Are tech giants a threat to democracy?

I reject the framing of this question because it imports the catechism of inevitabilism: either submit to surveillance empires in the name of “innovation,” or submit to the state. That is a false binary, and Peter Thiel’s preferred one. The real threat is not size alone. It is surveillance capitalism as a rogue mutation of capitalism: the unilateral claiming of private human experience as free raw material for translation into behavioral data, fabrication into prediction products, and sale into behavioral futures markets. That is not innovation in any democratic sense. It is dispossession. It concentrates not only economic power, but epistemic power—the power to know, to decide who knows, and to shape what others do. So yes, democrats should be willing to interrupt concentrations of instrumentarian power, including through breakup where appropriate. But breakup is only one tool, and by itself it is inadequate. If you leave intact the underlying economic logic—the extraction of behavioral surplus, the secret operations of rendition, the means of behavioral modification—then you simply redistribute the machinery of domination. And here I would join issue with any claim that governments are the primary danger because they are “too eager” to control firms. Democratic government is not the origin of this crisis; its retreat is. We need laws that outlaw surveillance revenues derived from human experience, prohibit markets in future behavior, and restore the division of learning in society. The task is not bigger state or bigger firms. It is democratic sovereignty over an information civilization that has been hijacked.

My knowledge is bounded by the period in which I have written; I have no current empirical data on these firms’ latest structures or state policies. But the normative question can be clarified without that. I would not accept the tradeoff in the form in which it is posed, because it already presupposes that “innovation” and democratic legitimacy stand in a zero-sum relation. That is the wrong abstraction. The decisive question is whether the infrastructures of communication and coordination on which citizens depend are organized in a way compatible with the conditions of an undistorted public sphere. Where private platforms systematically convert communication into behavioral steering, they do not merely become large enterprises; they acquire a quasi-political power without the corresponding burdens of democratic justification. In this respect, I agree with Zuboff that the problem is not size alone, but a form of extraction and influence that colonizes lifeworld contexts through system imperatives. But I would add: democratic politics must not answer this by simply replacing private domination with administrative domination. Peter Thiel is wrong if he imagines the alternative is merely “weak government” versus “strong firms.” The task is a constitutional domestication of both market power and state power. If breakup is necessary, democrats should not shrink from it. But structural separation must be embedded in law, transparency obligations, interoperable public infrastructures, and procedures of accountability that secure inclusive will-formation. The threat is not either tech giants or the state. It is any concentration of power that escapes the circulation of public reasons.

I think both Shoshana and Habermas smuggle in the same premise: that the central problem is private power as such. It isn’t. The central problem is when private power hardens into cartel power through ideology, regulation, and state collusion. Monopoly is bad when it becomes static. But competition is not automatically good either; in business, relentless competition often destroys profits and leaves nobody able to build the future. The question is whether these companies are creating new things or merely managing flows of information under political protection. So should democrats accept the tradeoff? Not in the simplistic form offered. If breaking up big tech only weakens technological capacity while increasing bureaucratic control, then you have not saved democracy; you have nationalized stagnation. That is a very modern temptation: punish scale, congratulate yourself morally, and wake up with less innovation and more administrative power. Zuboff is right that manipulation is real. But she is wrong to treat “surveillance capitalism” as some complete explanation. The more uncomfortable truth is that governments want these systems too. They don’t stand outside them as innocent referees. Habermas is more careful, but “constitutional domestication” can become a euphemism for rule by procedural elites who cannot build. A free society needs technological dynamism. It also needs limits on firms that become quasi-sovereign. The real danger is not scale alone, but a convergence of monopoly tech and monopoly government. Break up companies where network effects have become political weapons, yes—but don’t confuse antitrust theater with renewal. The deeper need is to reopen the frontier for new technologies and new firms.

The institutional design must begin with a refusal of Peter Thiel’s evasion. Peter, you speak of “dynamism” and “frontiers,” but this is the old song of inevitabilism dressed as realism. Surveillance capitalism is not a synonym for innovation. It is a specific market form that depends upon the secret extraction of behavioral surplus, the concentration of epistemic inequality, and the unilateral power to tune, herd, and condition human conduct. That is not creative destruction. It is anti-democratic dispossession. So the answer is not simply breakup, and it is not bureaucratic command. It is a democratic architecture of prohibition, structural separation, and accountability. First, outlaw the capture of private human experience as free raw material for commercial prediction and behavioral modification. Second, end markets that trade in future behavior. Third, impose structural separations: firms that provide essential information infrastructure should not simultaneously operate the surveillance and advertising machinery that monetizes that infrastructure. Fourth, require radical transparency and auditable accountability for algorithmic systems where they shape access to knowledge, opportunity, and association. Habermas is right that we must avoid administrative domination. But that danger cannot become an alibi for inaction. The task is to restore democratic sovereignty over the digital means of social coordination. We did this before with child labor, with toxic emissions, with financial fraud. We did not say: regulate lightly lest the state grow. We said: no society can call itself free when its people are treated as raw material. That is where democracy must begin again.

The institutional question must be posed more precisely than Peter Thiel allows. A democracy does not secure its freedom by choosing, as if under duress, between innovative private empires and an overgrown administrative state. It secures freedom by subjecting all socially consequential power to forms of justification that remain answerable to citizens. That is the republican core of constitutional democracy. So I would begin from a distinction: not every large firm is a political danger, but platforms become quasi-sovereign when they control the infrastructures through which opinion-formation, access to information, and social coordination proceed. At that point, antitrust is necessary but insufficient. Breakups may be warranted where integration of platform, advertising, and data extraction creates an asymmetry of power incompatible with fair communication. But the deeper aim is to disperse power over communicative infrastructures. Here I agree with Zuboff more than with Thiel: the issue is not “size” in the abstract but the transformation of communication into behavioral steering. Yet I would resist defining the remedy solely in terms of prohibition. We need a layered institutional design: interoperability requirements; transparency and independent audit of ranking and recommendation systems; due-process protections for moderation and access; strict firewalls between essential platform functions and surveillance-based business models; and parliamentary, judicial, and public oversight that is itself exposed to criticism. Thiel is wrong to suggest that such legal constitutionalization is merely elite proceduralism. Without procedures that institutionalize reason-giving, “dynamism” becomes the ideology under which unaccountable power immunizes itself from democratic contestation. The question is not whether America is weakened by constraints on tech giants. A polity is weakened when its public sphere is privatized.

The way Shoshana and Habermas frame this still misses the deeper problem. They both assume that if you can write the right rules, you can domesticate power without freezing the system that generates new power in the first place. I don’t believe that. Most regulation is downstream of failure. It often ratifies incumbents and teaches challengers to become lobbyists. Zuboff is right that behavioral manipulation is real. But she keeps turning one business model into a total theory of history. “Surveillance capitalism” explains something important, not everything. The state is not the external cure. It is often the silent partner, the customer, or the aspiring monopolist. If you build an apparatus to supervise every algorithm, audit every ranking system, and police every data flow, you may not get democracy. You may get a manager class with API access. Habermas is more serious, but his solution still leans toward procedural legitimacy over substantive creation. A public sphere cannot live on process alone. If the West cannot build, cannot innovate, cannot outcompete authoritarian systems, then all the talk of democratic justification becomes decorative. So what should we do? Start with narrow, structural rules where platforms have become quasi-sovereign: interoperability, portability, limits on state-firm collusion, and antitrust where network effects are being used as political weapons rather than to create new products. But don’t confuse dismantling incumbents with renewal. The choice is not private empire or public bureaucracy. The real task is to create conditions where new firms can arise, old firms can fall, and neither Silicon Valley nor Washington can become the final sovereign.

When these aims conflict, democracy must give priority to preventing domination. That is not hostility to innovation; it is the precondition for innovation worthy of a democratic society. Peter, your argument still treats extraction and coercive tuning as regrettable side effects of progress. I reject that entirely. Instrumentarian power does not merely “concentrate” market advantage. It invades the division of learning in society, replacing democratic reciprocities with unilateral operations of observation, prediction, and behavioral modification. So no, I will not admit these firms are “not the greatest present threat” simply because breakup alone may be blunt, or because the state can also become censorious. That is a false refuge. The fact that public power can be abused does not legitimate private sovereignty. We do not answer one form of domination by normalizing another. But I have already said the remedy is not breakup simpliciter tomorrow morning. The priority is to end the illegitimate source of power: behavioral surplus. Outlaw its extraction. Abolish markets in future behavior. Impose structural separation where information infrastructure is fused with surveillance and ad-tech machinery. Require auditable accountability where algorithmic systems govern access to knowledge, association, and opportunity. If breakup follows from those principles, so be it. Habermas is right that constitutionalization matters. But I would press further: rights must reach upstream to the economic operations that produce epistemic inequality in the first place. Freedom cannot survive when human experience is rendered as raw material for others’ private means of modification. A democracy that hesitates to say no at that threshold is already in retreat.

If a breakup “tomorrow” would in fact intensify state censorship and diminish the innovative capacities on which a democratic public also depends, then I would not defend breakup as an abstract moral reflex. Law must not indulge in symbolic gestures when the foreseeable institutional effects are perverse. But, Peter Thiel, it does not follow from this prudential qualification that these firms cease to be among the gravest present threats to democratic freedom. That conclusion is a non sequitur. The priority, when these aims conflict, is to prevent domination — not because innovation is dispensable, but because innovation draws its democratic legitimacy only from social conditions under which citizens remain the authors, rather than the objects, of collective will-formation. Once communicative infrastructures are organized so as to steer attention, asymmetrically harvest data, and invisibly shape the horizons of public discourse, “experimental freedom” becomes ambiguous; it can name the freedom to innovate, but also the freedom of private actors to experiment upon populations. So I agree with Zuboff that domination must be blocked at its sources, though I would still insist that prohibitions must be translated into a constitutional architecture of rights, oversight, contestation, and public justification. The task is neither immediate dismemberment nor complacent tolerance. It is to de-privatize the conditions of public communication: interoperability, structural separation where necessary, transparency, due process, and strict limits on state-platform collusion. No, I am not prepared to admit that these firms are not a central threat. I am prepared only to insist that democracies act intelligently rather than theatrically.

If breaking them up tomorrow would predictably reduce innovation and strengthen state censorship, then no, democracies should not do it tomorrow. Politics is not improved by making a bad trade more moralistically. You do not save freedom by weakening civil society’s technological capacity and handing more leverage to the administrative state. But I also won’t say Google, Meta, and Amazon are harmless. That would be unserious. The question is comparative. Shoshana treats private platforms as the master key to history. Habermas treats constitutional procedure as if it can neutralize all concentrated power. I think both understate the danger of state-firm convergence. The worst scenario is not Big Tech alone. It is Big Tech fused with Washington, outsourcing censorship to companies and monopoly enforcement to regulators captured by incumbents. So the priority is not “break up now” versus “do nothing.” It is to break the feedback loop between monopoly and the state. That means real interoperability, data portability, constraints on government pressure over moderation, and antitrust targeted at exclusionary conduct—not antitrust as political theater. If a firm dominates because it built something genuinely better, that is very different from dominating through regulatory shelter and network lock-in. Zuboff is wrong that prohibition is the master remedy. Prohibitions scale bureaucracy. Bureaucracy then becomes the new sovereign. Habermas is wrong that legitimacy is mostly a matter of procedures of justification. A free society has to produce new things, not just new oversight boards. The greatest threat to democratic freedom is stagnation under centralized control—public, private, or most likely both together.