she argues performs moral superiority through consumption patterns experienced as ethical conviction but functioning as class distinction. The boycott signals who you are. The cloud contract is invisible and therefore carries no signal value. You cannot perform your opposition to something your interlocutor cannot see.
Social psychology provides the mechanism by which the surface substitutes for the substrate. Research on moral licensing (Engel and Szech, PLOS ONE, 2019) demonstrates that engagement in one ethical behaviour reduces motivation regarding ethical behaviour in other domains. When a product improves on one ethical dimension, consumers care less about its other ethical failings. The researchers identify "static moral self-licensing": fulfilling one ethical criterion is sufficient for maintaining a high moral self-image, making it easier to ignore others. The effects spill across domains and compound. External observers, even when financially incentivized for accuracy, remain "completely oblivious" to the licensing effect.
Research on climate behaviour (Giesekam et al., Global Environmental Change, 2022) confirms the pattern: reminders of past climate-friendly actions significantly reduced discomfort about ongoing harmful behaviours. Being reminded of not flying reduced motivation to change meat consumption. The ethical act licensed the ethical omission.
The Starbucks boycott, read through this literature, does not merely fail to reach the infrastructure. It may actively reduce the motivation to engage with it. The ethical act at the visible level produces a moral self-image sufficient to license continued dependence at the invisible level. The boycotter has done something. The something generates the psychological experience of adequacy. The adequacy reduces the discomfort that might otherwise motivate engagement with the structural questions: what does the cloud run on, who holds the contracts, where does the surveillance data route. The boycott is not merely insufficient. Through the mechanism of moral licensing, it functions as a substitute for the structural engagement it claims to serve. What Debord called recuperation at the systemic level, moral licensing describes at the individual psychological level: the ethical performance generates the self-image that permits continued participation in the system being opposed. The critique becomes its own anaesthetic.
Why apartheid was different
The comparison to the South African anti-apartheid movement is instructive precisely because the structural conditions were different, and the differences explain why one form of pressure produced systemic change and the other, in its current form, structurally cannot.
The anti-apartheid boycott, divestment, and sanctions campaign brought measurable, escalating pressure on the South African economy. Citibank’s 1985 decision to call in loans was a turning point. The US Congress passed the Comprehensive Anti-Apartheid Act in 1986, overriding President Reagan’s veto. Scholarly analysis (Waterloo Human Rights, 2019) confirms that the combined campaign "infiltrated South Africa socially, economically, and politically, causing discontent within the state and eroding domestic support for apartheid."
Three structural conditions enabled that effectiveness. None obtains in the present case (Clarno and Vally, 2023):
First, the targeted economy was isolable. South Africa’s economy could be severed from global capital flows, trade networks, and institutional relationships without simultaneously severing the infrastructure on which the boycotter’s daily life depended. You could divest from South African firms without discovering that your email, your banking, and your cloud storage ran on South African infrastructure. The target was legible and separable. In the present case, the target economy is not a distant state. It is the infrastructure of the boycotter’s own digital existence. The entanglement is not commercial. It is architectural.
Second, state actors joined. The anti-apartheid campaign achieved structural effect not through consumer boycotts alone but through the accumulation of state-level action: arms embargoes, trade restrictions, diplomatic isolation, the withdrawal of institutional capital. Scholars of the movement note that the decisive leverage came when governments moved. In the present case, some state actors have moved: Canada halted arms shipments (February 2024); the Netherlands was court-ordered to halt F-35 part exports; Belgium, Spain, and Italy suspended new weapons sales; the UK suspended 30 of 350 licences. These are structural measures. They are also partial, and the largest supplier, the United States, has not joined.
Third, internal resistance was organized around a single political movement with clear strategic goals and a negotiating position. The ANC could receive the fruits of external pressure and convert them into political change. The Palestinian BDS campaign operates in a context of internal fragmentation, territorial partition, and a targeted state that is not merely adjacent to the global infrastructure economy but embedded in its substrate.
The lesson is not that boycotts are futile. The lesson is that the structural conditions under which consumer boycotts produce systemic change require the target economy to be isolable from the boycotter’s own infrastructure. When the infrastructure is the entanglement, the boycott reaches the retail surface while the substrate, where the contracts, the surveillance systems, and the targeting architecture operate, continues undisturbed.
Where the frameworks meet
Jodi Dean’s concept of "communicative capitalism" describes the condition precisely. Advanced capitalism co-opts the ideals of participation and free expression: the more messages circulate, the less political force any individual message carries. Communication proliferates. Response does not (Dean, 2009, pp. 22-30). The solidarity post circulates through the same algorithmic distribution system that serves advertisements, entertainment, and the platform’s own revenue model. The solidarity and the commodity travel on the same rails, are served from the same data centres, and are monetized by the same attention economy. The medium is not neutral. The medium is the infrastructure. And the infrastructure holds contracts with the systems being protested.
Byung-Chul Han, in Psychopolitics (2017), identifies the subject this system produces. Neoliberalism governs not through discipline or prohibition but through the psyche: the subject is encouraged to experience freedom, self-expression, and agency as the medium through which power operates. "The freedom of the subject of achievement is itself a form of coercion" (Han, 2017, pp. 1-3). The protest is experienced as self-expression. The boycott is experienced as ethical consumption. The social media post is experienced as political speech. All three operate within the logic of the system they claim to oppose, on platforms owned by the companies they claim to resist, generating the engagement metrics that fund the infrastructure they claim to challenge. The subject does not experience constraint because the system has organized resistance as a form of participation. The participation is the constraint.
Mark Fisher’s Capitalist Realism (2009) named the condition within which all of this operates: the widespread sense that capitalism is not merely the dominant system but the only conceivable one. Fisher identified "reflexive impotence" as its characteristic affect: the awareness that things are bad, that action is unlikely to change them, and that one acts anyway in ways that confirm rather than challenge the impotence. "What if you held a protest and everyone came?" he asked (Fisher, 2009, ch. 5). May 1968 supplied the answer: the system’s capacity to absorb does not diminish with scale. It increases. A larger protest generates more content, more circulation, more evidence of democratic tolerance, and no additional structural consequence. The absorption is proportional to the spectacle.
The convergence across these frameworks is the argument. Adorno and Horkheimer identified the absorption of critique into the culture industry. Debord extended it to social life and named the mechanism: recuperation. Marcuse identified the permission structure: repressive tolerance. Arendt distinguished action from labour and showed what political action requires. McLuhan showed that the medium determines the structure of the interaction. Jameson named the cognitive limit that keeps the infrastructure invisible. Fisher named the psychological condition: reflexive impotence. Dean named the medium: communicative capitalism, in which messages proliferate and response does not. Han named the subject: the achievement-subject who experiences resistance as self-optimization. The social psychologists named the licensing mechanism that converts the ethical gesture into the permission to continue.
Together, they describe a system in which dissent is not suppressed but metabolized: converted from a threat into a nutrient, from a challenge to the structure into a confirmation of its openness. The protest feeds the system’s self-image as tolerant. The boycott feeds the boycotter’s self-image as ethical. The engagement feeds the platform’s revenue. The infrastructure continues.
What would be structural
This analysis would be guilty of its own critique if it produced only the feeling of insight. Some structural action already exists, and the evidence is specific.
The ICJ case (South Africa v. Israel) is structural because it operates within an institutional framework that produces binding legal obligations. The May 2024 order to halt operations in Rafah was not content. It was a legal instrument. Universal jurisdiction cases in national courts have produced arrest warrants and travel restrictions. These proceedings are slow, procedurally complex, and invisible to algorithmic distribution. They are the only mechanism that produces legal consequences for state conduct.
Arms export restrictions by Canada, the Netherlands (court-ordered, finding "clear risk" of international humanitarian law violations), Belgium, Spain, Italy, and partially the UK are structural because they restrict the material flow that sustains military operations. They were produced by legal challenges, parliamentary votes, and government reviews, not by consumer boycotts. They do not generate engagement. They restrict supply chains.
Investigative journalism that makes the infrastructure legible is structural because it is the only mechanism that has produced infrastructure-level change in this context. Microsoft’s partial restriction of Israeli military access to Azure came not from consumer pressure but from Guardian and link:https://npr.org/2025/08/27/nx-s1-5518786/Associated Press] investigations that documented the surveillance architecture. The journalism performed what Jameson’s cognitive mapping describes as the analytical task: making the invisible system visible, rendering the substrate legible, so that what was un-opposable because it was unseen becomes opposable because it is documented. The destruction of documentation infrastructure, the killing of 248 journalists, is the system’s response to the one form of action that breaches the threshold of cognitive mapping. The killing is the measure of the threat.
The Norwegian Government Pension Fund’s exclusion of weapons producers represents institutional divestment at sovereign scale. University divestment from fossil fuels, which took decades of patient institutional labour, provides the operational model. The mechanism is the same whether the target is carbon or contracts: sustained, unglamorous pressure on the institutions that allocate capital, operating on timescales measured in years rather than news cycles.
The common feature of structural action is that it is institutional, slow, procedurally demanding, materially risky, and invisible to the spectacular frame. It does not lend itself to algorithmic distribution. It does not accumulate cultural capital for the participant. It produces material consequences through material mechanisms. The spectacular frame cannot accommodate it, and that inability is not a limitation of the frame. It is its function.
What the system permits
Vocabularies manage situations rather than describe them, and the management serves interests not always identical to those of the people speaking.
The narrative of solidarity manages the moral self-image of the participant without altering the structural conditions of the situation being protested. The vocabulary is "I stand with." The infrastructure is AWS, Azure, and Google Cloud. The vocabulary operates at the surface, where it is visible and permitted. The infrastructure operates at the substrate, where it is invisible and un-opposable. The gap between them is not a failure of conviction. It is a structural property of the economy in which the conviction is expressed. And the vocabulary manages the gap by providing the sensation of political action at the level where political action does not threaten the substrate.
The pattern across domains is identical. A vocabulary claims to describe a condition. In practice, it manages the condition. The management produces the experience of agency without its structural exercise. And the gap between experience and exercise is not an accident but an architecture: the designed relationship between what is visible and what is infrastructural, between what is permitted and what is structural, between the domain of consumer choice and the domain of systemic dependency.
The question is not whether to protest. Protest is a condition of democratic life, and the right to it is not negotiable. The question is whether the form protest has taken in this historical moment, ceremonial, mediated, algorithmically distributed, morally licensing, and architecturally entangled with the infrastructure it claims to oppose, is adequate to the conditions it addresses. The evidence presented here suggests it is not. Not because the participants lack sincerity, but because the system has organized itself so that sincerity, expressed through the channels it provides, produces engagement rather than consequence. The channels are the infrastructure. The engagement is the revenue. The revenue funds the contracts. The contracts are the conditions being protested. The circuit closes.
The people who produce consequence are not, for the most part, the ones with the largest platforms. They are the ones with the largest exposure: the legal teams who file applications under the Genocide Convention, the parliamentarians who vote to restrict arms exports, the journalists who document what a million intercepted phone calls per hour are used for, the aid workers who deliver sustenance in zones where delivery is lethal. Their work is material. Their risk is real. And the system’s response to them, defunding, termination, bombardment, is the clearest available measure of what it considers a threat.
The system does not fear your protest.
It runs on the same server.