Power, Topology, the Dawn of AI and the End of Shared Governance, 2025–2028
There is a scene that recurs in the political science literature on regime change, drawn from Alexis de Tocqueville, from Crane Brinton, from Juan Linz: the moment when the erosion of norms, which has proceeded for years beneath the surface of institutional life, breaks through. It does not announce itself. The courts still meet. The legislature still convenes. The flag still flies. But the animating logic that gave those institutions their authority — the shared agreement that the rules apply to everyone, that outcomes must be legitimate even when they are unfavorable, that power is held in trust — has quietly departed. What remains is the scaffolding of a republic inhabited by the logic of something else.
We are in that scene now. And the question for those who study how power actually operates — not as civics-class abstraction but as a lived social force — is not whether something fundamental has changed. It plainly has. The question is what comes next, and what shape it takes between now and the end of 2028.
This attempts that analysis. It draws on two intellectual traditions that are rarely placed in conversation: the network sociology of Manuel Castells, whose work on the network society illuminated how control over informational flows reshapes social topology, and the structural realism of Kenneth Waltz, who insisted that the distribution of power in any system — domestic or international — shapes behavior in ways that transcend the intentions of individual actors. These two are taken in light of how technologically-advantaged elites have historically solved the governance problem of preventing both atomization and mass coordination — and how artificial intelligence may be making that ancient project newly scalable.
The synthesis of these perspectives points toward a trajectory that is neither apocalyptic nor reassuring. It is, in Waltz's sense, structural: the logic of the system produces outcomes that no single actor fully intends, and that prove very difficult to reverse once set in motion.
I. The Topology Has Already Changed
Castells argued, in his trilogy on the Information Age, that what mattered in the emerging network society was not simply who held power in the traditional sense — who controlled armies, treasuries, and courts — but who controlled the "switches" that connected networks together, and who had the power to "program" those networks with their values and interests. Political power in the network society was increasingly a function of narrative generation capacity, attention routing, and the ability to shape the epistemic defaults through which citizens understood their world.
This framework, developed in the late 1990s, has proven embarrassingly prescient. What we have witnessed since approximately 2015, and with accelerating intensity since January 2025, is a systematic capture of the major switches in the American political network. This has not occurred through a single dramatic seizure. It has occurred through a patient, overlapping series of structural transformations.
The federal judiciary, whose independence from political pressure was the foundational assumption of the entire constitutional order, now reflects a generation of ideologically curated appointments made with a clarity of purpose that the architects of judicial independence would have found alarming. The Supreme Court has, in a series of decisions culminating in its immunity ruling of 2024, substantially reconceived the relationship between the executive and the law — effectively insulating presidential conduct from criminal accountability in ways that shift the constitutional balance more than any formal amendment could have done. The effects will outlast any individual president by decades.
The Executive Branch, meanwhile, has discovered and exploited a zone of administrative authority that existing law left underspecified. The use of emergency powers declarations, impoundment of congressionally-appropriated funds, recess appointments made during contested legislative sessions, and the systematic replacement of career civil servants with political loyalists has produced an executive apparatus whose responsiveness to institutional constraint is qualitatively different from anything in the postwar period. This is not merely aggressive interpretation of existing authority. It is the construction of new facts on the ground that subsequent administrations will inherit as precedent.
Congress, for its part, has largely abdicated. The reasons are structural, not merely personal. The geographic sorting of the electorate into homogeneous partisan districts has made electoral survival contingent on appeasing a primary electorate that rewards ideological performance and punishes institutional loyalty. Members who might, in an earlier era, have defended congressional prerogatives against executive encroachment have instead calculated, correctly from a personal career standpoint, that the costs of crossing their party's executive are greater than the diffuse benefits of preserving institutional norms that their constituents neither understand nor value.
The result is a structure that Waltz would recognize. The distribution of effective power in the American system has shifted, and that shift creates new incentive structures that will shape the behavior of all actors in the system, including actors who did not intend or desire the shift.
II. The Network Architecture of Managed Pluralism
Here is not a prediction but a diagnosis of a mechanism that has been unfolding in the political domain long before it became a question of artificial intelligence.
The central insight is that historically, dominant classes survive when they solve two tensions simultaneously: preventing the atomization that breeds revolt, and preventing the unified mass coordination that threatens control. The sweet spot is what might be called "small-scale belonging without large-scale solidarity" — communities that are internally coherent and emotionally satisfying, but collectively unsynchronizable.
This is not a new observation, but the key is the mechanism by which it is achieved: not through crude propaganda or overt suppression, but through architecture. Control over community formation algorithms, discourse moderation norms, identity reinforcement feedback, cross-community exposure levels, and narrative salience weighting. These are not merely media tools. They are "social topology design" — the capacity to shape how humans cluster.
What is striking, when one applies this to the current American political landscape, is how completely this architecture has already been constructed — not primarily through AI, which is still arriving, but through the prior generation of social media platforms, curated information ecosystems, and the political entrepreneurs who learned to exploit them.
The American electorate does not experience itself as fragmented. Each fragment experiences itself as a coherent community with shared values, legitimate grievances, and clear villains. The MAGA ecosystem and the progressive ecosystem are not simply opposed; they are internally coherent and collectively unsynchronizable. They share no common epistemic ground from which coordinated political action against elite interests could be organized. Their grievances are real and often overlapping in their underlying causes — wage stagnation, housing unaffordability, institutional distrust, cultural displacement — but their framing of those grievances has been so thoroughly customized that cross-community solidarity is structurally impeded.
This as a feature, not a bug, of elite stabilization strategies. Whether it is intentional or emergent is, for structural analysis, beside the point. The Waltzian insight is that when a structure produces outcomes that serve powerful interests, it tends to be reproduced regardless of anyone's conscious design.
III. The Political Economy of Fracture: 2025–2028
Against this structural backdrop, what does the evidence suggest about the trajectory of American economic and political life through the end of 2028?
The economic dimension is where the structural contradictions are most acute. The tariff regime inaugurated in 2025 represents not merely a policy choice but a fundamental reorganization of the assumptions underlying the postwar international economic order — an order from which American capital, and through capital American workers, derived enormous benefits. The near-term effects are inflationary, supply-chain-disrupting, and recession-generating in ways that the heterodox economic nationalism undergirding the policy does not adequately account for.
More importantly, the political economy of the tariff regime creates a peculiar distribution of costs and benefits. The sectors and communities that were promised revival — manufacturing rust belt, extractive industries — face a timeline of industrial reorganization that extends well beyond 2028 even under optimal conditions. The costs of higher prices, reduced trade volume, and retaliatory measures are distributed broadly and immediately. The benefits, if they materialize, are concentrated and delayed.
This creates a structural political problem that the managed-pluralism architecture will struggle to contain. Economic pain that is both concrete and widespread tends to generate the kind of cross-community grievance that is hardest to manage through narrative differentiation. The architecture collapses when economic inequality becomes visible across villages. When your MAGA neighbor and your progressive neighbor are both struggling with the same gas and grocery prices, the narrative infrastructure that keeps them from coordinating faces its most serious test.
The Federal Reserve, already in a constrained position between inflation risk and growth risk, will find its independence tested further. The trajectory points toward a stagflationary environment — the worst political economy for an incumbent governing coalition, because it denies the comfort of being able to address one problem without worsening the other.
The institutional dimension will play out along two fault lines.
The first is the judiciary. The federal court system, from district courts through the circuit courts to the Supreme Court, will be the primary arena in which the redefinition of executive authority either gets ratified or contested. The current trajectory suggests ratification, not contestation. The legal theories that have expanded executive immunity, reduced the scope of the administrative state, and reconceived the unitary executive are sufficiently entrenched in enough judicial seats that significant reversal before 2028 is structurally improbable. What is more likely is a further elaboration of these doctrines — the construction of legal architecture that normalizes what was, as recently as 2020, considered outside the bounds of permissible interpretation.
The second fault line is the use of federal power against political opponents, institutions, and civil society organizations. The use of regulatory and prosecutorial authority as instruments of political consolidation has historical precedents in other democratic backsliders — Hungary, Turkey, Poland under PiS — and the structural logic, once begun, tends toward escalation. Each exercise of such power normalizes the next. The professional norms that once constrained such use — Justice Department independence, FBI separation from White House direction, the IRS's political neutrality — have been eroded enough that the remaining constraints are primarily personal, not institutional. Personal constraints, as any political scientist will tell you, are the least durable kind.
The social dimension is where Castells is most illuminating. The network topology of American political life is being permanently restructured. The platforms that mediate political communication have, through a combination of algorithmic logic and owner preference, shifted from a posture of managing political content to one of amplifying it along lines that favor certain kinds of engagement — outrage, tribal affirmation, conspiratorial pattern-finding — over others.
The emerging AI layer will accelerate this dynamic in ways that are only beginning to be visible. The coming capacity for narrative generation at scale and epistemic framing defaults as a form of structural power. We are at the threshold of that capacity becoming operationally real. Before the 2026 midterms, and certainly before the 2028 election cycle, AI-generated political content will be available at sufficient scale and sophistication to fundamentally alter the information environment within which citizens make political judgments.
The significance of this is not primarily about disinformation in the crude sense. It is about the industrialization of the managed-pluralism architecture. What previously required skilled human operators — the propagandists, the consultants, the social media strategists — will increasingly be automatable. The cost of maintaining the "virtual villages" that keep communities internally satisfied but externally non-coordinating will drop dramatically. This changes the structural economics of political control in ways that favor incumbents and disadvantage movements that depend on the organic generation of political solidarity.
IV. The Waltzian Prognosis: Structural Outcomes
Waltz was insistent that structure does not determine outcomes at the level of individual events. It shapes the range of likely outcomes, the incentives facing actors, and the probability distribution over possible futures. With that caveat, several structural predictions follow with reasonable confidence from the analysis above.
First, the institutional consolidation will continue. The incentive structures facing Republican elected officials, federal judges, and executive appointees all point in the same direction: toward accommodation of expanded executive authority rather than resistance to it. Resistance is personally costly and institutionally ineffective given the current distribution of power. This is the central Waltzian insight: actors in a system behave as the system's incentives direct them to, regardless of their personal preferences. The Republican senator who might, in a different structural context, defend congressional prerogatives will calculate that the personal and political costs of doing so exceed the benefits, and will accommodate. This prediction has been borne out repeatedly since 2017, and there is no structural reason to expect a different pattern.
Second, the Democratic opposition will remain fragmented and reactive. The managed-pluralism dynamic cuts both ways. The progressive coalition is itself fractured along lines — race, age, class, cultural politics — that make unified strategic action difficult. The 2026 midterms will offer an opportunity for electoral pushback, and the historical pattern of midterm losses by the party holding the White House favors Democratic gains. But gains in a gerrymandered House and in a Senate map that favors Republicans structurally will not produce the kind of veto-wielding majority that could check executive consolidation. What is more likely is a messy, contested legislative environment that produces gridlock rather than correction.
Third, the international dimension will feed back into domestic politics in destabilizing ways. The reorientation of American foreign policy — the weakening of NATO commitments, the transactional approach to alliances, the accommodation of authoritarian leaders — is not merely a policy choice. It is a restructuring of the international order that will have consequences extending well past 2028. European allies are already engaged in a fundamental reassessment of their security dependence on the United States. The medium-term result is a more multipolar security environment in which American influence is reduced. This creates political vulnerabilities that domestic actors will exploit, and foreign actors, most notably Russia, but also China and others, will investigate.
The Waltzian framework suggests that the international and domestic dimensions of power consolidation are not independent. Authoritarian-leaning executives have consistently employed external threat environments, whether fabricated or genuine, to rationalize internal consolidation. The structural incentive to do so will be present throughout the period in question.
Fourth, and most consequentially, the 2028 election will occur in a fundamentally different institutional context than any election in living memory. The mechanisms for contesting electoral outcomes, including the courts, the Justice Department, and the state-level administrative apparatus, will have been substantially reshaped by four years of appointments and norm revision. This does not mean that a contested outcome is inevitable or that democratic processes will fail. But it means that the structural safeguards that functioned, barely, in 2020 will be weaker, and that the actors who might exploit ambiguity will have both greater capacity and greater legitimacy — in the sense of normalized precedent — to do so than they did in the prior cycle.
V. The Failure Modes and the Question of Collapse
These are the conditions under which the managed-pluralism architecture collapses: when villages radicalize beyond the damping capacity of the system, when translation layers lose trust, when economic inequality becomes visible across tribal lines, and when elites over-optimize their segmentation strategies.
All four of these failure modes are present in the current American context, in incipient form.
The radicalization of political communities is not a background trend; it is the intended product of the information architecture. The problem, from an elite stabilization standpoint, is that this process has a momentum that is difficult to manage. The factions created to prevent mass coordination against elite interests can themselves become agents of destabilization when their grievances are sufficiently acute and their organizational capacity sufficiently developed. The January 6th insurrection was, in miniature, an example of this dynamic: a mobilized community that had been successfully oriented against a political opponent rather than against economic structures, but whose mobilization produced a kind of chaos that even its architects did not fully intend or control.
The translation layers — the shared institutional norms, the independent press, the professional associations, the nonpartisan civil service — that allow diverse communities to operate within a common framework are eroding. Their erosion is the product of deliberate strategy in some cases and structural pressure in others. But the result is the same: the ligaments of common political life are weakening, and the question of what holds the system together when they fail has no obvious answer.
The economic visibility problem is the most acute near-term risk. As noted above, the tariff regime and its downstream effects create a distribution of economic pain that cuts across the tribal segmentation of the political landscape. The political question is whether the narrative architecture — the managed pluralism — can sustain its community coherence in the face of concrete shared economic hardship. The historical evidence is mixed. Sometimes economic crisis produces the cross-tribal coordination that the structure is designed to prevent. More often, it produces intensified scapegoating within existing tribal frameworks, as communities look inward and downward rather than upward and outward for explanations of their distress.
VI. Conclusion: The Topology of What Comes Next
What does all of this add up to? Not, on the evidence of structural analysis, a clean break — neither a democratic restoration nor a formal authoritarian consolidation. The American system has too much institutional inertia, too many distributed power centers, and too much civil society resilience for either extreme. What it adds up to is something more unsettling: a stable-but-degraded equilibrium in which the forms of democratic governance persist while the substance is progressively hollowed out.
In Castells' terms: the switches are captured, the topology has been redesigned, and the network programs itself with values and distributions that serve a smaller and smaller circle of beneficiaries, while the nodes at the periphery experience strong local belonging and attenuated macro-political agency. People will feel heard within their communities and powerless within their government. Progress in lived meaning at the price of stasis in structural power.
In Waltz's terms: the distribution of power in the system has shifted, and it will require a countervailing shift — whether through electoral realignment, institutional recovery, economic crisis, or some combination — to produce a new equilibrium. Such shifts do occur. History is not linear and structures do not persist indefinitely. But Waltz's insight was that structure shapes the probability of outcomes, and the current structure does not favor rapid correction.
Perhaps AI could allow dominant groups to solve the ancient governance problem — how to let people feel deeply embedded without letting them unite. What the analysis above suggests is that this problem has already been substantially solved through prior-generation technology and political architecture. The AI layer that is now arriving will not create this dynamic. It will industrialize and scale it.
To understand the period between now and the end of 2028 is to understand that the American republic is not in a crisis that will resolve itself through the normal mechanisms of democratic accountability. It is in a structural transformation that those mechanisms were designed for a different era and are, by themselves, insufficient to address. The question is not whether this transformation will continue — structurally, it will. The question is whether the countervailing forces, such as economic pressure, institutional resilience, civil society mobilization, and the fundamental unpredictability of a free people, will be sufficient to generate a new and more equitable equilibrium before the transformation becomes self-reinforcing in ways that are very difficult to reverse.
History does not offer a confident answer. Structure suggests we should not be complacent waiting for one.
Comments
Post a Comment