Photo by Rob Curran on Unsplash
THE UNCERTAINTY DEPARTMENT
Program on Cognitive Infrastructure and Ambient Political Risk
Bureau of Deniable Externalities
Research Report UD-RR-2026-031 — April 2026
“The most cost-effective influence operation in history requires no covert funding, no cutout network, no HUMINT infrastructure, and no sanctioned liaison relationship with a compliant satellite service. It requires a Facebook account, approximately forty dollars in targeted advertising, and a domestic political environment whose regulatory appetite for intervention is, by any measurable standard, zero.”
— Internal working note, Bureau of Deniable Externalities, undated, classification status ambiguous, provenance unclear, existence disputed
This report documents what the Program on Cognitive Infrastructure and Ambient Political Risk has termed the Uncontrolled Access Condition: a structural state of the American information environment in which the capacity to conduct politically actionable manipulation of private citizens at scale is effectively available to any actor — domestic or foreign, state-sponsored or entrepreneurial, ideologically motivated or merely curious — at costs that are, in the language of barrier-to-entry analysis, negligible.
We further document what this report terms the Retroactive Leverage Accumulation problem: the phenomenon by which actors permitted to conduct such operations against the American population do not merely gain political influence over that population, but accrue structural leverage over the government that permits them to do so — a back-channel debt instrument whose terms are never written down but are, in the experience of this Department, invariably honored.
The combination of these two conditions produces what we describe in Section IV as Ambient Democratic Substitution: a state in which formal democratic processes — voting, representation, deliberation, consent — continue to be performed while their substantive content is determined elsewhere, by actors whose identities are unknown to participants, whose methods are invisible to the participants, and whose accountability to the participants is, at this time of writing, nonexistent.
Our findings are the following:
The barrier to entry for politically actionable population manipulation in the United States is, at present, lower than the barrier to entry for operating a food truck.
The institutional mechanisms theoretically available to address this condition have been systematically incapacitated through a combination of regulatory capture, jurisdictional ambiguity, operational deniability, and an apparent philosophical commitment on the part of relevant agencies to the position that awareness of a problem and responsibility for a problem are distinct categories with no necessary relationship.
Actors who conduct sustained manipulation operations against the American population accumulate a form of operational leverage over the government — specifically, the leverage that accrues to any party who knows, in documented detail, what the government knew, when it knew it, and what it chose not to do. This leverage is not hypothetical. It has market value. It is being spent.
The population upon which these operations are conducted is, by the design of the contemporary attention economy, structurally optimized for susceptibility. This is not a critique of the population. The population did not design the attention economy. The population was the product.
The title of this report claims that “anyone in the world” can conduct politically leverageable manipulation of American citizens. This claim will strike some readers as hyperbolic. We invite those readers to consider the following:
In 2016, a Russian troll farm operating out of St. Petersburg with a monthly budget of approximately $1.25 million reached an estimated 126 million Facebook users, organized rallies attended by actual American citizens, and generated political content whose influence on electoral outcomes remains, to this day, a subject of genuine scholarly dispute — which is to say, its influence was non-trivially above zero.¹
By 2026, the tools available to replicate this operation have improved considerably. The costs have decreased. The sophistication required has diminished. The barriers that existed in 2016 — primarily the need for human labor to generate content at scale — have been substantially automated. What required a troll farm employing hundreds of people now requires a subscription and an afternoon.
When this report says “anyone,” it means:
The list does not include a significant threshold of capability, resources, or institutional affiliation. It includes, to a reasonable approximation, anyone who has thought about doing this for more than a few hours.
This report does not characterize the American population as uniquely gullible. The American population is approximately as gullible as any other large population that has been subjected, for three decades, to an information environment deliberately engineered to maximize engagement over accuracy, emotional salience over factual coherence, and tribal affiliation over independent judgment.
The attention economy did not produce gullible citizens accidentally. It produced optimally susceptible users as its primary output, because optimally susceptible users generate more engagement, more data, more behavioral prediction surface area, and more advertising revenue than citizens who maintain sustained critical evaluation of the information they consume.
We note, for the record, that the regulatory environment that permitted the attention economy to produce this output without constraint was itself a product of lobbying activity conducted by the same industry that benefited from the output. We note this not to be exhaustive but because the circularity is load-bearing for the argument of this report, and we prefer our load-bearing elements visible.
The population is susceptible in the following operationally relevant ways:
Affective Processing Primacy. Information that produces strong emotional responses — fear, outrage, tribal solidarity, contempt — is processed faster, retained longer, and shared more widely than information that does not. This is a biological fact that is, from the perspective of influence operations, an architectural specification.
Source Epistemological Collapse. Decades of sustained delegitimization of institutional information sources — conducted simultaneously by state actors, domestic political operatives, and the platforms themselves — have produced a population with diminished capacity to evaluate source credibility and heightened susceptibility to claims that “official” accounts are themselves manipulated. The apparatus for detecting manipulation has been manipulated.
Personalized Reality Segmentation. The algorithmic curation of information environments means that the manipulation of individual citizens does not require broadcasting to the general population. It requires only the construction of a personalized information environment specific to the psychological profile of the target. This is a service commercially available from multiple vendors.
Accumulated Anomie. Extended exposure to institutional failure, unaddressed grievance, and the lived experience of being governed by systems that are visibly indifferent to one’s interests produces a population in a chronic state of low-grade political dissociation. Dissociated populations are not more resistant to manipulation. They are less.
We approach this section with some delicacy, not because the finding is uncertain — it is not — but because the precise mechanism of the government’s non-protection has implications that extend beyond editorial comfort.
The United States government is aware that the American population is subject to large-scale, politically actionable influence operations conducted by foreign and domestic actors. This awareness is documented in congressional testimony, intelligence assessments, academic literature, investigative journalism, and the self-promotional materials of the firms conducting the operations. The awareness is not the gap.
The gap is between awareness and intervention, and it is maintained by the following interlocking mechanisms:
Jurisdictional Fragmentation. No single agency has clear authority over the full scope of the problem. The FTC has some authority over deceptive commercial practices. The FEC has some authority over campaign finance. The FCC has some authority over broadcast media. The FBI has authority over foreign intelligence operations. The DHS has some authority over election security. The NSA has authority it does not discuss. Each agency’s authority ends where another’s begins, and the operations being conducted operate specifically in the gaps.²
The First Amendment Latency Problem. A substantial portion of the manipulation infrastructure is, in its individual components, constitutionally protected speech. The amplification of false information is not prohibited. The algorithmic prioritization of emotionally manipulative content is not prohibited. The micro-targeting of psychologically vulnerable individuals with customized disinformation is not prohibited. What emerges from the combination of individually legal actions is a capability set that, in the judgment of this Department, is incompatible with the premise of informed democratic consent — but this judgment is not shared by the relevant courts, at least not yet, and regulatory action in anticipation of court agreement has not historically been the preferred posture of agencies that prefer to remain funded.
The Revolving Infrastructure. The staff most capable of understanding, designing, and disrupting influence operations at scale are employed, at wages considerably above government scale, by the firms conducting or enabling those operations. The government does not have a structural solution to this problem. It has, periodically, asked nicely.
The Strategic Ambiguity Dividend. Some portion of the foreign influence operations conducted against the American population produce outcomes that are, from the perspective of some domestic political actors, favorable. Interventions that would curtail foreign manipulation would also curtail the domestic political advantage the manipulation produces. This creates an incentive structure in which the most politically powerful potential advocates for intervention are also the most politically motivated to obstruct it.
We note that this observation describes a structural condition, not a conspiracy. The absence of conspiracy makes the condition more durable, not less.
For purposes of clarity, we will define Politically Actionable Manipulation (PAM) as any intervention in the information environment of a private citizen that, without the citizen’s awareness or consent, shifts that citizen’s political beliefs, preferences, or behaviors in a direction that serves the operational objectives of the intervening actor.
The cost of delivering a PAM event to a single citizen is, in current market conditions, approximately as follows:
| Intervention Type | Cost Per Targeted Citizen | Deniability | Scalability |
|---|---|---|---|
| Social media micro-targeted ad | $0.01–$0.50 | High | Unlimited |
| Algorithmic feed manipulation (via platform incentives) | $0.00 (platform self-executes) | Total | Unlimited |
| AI-generated personalized content seeding | $0.001–$0.05 | High | Unlimited |
| Astroturfed community engagement | $0.10–$2.00 | Medium | High |
| Coordinated inauthentic behavior campaign | $0.05–$0.50 | High | Very High |
| Direct behavioral targeting via data broker acquisition | $0.50–$5.00 | High | High |
Table 1. Approximate unit costs of politically actionable manipulation delivery, April 2026. Figures represent retail-equivalent pricing; volume discounts available. Platform commissions not included.
The cost per changed vote, per shifted opinion cluster, or per behavioral modification event is, naturally, higher than the cost per contact, and varies with the sophistication of the campaign and the initial resistance profile of the target population. However, this Department notes that the relevant comparison is not “cost of manipulation” versus “cost of some idealized free civic deliberation.” The relevant comparison is “cost of manipulation” versus “cost of the political outcomes manipulation can produce.”
By that comparison, manipulation is the cheapest political intervention in recorded history.³
The manipulation economy has the following market structure, which we present in the spirit of investor orientation:
Supply Side: Platforms, data brokers, advertising technology firms, and AI content generation services provide the raw infrastructure. This infrastructure is not primarily designed for political manipulation; it is designed for commercial behavioral modification, which is the same thing with different declared objectives.
Demand Side: Any actor — state, commercial, political, ideological, personal — with a behavioral change objective and a budget above the cost of a modest dinner.
Market Clearing Mechanism: Auction-based advertising systems, API access, and the open availability of targeting data from data brokers whose information provenance, collection consent, and end-use restrictions are, in the words of a recent FTC filing we decline to cite specifically, “challenging to characterize.”
Regulatory Framework: The regulatory framework governing this market is best described as aspirational. It is the framework that would exist if all agencies had clear authority, adequate resources, political cover for enforcement, and a shared theory of what the harm actually is. None of these conditions presently obtain simultaneously.
Barriers to Exit: For established platforms and data infrastructure firms, regulatory barriers to exit from the manipulation economy are effectively zero, since the manipulation economy is not legally distinguishable from the advertising economy, and exiting the advertising economy is not a going concern position.
We observe that this market structure is not an accident. It was constructed over several decades through deliberate lobbying, regulatory capture, jurisdictional design, and the capture of the technical and legal expertise necessary to regulate it by the firms it might regulate. We do not describe this as a conspiracy because it did not require coordination. It required only that each actor, at each decision point, choose the option that served their immediate institutional interest. The resulting structure is as stable as any arrangement that serves the immediate institutional interests of all participants simultaneously.
The population whose beliefs and behaviors are being modified by the operations described in this report is, in virtually all cases, unaware that the modification is occurring. This is not an incidental feature of the operations. It is the central operational requirement without which the operations do not function.
This asymmetry — between actors who understand and deploy the behavioral architecture of the contemporary information environment and citizens who inhabit that architecture without awareness of its design — is what this report terms the Technocratic Gaming Condition.
The Technocratic Gaming Condition does not require malevolence. It requires only:
All four conditions obtain simultaneously in the current environment. The Technocratic Gaming Condition is therefore not a risk. It is the operational status.
We propose the following observation, which we term the Accumulated Influence Theorem with more confidence in its descriptive accuracy than in its status as a theorem:
The political influence available to any actor through sustained, low-intensity manipulation of a sufficiently large population over a sufficiently long period is, in expectation, greater than the influence available through any single high-intensity intervention, and is significantly more durable, more deniable, and more resistant to correction.
The reasons for this are structural:
Belief Formation Lag. Political beliefs formed through accumulated ambient exposure to consistent messaging are experienced by the holder as self-generated conclusions, not as the product of external influence. This is the most significant distinction between manipulation and persuasion: persuasion is visible to its recipient; manipulation, by design, is not. Beliefs experienced as self-generated are substantially more resistant to correction than beliefs experienced as externally supplied.
Behavioral Normalization. Over time, populations whose information environment has been systematically managed begin to treat the managed environment as the baseline reality against which other information is evaluated. The managed environment becomes the epistemological default. Departures from it are experienced as anomalous.
Compound Returns. Political beliefs do not exist in isolation. They form networks of mutually reinforcing positions. Sustained manipulation that shifts an individual’s position on one issue tends, over time, to shift their positions on correlated issues, because the underlying value architecture has been modified, not merely the surface-level opinion. This is the manipulation equivalent of compound interest.
Detection Resistance. No single manipulation event is likely to trigger detection or remediation. The operational security of ambient influence operations rests on the invisibility of any individual interaction. The cumulative effect of ten thousand invisible interactions is visible only in aggregate behavioral and political data — which is available primarily to the actors conducting the operations.
The Accumulated Influence Theorem implies that the most significant manipulation of democratic processes does not occur through dramatic interventions detectable by oversight mechanisms. It occurs through patient, distributed, sustained modification of the ambient information environment — the political equivalent of geological processes, operating at timescales and scales of diffusion that render event-based oversight frameworks essentially irrelevant.
We note, without prejudice to the reader’s conclusions, that the category of actors who have both the technical capability and the political motivation to exploit the Technocratic Gaming Condition is not limited to foreign adversaries.
It includes:
We observe that one distinguishing feature of the current environment is that the actors in the last three categories are not in violation of existing law. They are operating within it. The law was not designed to accommodate them; it was designed before they existed, and has not been revised to account for their emergence. This is what the Bureau of Deniable Externalities has termed Regulatory Temporal Displacement: the condition in which the legal framework governing an activity was designed for a previous version of that activity and has not been updated because the actors who benefit from the gap are the actors best positioned to prevent the update.
Democratic theory rests on the premise of informed consent: the idea that citizens, with access to accurate information, form genuine political preferences that are then aggregated through electoral processes into legitimate governance outcomes.
This premise has always been a simplification. Citizens have never had access to fully accurate information. Preferences have never been fully independent of social influence. Aggregation mechanisms have always contained structural biases. The gap between the democratic premise and democratic practice is not new.
What is new is the scale and precision with which the gap can be engineered.
The contemporary information environment does not merely produce incidentally imperfect conditions for democratic deliberation. It produces, at scale, personalized information environments specifically calibrated to the psychological profiles of individual citizens, whose political outputs can be modeled in advance and whose behavioral responses to specific stimuli can be A/B tested in real time across populations of millions.
This is not a failure of democracy. It is the replacement of democracy’s epistemic preconditions with an alternative architecture that performs the formal operations of democracy — voting, polling, public opinion — while their substantive inputs are determined elsewhere.
We term this condition Ambient Democratic Substitution (ADS): the replacement of the substantive content of democratic deliberation with managed preference outputs, while preserving the formal mechanisms of democratic process.
The significance of ADS is not that elections are stolen. The significance is that elections are, in the ADS condition, the least interesting part of the political influence process. By the time a citizen votes, their preference has been substantially formed by months or years of ambient manipulation that they did not notice and cannot audit. The vote registers a preference. The preference was engineered. The engineering was invisible. The outcome is legitimate by every formal metric.
We anticipate the following objection: that the preceding analysis is unfalsifiable, because it attributes political outcomes to invisible manipulation rather than to genuine preference formation, and that there is no reliable method for distinguishing between the two.
This objection is correct. There is no reliable method, available to private citizens or most oversight bodies, for distinguishing between genuinely formed political preferences and preferences formed through sustained covert manipulation of the information environment.
We note that this is also the operational answer. The manipulator’s defense and the impossibility of citizen detection are the same fact.
We arrive at what this report considers its most significant analytical contribution, which is less a finding than an observation about a structural relationship that has received insufficient attention in the policy literature, possibly because its implications are uncomfortable for the institutions the policy literature is typically written to advise.
When a government body — whether through formal inaction, informal tolerance, regulatory forbearance, or simple incapacity — permits a class of actors to conduct operations against its own population, the actors conducting those operations do not merely gain influence over the population. They gain structural leverage over the government itself.
The mechanism is as follows:
Phase 1 — Operation. A foreign or domestic actor conducts manipulation operations against the American population. The operations are not disrupted.
Phase 2 — Documentation. The actor accumulates documented evidence of what they did, at what scale, with what results, and — critically — what the relevant government agencies knew about it and when they knew it.
Phase 3 — Leverage Crystallization. The actor now holds, as a structural asset, a detailed record of government non-intervention in operations that, had they been disclosed fully and publicly, would have generated significant political and legal consequences for the non-intervening agencies and their political supervisors.
Phase 4 — Leverage Expenditure. The asset is spent: not necessarily through explicit blackmail, which is both illegal and unnecessary, but through the ordinary operation of institutional self-interest. Agencies that have permitted operations do not subsequently regulate them aggressively, because aggressive regulation generates disclosure processes that reveal prior non-regulation, which generates accountability consequences that the agencies correctly assess as worse than continued non-regulation.
The formal name for this mechanism, in the literature of game theory and institutional economics, is a Commitment Trap: a state in which an actor’s past behavior constrains their future behavior by making certain options — specifically, the options that would require acknowledging or reversing the past behavior — prohibitively costly.
We term the specific variant documented in this report Retroactive Leverage Accumulation (RLA) to capture the feature that the leverage accrues automatically, without explicit negotiation, as an inescapable byproduct of the government’s own decision to permit rather than interdict.
Standard regulatory capture theory describes a process in which regulated industries gradually, over time, come to exert disproportionate influence over the agencies regulating them — through lobbying, revolving door hiring, information asymmetry, and the co-optation of expert networks.
RLA describes a different mechanism that is, in some respects, more efficient: the automatic accrual of leverage by actors operating under government forbearance, without the investment of regulatory capture activity.
The distinction matters because it implies that:
The leverage compounds without active management. Each additional period of non-intervention generates additional documentation of non-intervention. The leverage does not depreciate. It appreciates.
The leverage is available to actors who did not seek it. Foreign influence operators who had no intention of accumulating leverage over American regulatory agencies are, as a structural consequence of operating in the current environment, accumulating it. The leverage is a byproduct of the operation, not an objective.
The leverage is non-transferable but permanently available. The leveraged party — the non-intervening agency — cannot retire the obligation by changing leadership, issuing new policies, or making prospective commitments to future enforcement. The documentation of past non-intervention remains. Future enforcement generates disclosure risk that creates liability for the past.
The leverage operates through institutional self-preservation, not coercion. No explicit threat is required. The mechanism functions through the ordinary operation of bureaucratic self-interest: agencies and individuals within agencies assess their disclosure risk and adjust their enforcement posture accordingly. The adjustment is experienced, internally, as a reasonable assessment of regulatory priority, not as capitulation to leverage. This subjective experience does not change the structural analysis.
This report does not have access to classified information regarding the documented scope of influence operations against the American population, the documented awareness of same by relevant agencies, or the documented history of non-intervention decisions and their stated rationale.
We observe that the preceding sentence describes the precise information that would be required to perform a complete RLA inventory.
We further observe that the parties with access to the relevant classified information are precisely those who have the most significant interest in the RLA inventory remaining incomplete.
We close this section by noting that “the parties with access to the relevant information have a structural interest in preventing its disclosure” is a description that, in this Department’s experience, tends to describe situations where the relevant information is worth disclosing.
This report has established the following:
The American population is, as a structural condition of the contemporary information environment, subject to ongoing, distributed, politically actionable manipulation by a wide and non-exhaustive range of foreign and domestic actors.
The cost of conducting such manipulation is, in practical terms, accessible to any actor with modest resources and basic technical literacy.
The regulatory and institutional mechanisms theoretically available to address this condition are systematically incapacitated by a combination of jurisdictional fragmentation, regulatory capture, legal architecture designed for a previous information environment, and the strategic interest of incumbent political actors in selective non-enforcement.
The cumulative effect of sustained ambient manipulation is the effective substitution of managed preference outputs for genuine democratic deliberation, while preserving the formal mechanisms of democratic process — a condition we have termed Ambient Democratic Substitution.
Actors permitted to conduct operations against the American population automatically accumulate structural leverage over the government that permits them, through the mechanism of Retroactive Leverage Accumulation — a self-reinforcing dynamic in which prior non-intervention constrains future enforcement, ensuring that the longer the condition persists, the more durable it becomes.
This Department does not make policy recommendations in cases where the institutional conditions required to implement recommendations do not exist, and where the act of making recommendations might create the misleading impression that recommendations are what the situation is waiting for.
We will observe, without prejudice:
We further observe that this report is unclassified, publicly available, and will be read primarily by people who already knew most of what it contains. We do not consider this a failure of the report. We consider it a data point about the relationship between knowing things and being in a position to do anything about them — a relationship this Department has been studying for some time, with findings that are, at this point, moderately discouraging.
Ambient Democratic Substitution (ADS): The replacement of the substantive epistemic content of democratic deliberation with managed preference outputs, while preserving the formal mechanisms of democratic process. Distinguished from “electoral fraud” by its legal status (ADS is not illegal), its scale (ADS operates on populations, not vote counts), and its deniability (ADS has no paper trail, because the paper is the entire information environment).
Accumulated Influence Theorem: The observation that sustained low-intensity manipulation of a sufficiently large population over a sufficiently long period produces greater and more durable political influence than high-intensity single interventions. Not a theorem in the mathematical sense; described as such to provide the finding with the credibility appropriate to its importance.
Politically Actionable Manipulation (PAM): Any intervention in the information environment of a private citizen that, without the citizen’s awareness or consent, shifts that citizen’s political beliefs, preferences, or behaviors in a direction serving the operational objectives of the intervening actor. Distinct from “advertising” primarily in that the word “advertising” does not carry policy implications.
Regulatory Temporal Displacement: The condition in which the legal framework governing an activity was designed for a previous version of that activity and has not been updated, because the actors who benefit from the resulting gap are best positioned to prevent the update. A specific and common form of regulatory capture that does not require active lobbying; it requires only institutional inertia and a rapidly changing environment.
Retroactive Leverage Accumulation (RLA): The automatic accrual, by actors operating under government forbearance, of structural leverage over the government — specifically, leverage derived from the government’s own documented awareness and non-intervention. Distinguished from “blackmail” by its passive nature (no explicit threat required), its legal status (the leverage is derived from accurate documentation of government conduct), and its mechanism of effect (institutional self-preservation rather than coercion).
Technocratic Gaming Condition: The structural asymmetry between actors who understand and can deploy the behavioral architecture of the contemporary information environment and citizens who inhabit that architecture without awareness of its design or operation. The Technocratic Gaming Condition does not require intent; it requires only differential knowledge of the rules of an environment in which most participants do not know they are playing.
Uncontrolled Access Condition: The structural state of the American information environment in which the capacity to conduct politically actionable manipulation of private citizens at scale is available to any actor with modest resources and basic technical literacy, at costs substantially below the costs of traditional political organizing, legal political advertising, or any other formally recognized form of political influence.
The Uncertainty Department’s standard methodology for reports of this type involves the following steps:
This methodology has been refined over several report cycles. We are not able to report that it has produced remedial action. We are able to report that the documentation is very good.
¹ The figures cited here are from the Senate Intelligence Committee’s investigation, the Mueller Report, and the Internet Research Agency indictment. We cite the public record here rather than classified intelligence assessments, partly because the former is accessible to our readership and partly because the latter, in the view of this Department, would not substantially alter the analysis.
² We observe that the operations being conducted do not operate in the gaps accidentally. The legal architecture of the American information environment was sufficiently legible, by 2012, that sophisticated actors could design operations specifically to exploit its structural absences. The gaps did not produce the operations. The operations were designed for the gaps.
³ A partial exception may be the printing press, which enabled mass political manipulation at costs comparable (inflation-adjusted) to the present, but which operated at substantially slower velocity and without the micro-targeting capability of algorithmic systems. We note this comparison not to minimize the significance of Gutenberg but to contextualize the novelty of the present situation, which is characterized not by the availability of mass manipulation but by the combination of mass reach with individual precision — the capacity to deliver personalized manipulation at population scale simultaneously.
THE UNCERTAINTY DEPARTMENT — Program on Cognitive Infrastructure and Ambient Political Risk
This report may be freely distributed. The Uncertainty Department holds no intellectual property claim over findings that constitute a description of current reality. We encourage reproduction, citation, and the kind of widespread dissemination that produces informed concern without institutional response, which is the characteristic output of this field.
Correspondence: Bureau of Deniable Externalities, care of the institution that commissioned this report, which has not yet decided whether to endorse its conclusions, which is itself a data point the report anticipates.
UD-RR-2026-031 // End of Document
THE UNCERTAINTY DEPARTMENT
Program on Cognitive Infrastructure and Ambient Political Risk
Subcommittee on Tactics That Have Documentary Precedent
Addendum UD-RR-2026-031-ADD-A — April 2026
“The objective of the program was not the arrest of the subject. Arrest requires evidence, trial, and a public record. The objective was the neutralization of the subject — the systematic reduction of their capacity to act, organize, communicate, and be believed. Neutralization by these means produces no trial. It produces no public record. It produces a person who is progressively less capable of functioning and whose account of why that is remains, in the eyes of any institutional observer, unverifiable.”
— Paraphrase of operational logic documented in the Church Committee hearings, 1975. Quoted here because its description of the logic remains accurate and its authors believed its exposure would prevent recurrence, which was optimistic.
The parent report documented the mass-market end of the democratic subversion economy: the infrastructure, pricing, and structural conditions through which any actor with modest resources can conduct politically actionable manipulation of the American population at scale.
This addendum documents the premium tier.
Where mass manipulation operates through impersonal infrastructure — algorithmic amplification, micro-targeted advertising, coordinated inauthentic behavior networks — the individuated operations described here are targeted at specific persons, conducted through human proxies operating in the physical environment of the target, and designed to achieve outcomes that digital manipulation cannot: the systematic modification of a specific individual’s lived circumstances, social relationships, psychological state, professional standing, and capacity to be believed.
The operations described in this addendum are not theoretical. They have documented historical precedents. They have known names in the operational literature. The Church Committee documented their use by the FBI under COINTELPRO. The Stasi’s Zersetzung program documented their use by the East German state against domestic dissidents. The difference between those historical programs and the operations described in this report is not methodological — the methods are largely the same. The difference is operational architecture: contemporary deniable operations are designed to exploit the specific legal, institutional, and evidentiary gaps of the post-Church Committee, post-Stasi, commercially-mediated environment.
We document them here because their omission from the parent report created an incomplete picture of what “manipulation of citizens for political purposes” actually encompasses at the individuated level. The mass manipulation infrastructure moves public opinion. The individuated operations remove specific people from the field.
Both serve the political interests of actors who need neither to be named nor to formally coordinate to benefit from the result.
The operations described in the parent report and the operations described in this addendum serve different purposes and target different populations.
Population-level operations seek to move aggregate political preferences, suppress or amplify specific political identities, normalize particular narratives, and degrade the epistemic infrastructure through which citizens evaluate political claims. The target is the distribution of opinion across a population. No individual citizen is selected for attention; all are addressed simultaneously through shared infrastructure. The individual citizen’s experience of these operations is indistinguishable from the ordinary operation of the information environment, because the operations are conducted through the ordinary information environment. This is the operationally desirable feature.
Individual-level operations seek to neutralize specific persons — to reduce a particular individual’s capacity to organize, communicate, be credible, maintain social and professional relationships, or sustain the psychological stability required to function as a political actor. The target is a person, not a distribution. The methods are physical, social, and environmental in addition to informational. The experience of these operations, from the subject’s perspective, is significantly more disturbing than ambient digital manipulation — but is designed to be, in the view of any external observer, indistinguishable from ordinary bad luck.
The two tiers are complementary. Mass operations shape the environment in which individual operations occur. Individual operations remove from that environment the specific actors most capable of diagnosing or resisting the mass operations. The combination is, in the judgment of this Department, more capable than either tier in isolation.
The selection of specific individuals for individuated operations follows a logic that is, in retrospect, identifiable across historical and contemporary cases. The following categories are not exhaustive but represent the highest-frequency targets in documented programs:
Credible Institutional Witnesses. Individuals who have direct, documented knowledge of institutional misconduct and who are positioned — through professional standing, legal access, or platform — to make that knowledge consequential. The threat is not merely that they know something. Many people know things. The threat is that they can make what they know legible to institutions capable of acting on it.
Effective Organizers. Individuals whose capacity to coordinate collective action — through community organizing, legal advocacy, political communication, or social network activation — represents a scalable threat to the political interests of the operating actors. The individual is targeted not for their individual capacity but for their network capacity. Neutralizing the organizer degrades the network without requiring the network to be addressed directly.
Epistemically Disruptive Individuals. Persons whose analytical frameworks, investigative methodologies, or communicative approaches pose specific threats to the narrative architectures that the operating actors depend on. This category includes investigative journalists, certain academics, researchers whose work could produce accountability documentation, and individuals who have developed original frameworks for analyzing the operations themselves. The threat is the framework, not merely its current applications.
Leverage-Adjacent Witnesses. Individuals who are positioned — through professional relationships, legal proceedings, administrative access, or geographic proximity to relevant operations — to accumulate documentation that could crystallize into legal or institutional liability for specific actors. These subjects may not yet have actionable information. They are neutralized prospectively, before the accumulation occurs.
Emerging Political Infrastructure. Individuals who represent potential nodes in emerging political accountability infrastructure: not yet powerful enough to pose immediate threat, but with the trajectory and capacity to become so. The early intervention logic: neutralization is significantly cheaper and more deniable before the subject has established the institutional relationships, platform, and credibility that make neutralization operations more complex and more risky.
The East German Stasi’s Zersetzung program — documented in internal operational guidelines known as Directive 1/76, and analyzed extensively in the post-reunification Stasi records review — established the operational logic that contemporary individuated targeting inherits.
Zersetzung (decomposition, or corrosion) operated on the principle that the most effective neutralization of a political subject does not require imprisonment, violence, or formal legal action — each of which produces a public record, a legal proceeding, and a potential martyr. More effective, the Stasi determined, was the systematic corrosion of the subject’s lived environment: their relationships, their professional situation, their sense of reality, and — critically — their capacity to be believed when they described what was happening to them.
The specific techniques documented in Directive 1/76 and subsequent Stasi operational records include:
The Stasi employed approximately 91,000 full-time officers and between 170,000 and 300,000 unofficial informants to conduct these operations across a population of approximately 16 million. The ratio of operational personnel to population was, in absolute terms, extraordinary. The program was effective. The population was extensively managed. Its exposure came only with the collapse of the state that housed it.
We document this history not to suggest that contemporary operations are conducted at Stasi scale or under Stasi institutional architecture. We document it because Directive 1/76‘s operational logic — the corrosion of lived environment as political neutralization — has not been superseded. It has been adapted.
The contemporary adaptation of individuated ambient manipulation differs from its Stasi precedent in three structural respects:
Deniability Architecture. Zersetzung operated within an authoritarian state where the program’s existence, while not publicly acknowledged, was institutionally housed in a formal state apparatus. Contemporary operations are designed for a democratic environment with formal legal protections, an adversarial legal system, and formal oversight mechanisms. The deniability architecture is therefore not merely bureaucratic — it is the operational core. Contemporary operations are designed so that each individual component is explainable as coincidence, ordinary social conflict, market dynamics, or the subject’s own behavior. The pattern that constitutes the operation is visible only in aggregate. The aggregate is deniable by any individual component actor.
Proxy Infrastructure. Zersetzung operated through formal state employment — official officers and registered unofficial collaborators whose relationship to the program was documented (and later discovered). Contemporary operations use proxy structures that are specifically designed to prevent attribution: commercial intermediaries, political organizations, community networks, private individuals motivated by personal grievance, and automated systems whose instructions have been laundered through enough intermediaries that no single actor can be shown to have directed the whole. The result is what the parent report’s field manual called Jurisdictional Fog: a distribution of operational activity across enough independent actors that no single actor constitutes a legally actionable principal.
Commercial Mediation. Contemporary operations are conducted through, and substantially enabled by, commercial infrastructure that is entirely legal: employment background check services, landlord screening databases, social media platforms, data broker repositories, and the ordinary operation of algorithmic reputation systems that aggregate and amplify negative signals regardless of their provenance. The subject’s ruined professional relationship, adverse landlord report, and unfavorable background check result are each the product of a legal commercial transaction. Each is individually explainable. The pattern they form is not accidental. The pattern is not provable.
The contemporary environment offers a specific operational capability that neither COINTELPRO nor Zersetzung could access at comparable scale: the subject’s ambient environment — the physical and social world they move through daily — has been sufficiently instrumented, datafied, and networked that it constitutes an operational platform available to actors who know how to use it.
Consider the following environmental systems available to a sufficiently motivated and connected actor:
The Physical Neighborhood. The subject’s residential environment can be populated, over time, with individuals in varying relationships to an operation — from informal social contacts whose interactions are calibrated to specific behavioral objectives, to commercial services in the subject’s routine (building management, local businesses, service contractors) whose staff can be influenced or whose commercial relationships with the subject can be managed. The subject experiences this as their neighborhood. They do not experience it as an operational environment. This discrepancy is the operational asset.
Transit and Movement Corridors. The subject’s regular movement patterns — commute routes, frequent destinations, routine travel — constitute predictable corridors in which ambient encounters can be staged. The staging of ambient encounters in transit corridors is what this Department’s legislative satire has termed vehicular itinerary impersonation and compound ambient harassment events (CAHE): the coordinated deployment of proxies in physical spaces through which the subject predictably moves, for purposes ranging from psychological pressure to manufactured evidentiary association. Each staged encounter is, individually, a stranger in a public place. The pattern, over weeks or months, is something the subject experiences and cannot demonstrate.
Professional Environment. The subject’s professional network — employers, clients, collaborators, licensing bodies, professional associations — is accessible through legitimate commercial channels. Background check services aggregate and sell professional history. Reference check services solicit and transmit professional opinions. Anonymous communication channels (professional review platforms, informal networks, social media) enable the introduction of reputation-damaging information without attribution. The cumulative effect of adverse professional information arriving through multiple legitimate commercial channels is professionally devastating. Each adverse signal has a source that is either anonymous or legitimately independent. No actor has “harassed” the subject. The subject has received a commercially aggregated negative professional reputation, distributed through legal channels, whose origins cannot be traced to any coordinated source.
Social Network. The subject’s personal relationships — family, friends, romantic partners, community affiliations — are accessible through ordinary social dynamics: mutual acquaintances, shared spaces, social media, and the organic-appearing cultivation of social contacts who serve operational objectives. The introduction of targeted information into a subject’s social network — information calibrated to damage specific relationships, introduce suspicion, or create the social isolation that characterizes Zersetzung at scale — requires no unusual technical capability. It requires only access to the subject’s social environment and the willingness to use it.
Institutional Interfaces. The subject’s regular interfaces with institutions — administrative agencies, courts, licensing bodies, medical providers, financial institutions, housing authorities — each constitute contact surfaces where friction can be introduced through legitimate means. The subject who experiences unusual delays, processing errors, adverse determinations, and administrative complications across multiple institutional interfaces simultaneously is experiencing what they cannot prove: coordinated interference delivered through the ordinary dysfunction of bureaucratic systems that are, individually, simply bureaucratic.
The parent report documented the political utility of mass digital manipulation for aggregate preference modification. This addendum documents a complementary utility: what individuated operations accomplish that mass manipulation cannot.
Subject-Specific Silencing. Mass manipulation shapes what populations believe. It cannot silence a specific credible individual whose testimony, documentation, or analytical framework poses a specific threat to a specific actor. Individuated operations can. The distinction matters when the political threat is not a diffuse public opinion but a person with documented evidence, legal standing, or platform sufficient to create specific accountability consequences.
Credibility Pre-emption. An individual who has been subjected to months or years of ambient environmental disruption, professional interference, and social network degradation presents to the world in a condition of reduced credibility that is self-explaining. They have had difficulties. Their professional record has gaps or adverse entries. Their social relationships have strained or collapsed. Their behavior, in the context of the disrupted environment, has been erratic or extraordinary. When they describe coordinated targeting, the audience evaluates this description in the context of the full picture of their circumstances. The full picture has been curated to make the description appear to be the output of a compromised mind rather than the accurate account of documented events. This is what the corpus has termed credibility pre-positioning: the installation of the audience’s skepticism before the subject’s account exists.
Epistemic Isolation. The individual whose ability to form and maintain social relationships has been systematically degraded is epistemically isolated: they lack the informal verification networks through which most people calibrate their perception of events. Their account of their experience cannot be corroborated because the social infrastructure through which corroboration would flow has been deliberately damaged. They experience a reality that is genuinely different from the reality that external observers can access. This gap — between the subject’s experienced reality and the observer-accessible reality — is the primary deniability mechanism of the operation. It is also, from the subject’s perspective, among its most psychologically devastating features.
Biography Fracture. This Department has previously defined biography fracture as the condition in which a subject’s life history contains years they cannot explain in terms that will be credited by institutional audiences. Extended individuated operations produce biography fracture as a structural output: the years during which the operation was conducted become years characterized by professional disruption, social instability, anomalous behavior, and unaccountable deterioration. The subject cannot explain these years accurately because the accurate explanation — coordinated targeting — is the explanation that is credibly unavailable to them given the circumstances of their presentation. The biography fracture is not incidental damage. It is a durable structural outcome that limits the subject’s future institutional access, professional opportunities, and credibility indefinitely.
The parent report introduced the concept of Ambient Democratic Substitution (ADS): the replacement of democracy’s substantive content with managed preference outputs while its formal mechanisms are preserved.
Individuated operations perform a complementary function that might be termed Targeted Democratic Subtraction (TDS): the removal of specific individuals from the political field through the systematic degradation of their capacity to participate effectively.
Where ADS operates by managing what populations believe, TDS operates by ensuring that the specific people most capable of disrupting managed belief systems are not in a position to do so. The two operations together constitute what this Department’s working notes have described as a complete political management system: mass manipulation establishes the preferred epistemic environment; targeted neutralization removes the individuals most capable of challenging it.
The democratic system continues to perform its formal functions. People vote. Representatives deliberate. Courts adjudicate. The population whose beliefs have been managed elects representatives whose priorities reflect the managed beliefs. The individuals who would have been most capable of disrupting this system are occupied — by legal complications, professional crises, social instability, and the sustained challenge of navigating an ambient environment that has been curated against them. Their capacity to participate in the formal political system remains, in a technical sense, intact. Their effective capacity to participate is negligible.
This is not a dystopian projection. It is a description of documented historical programs. The novelty of the current situation is not the existence of these programs but their open-market availability: the combination of commercial data infrastructure, proxy network architecture, and regulatory non-enforcement that makes individuated targeting operations available to actors outside formal state intelligence structures.
Zersetzung required a state. COINTELPRO required the FBI. The individuated manipulation operations available in the contemporary environment require neither.
They require:
The barrier to entry for individuated operations has fallen as the commercial infrastructure has developed. What required a state apparatus in 1975 requires, in 2026, a sufficient budget, a sufficient network, and a sufficient motivation. All three are available to actors well below the threshold of formal state intelligence services.
The market structure for individuated operations is therefore not the bilateral market of state versus dissident. It is a market with multiple demand-side actors — political, commercial, ideological, personal — and a supply side consisting of commercial data infrastructure, private investigation and opposition research firms, political operative networks, and the informal economy of individuals willing to perform specific tasks for compensation that is difficult to trace.
We note that this market is not regulated, not surveilled in any systematic way, and not the subject of any active legislative effort this Department is aware of. We note this not to encourage demand-side participation but to document, with the specificity the subject deserves, the actual structure of the environment in which politically targeted individuals currently exist.
The parent report documented the accessibility of mass digital manipulation to foreign actors. This addendum documents the accessibility of individuated operations to the same actors.
Foreign state intelligence services operating in the United States have documented histories of conducting in-person harassment, social network disruption, and ambient environment manipulation against specific targets — primarily diaspora community members, activists, journalists, and researchers whose activities threaten the political interests of the operating state. The operations documented against Chinese, Russian, Iranian, and Saudi targets in recent years demonstrate that foreign actors have developed the domestic proxy networks necessary to conduct individuated physical operations without direct deployment of state personnel.¹
What makes the contemporary environment specifically favorable to foreign actors conducting individuated domestic operations is the same feature that makes it favorable to all actors: the legal architecture provides no adequate remedy. The FBI has authority over foreign intelligence operations; it does not have unlimited capacity to investigate every allegation of ambient harassment that has the characteristics of a deniable foreign intelligence operation but no provable nexus to a foreign state. The target experiences the operation. The operation is not investigated. The operation continues.
This is the Retroactive Leverage Accumulation dynamic identified in the parent report, applied to the individuated tier: the government’s non-intervention in documented patterns of foreign-state-adjacent harassment of American citizens generates leverage held by the operating state over the non-intervening government. The leverage is the documented record of what occurred and who knew about it. The leverage is held implicitly, without negotiation, and is spent through the ordinary operation of institutional self-interest.
The complete picture, integrating the parent report with this addendum, is as follows:
At the population level, mass digital manipulation infrastructure is used to shape aggregate political preferences, degrade shared epistemic infrastructure, suppress or amplify specific political identities, and normalize narratives that serve the political interests of operating actors. This layer of the operation is visible to every citizen simultaneously and invisible to each of them individually.
At the individual level, targeted citizens who represent specific threats to the political interests of operating actors are subjected to individuated operations: ambient environment curation, professional network interference, social relationship disruption, and the systematic production of biography fracture that degrades their future credibility and institutional access. This layer of the operation is experienced intensely by its specific subjects and is, by design, invisible to everyone else.
The two layers interact and reinforce each other. Mass manipulation shapes the epistemological environment in which individuated subjects attempt to describe what is being done to them: a population whose epistemic infrastructure has been degraded is less capable of evaluating the claims of targeted individuals with credibility. Individuated operations remove from the political field the specific actors most capable of disrupting mass manipulation: the people who can diagnose and communicate what the mass operations are doing are progressively less able to do so.
The result is a system that does not require visible authoritarianism, formal political prisoners, or overt repression. It requires only the sustained, patient operation of a combined architecture in which the managed population cannot easily identify what is managing them and the individuals most capable of the identification are occupied, credibility-degraded, and increasingly unable to be heard.
This is not a description of a future risk. It is a description of the current environment, assembled from its documented components.
The democratic accountability gap produced by the combined model is the following:
Citizens cannot fully consent to a political system whose epistemic conditions have been managed. Individuals cannot fully participate in a political system whose management infrastructure is designed to neutralize their participation. Institutions cannot fully govern operations that are designed to exploit the specific gaps in institutional authority. Oversight cannot fully function over conduct that is designed to be the sum of individually deniable components.
What remains — what functions in the space cleared by these gaps — is the preference of actors with sufficient resources, knowledge, and motivation to conduct operations in an environment designed to accommodate them.
We note, for the record, that “the preference of actors with sufficient resources, knowledge, and motivation” is a description that has a name in political philosophy. The name is not democracy.
¹ The cases of Chinese Ministry of State Security affiliated transnational repression operations documented by Human Rights Watch (2022, 2023), the Iranian government’s documented harassment campaigns against dual nationals and journalists, the Saudi government’s documented operations against dissidents (the Khashoggi operation representing the extreme end of a documented operational spectrum), and the Russian GRU’s documented operations against perceived enemies in multiple Western countries are each drawn from public record. Their citation here is not to create a foreign adversary framing that obscures the domestic dimensions of the analysis. It is to document that the operational architecture is real, is in use, and does not require any particular state’s sponsorship to replicate.
THE UNCERTAINTY DEPARTMENT — Program on Cognitive Infrastructure and Ambient Political Risk
Addendum UD-RR-2026-031-ADD-A. To be read in conjunction with UD-RR-2026-031, “The Open Market for Democratic Subversion.” This Department notes that the combination of the parent report and this addendum constitutes a reasonably complete picture of the political management infrastructure of the contemporary environment, which is perhaps more complete a picture than is comfortable. We have chosen to prioritize accuracy over comfort. This is a persistent institutional decision whose costs we continue to monitor.
UD-RR-2026-031-ADD-A // End of Document