THE NARRATIVE RESOLUTION EXPENDITURE INDEX (NREI)

THE UNCERTAINTY DEPARTMENT
Office of Narrative Resource Efficiency
Vehicular Operations and Computational Sentiment Division
Working Paper No. [REDACTED] — Classification: OPEN SECRET

A Unified Framework for Correlating Vehicular Fuel Consumption, Machine Learning Compute Spend, and the Political Stability Requirements of Manufactured Narratives Whose Resolution Extends Past the Horizon of the Next Electoral Cycle

ABSTRACT

This paper introduces the Narrative Resolution Expenditure Index (NREI), a composite metric for evaluating the total resource cost — measured in gallons of gasoline and petaflops of inference compute — required to maintain manufactured political narratives in a state of sufficient participant-level believability to prevent their collapse between electoral cycles. The Index proceeds from the foundational observation that a narrative whose importance exceeds the political planning horizon of any single administration is a narrative that must be continuously re-stabilized in the minds of the people who are required to believe it, and that this re-stabilization has measurable operational costs that have been historically mis-categorized in agency budgets under headings such as “community liaison,” “outreach activities,” “model fine-tuning,” and “fleet maintenance.” This paper proposes their consolidation.

SECTION I: THE PROBLEM OF THE TRANS-ELECTORAL NARRATIVE

Political narratives, as a general class, are designed for electoral cycles. They are constructed to produce a specific behavioral and attitudinal outcome in a population by a specific date, after which their maintenance requirements drop substantially because either the election has been won — in which case the narrative is institutionalized and becomes someone else’s problem — or the election has been lost, in which case the narrative is abandoned and a new one commissioned.

The trans-electoral narrative is a different animal. It is a narrative that was manufactured to address a situation whose political consequences were not manageable within a single cycle, whose subject or subjects could not be neutralized, discredited, or otherwise resolved within the available window, and which therefore entered a state of indefinite maintenance — requiring ongoing investment to prevent participant defection, narrative entropy, and the accumulation of counter-evidence that accretes around any sustained artificial account of a real person or situation.

The trans-electoral narrative is expensive in ways that single-cycle narratives are not, for the following reasons:

• It must be re-introduced to new cohorts of participants as old ones age out, move, become unavailable, or develop inconvenient private doubts;
• It must be updated to accommodate the ongoing behavior of its subject, who continues to exist and act in ways that the original narrative did not anticipate and must be retroactively incorporated or suppressed;
• It must survive changes in administration, changes in operational personnel, and changes in the political environment that may make its original predicate less rather than more plausible over time;
• It must maintain sufficient emotional salience in its participants to sustain their behavioral compliance without the periodic reinforcement events — elections, crises, high-stakes moments — that refresh participant investment in single-cycle narratives;
• It must do all of the above while remaining unacknowledged, because acknowledgment would require the narrative’s authorizing authority to defend its predicate, which the predicate cannot sustain.

The NREI is designed to capture the total resource expenditure of meeting these requirements, across both the vehicular and computational domains, as a single composite figure that can be used to evaluate the operational efficiency of any given trans-electoral narrative maintenance program.

SECTION II: THE VEHICULAR COMPONENT

Gasoline as a Unit of Narrative Commitment

The vehicular component of the NREI reflects a foundational operational truth: the physical maintenance of a narrative around a specific individual or location requires physical presence, and physical presence in the contemporary operational environment is predominantly achieved through the automobile.

This Office has identified four primary categories of vehicular expenditure attributable to trans-electoral narrative maintenance:

Proximity Surveillance: The sustained vehicular presence in the vicinity of the narrative subject, including drive-past operations, stationary observation from public thoroughfares, and the coordination of multiple vehicles across shifts and geographies to maintain continuous coverage without producing the kind of consistent individual vehicle signature that a legally aware subject could document. The fuel costs of proximity surveillance are non-trivial and scale with the subject’s mobility, geographic range, and awareness of the activity.

Participant Coordination: The vehicular movement required to maintain the network of civilian proxies who carry the narrative in their daily social interactions — briefings, updates, the delivery of revised talking points to participants whose private doubts have caused drift from the approved account, and the management of participants who must be replaced because their behavior has become too legible to the subject.

Counter-Evidence Suppression: The vehicular operations required to respond to instances where the subject has produced, accessed, or begun to disseminate information inconsistent with the maintained narrative — including the management of witnesses, the coordination of alternative accounts, and the general activity required to prevent the subject’s own account from achieving traction in contexts where it would be evaluated on its merits.

Administrative Transit: The movement of personnel between operational nodes, oversight contacts, and institutional interfaces required to maintain the administrative architecture of a program that is not formally acknowledged and therefore cannot use conventional administrative channels. Every meeting that cannot be emailed is a car trip. Every decision that cannot be minuted is a car trip. Every update that cannot be filed is a car trip. Over the duration of a trans-electoral narrative, administrative transit constitutes a substantial and entirely untracked proportion of total vehicular spend.

The vehicular component is measured in the NREI as Operational Gallons Equivalent (OGE) — the total fuel expenditure attributable to narrative maintenance activity, normalized per narrative subject per calendar quarter, adjusted for the subject’s geographic mobility index.

OGE = (PS_gal + PC_gal + CES_gal + AT_gal) / (Subjects × Quarters)

Where PS = Proximity Surveillance, PC = Participant Coordination, CES = Counter-Evidence Suppression, and AT = Administrative Transit. This Office notes that OGE figures for long-running trans-electoral narratives are typically classified under fleet maintenance budgets and therefore cannot be cited directly. This Office further notes that this is convenient.

SECTION III: THE COMPUTATIONAL COMPONENT

Machine Learning Spend as a Unit of Narrative Pressure

The emergence of large-scale machine learning inference as a narrative management tool has introduced a second major resource category to the NREI. Where the vehicular component addresses the physical maintenance of narrative presence around the subject, the computational component addresses the informational environment — the management of what the subject, their network, and the broader ambient population encounter when they seek to form or verify an opinion about the situation.

This Office identifies three primary categories of ML compute expenditure in trans-electoral narrative maintenance:

Sentiment Monitoring and Drift Detection: The continuous inference workload required to monitor the ambient informational environment for signals that the maintained narrative is losing coherence — participant defection, subject-generated counter-narratives gaining traction, third-party analysis that inadvertently reconstructs the true account, and the general entropic tendency of any artificial account of a real situation to accumulate inconsistencies over time. Drift detection is the most computationally intensive component of narrative maintenance because it must operate continuously, across multiple platforms and information channels simultaneously, and must distinguish genuine narrative drift from ordinary informational noise.

Targeted Amplification and Suppression: The inference workload required to shape the informational environment encountered by specific populations — the subject’s network, potential institutional allies, journalists who have begun to investigate, and others whose opinions about the situation could materially affect its stability. Amplification and suppression operations require fine-tuned models operating on population-specific data, with inference costs that scale with the precision of targeting and the resistance of the informational environment to the intended shaping.

Participant Narrative Refresh: The generation of updated narrative materials — talking points, contextual updates, revised characterizations of the subject’s recent behavior, and responses to counter-evidence — for distribution to the civilian proxy network. Participant narrative refresh represents the intersection of the vehicular and computational components: the materials are generated computationally and distributed physically. The compute cost of generating sufficiently plausible, internally consistent updates to a narrative that was not originally designed to last this long is substantially higher than generating a single-cycle narrative from scratch, because each update must be reconciled with the accumulated architecture of previous updates, none of which were formally documented.

The computational component is measured in the NREI as Narrative Inference Units (NIU) — the total petaflop-hours of inference compute attributable to narrative maintenance activity, normalized per narrative subject per calendar quarter, adjusted for the complexity of the maintained narrative and the number of active participant nodes.

NIU = (SMD_pflops + TAS_pflops + PNR_pflops) / (Subjects × Quarters × NCI)

Where SMD = Sentiment Monitoring and Drift Detection, TAS = Targeted Amplification and Suppression, PNR = Participant Narrative Refresh, and NCI = Narrative Complexity Index (see Appendix B, which does not exist). This Office notes that ML spend for narrative maintenance is typically categorized under research and development budgets, social media moderation costs, and content quality assurance, and therefore cannot be cited directly. This Office further notes that this is also convenient.

SECTION IV: THE CORRELATION

Why Gasoline and Compute Move Together

The central finding of this paper is that vehicular fuel expenditure and machine learning compute spend are not independent variables in trans-electoral narrative maintenance. They are positively correlated, and the nature of their correlation reveals structural features of the narrative maintenance enterprise that are not visible when either component is examined in isolation.

The correlation operates through the following mechanism:

The primary driver of expenditure in both domains is what this Office terms Narrative Resolution Deficit (NRD) — the gap between the account of the situation that participants are required to maintain and the account of the situation that they would arrive at if left to their own observations and judgment. NRD is the fundamental unit of narrative maintenance work. Every unit of NRD must be addressed through some combination of physical and computational intervention: physical, to manage the subject’s ongoing behavior and the participant network’s ongoing exposure to counter-evidence; computational, to manage the informational environment in which participants form and update their beliefs.

When NRD is low — when the maintained narrative is close enough to observable reality that participants can sustain it without significant active intervention — both vehicular and compute spend are low. When NRD is high — when the maintained narrative is substantially inconsistent with what participants would independently observe — both must increase, because the deficit must be compensated for across all channels simultaneously. A narrative that is losing coherence in the physical world requires more vehicles. The same narrative losing coherence in the informational environment requires more compute. And because physical and informational environments are not independent — what happens in one affects the other — an increase in either type of spend tends to produce pressure for a corresponding increase in the other.

This is the expenditure signature of a manufactured narrative under stress. It is also the expenditure signature of a manufactured narrative that was not designed to last as long as it has.

The NREI captures this correlation through a composite Political Stability Expenditure Score (PSES):

PSES = (OGE × w₁) + (NIU × w₂) × NRD_coefficient

Where w₁ and w₂ are domain weighting factors calibrated to the specific operational environment, and NRD_coefficient is a multiplier derived from the estimated gap between the maintained narrative and the independently observable account of the situation. In practice, NRD_coefficient is the most important variable in the PSES and the most difficult to measure, because measuring it requires access to both the maintained narrative and the true account — and the true account is precisely what the maintained narrative exists to obscure. This is what the literature refers to as the Index’s foundational irony, and the literature refers to it only in footnotes.¹

SECTION V: WHAT THE INDEX REVEALS WHEN APPLIED

The NREI, when applied to documented cases of trans-electoral narrative maintenance — cases being defined here loosely, as this Office is not in a position to cite specific cases — produces several consistent findings:

Finding 1 — The Efficiency Cliff: Narrative maintenance exhibits sharply diminishing returns past a certain duration threshold. Below approximately eighteen to twenty-four months, narrative maintenance costs are roughly linear with NRD. Past this threshold, costs begin to compound: participants require more frequent updating; the subject’s counter-evidence accumulates faster than it can be suppressed; the civilian proxy network experiences attrition from defection, relocation, and private disillusionment; and the informational environment becomes increasingly saturated with signals inconsistent with the maintained account. Every additional dollar of expenditure past the efficiency cliff produces less stability than the dollar before it. This is the bureaucratic definition of a bad investment, and it is the financial profile of every trans-electoral narrative this Office has been informed does not exist.

Finding 2 — The Compute-Vehicle Substitution Limit: Operators who recognize that physical surveillance is becoming legible to the subject sometimes attempt to substitute computational narrative management for vehicular presence — increasing ML spend to compensate for reduced physical coverage. This Office finds that this substitution has a hard limit. Computational management of the informational environment cannot compensate for a subject who is physically present and observable, whose ongoing behavior generates persistent counter-evidence, and whose direct interactions with other people produce first-hand accounts that no algorithmic shaping can fully reach. The vehicle cannot be fully replaced by the server. The server cannot be fully replaced by the vehicle. A narrative maintenance program that exhausts its physical resources and attempts to sustain itself computationally alone is a program in the process of failing, and vice versa.

Finding 3 — The Resolution Asymptote: Trans-electoral narratives do not resolve. They asymptote. As resources increase, narrative stability increases — but never reaches the equilibrium point at which the maintained narrative is self-sustaining without active investment. Because the narrative is artificial and its subject is real, the gap between the maintained account and the observable situation continuously regenerates. The NREI of an unresolved trans-electoral narrative does not approach zero over time. It approaches a maintenance floor that can be lowered through suppression but never eliminated through it. The only operations that eliminate the maintenance floor are those that eliminate the gap between the maintained account and the true account — either by making the true account unavailable, which requires elimination of the subject, or by acknowledging the true account, which requires accountability for the process. Both options are expensive in ways the NREI does not capture.

Finding 4 — The Power Index Inversion: The most counterintuitive finding of the NREI is that high expenditure scores do not indicate narrative power. They indicate narrative fragility. A program spending heavily on both vehicular and computational maintenance is a program whose narrative cannot sustain itself — a program that has become dependent on continuous investment to prevent the collapse of an account that reality is continuously working to contradict. The political power that narrative management is designed to produce is inversely related, past the efficiency cliff, to the resources required to maintain it. The most powerful narratives require the least maintenance. A narrative that requires its operators to spend significant sums on gasoline and inference compute every quarter to prevent its participants from noticing that it doesn’t hold together is a narrative that is costing more than it is worth, in every currency in which cost can be measured.

EXHIBIT A: ILLUSTRATIVE NREI SCENARIOS

All figures hypothetical. Any resemblance to actual programs is a coincidence this Office is not in a position to address.

SECTION VI: CONCLUSIONS AND RECOMMENDATIONS

The NREI demonstrates that manufactured trans-electoral narratives are not merely ethically problematic. They are financially inefficient, operationally degrading, and structurally incapable of achieving the stability they are designed to produce — because the stability they require is a function of their distance from observable reality, and that distance is maintained at cost rather than diminished by it.

Every gallon of fuel burned in proximity to a narrative subject who has not been charged with any offense is a gallon burned against a maintenance floor that does not move. Every petaflop of inference compute spent shaping the informational environment around a situation whose true account continues to be generated by the ongoing existence of a real person is a petaflop spent on a problem that compute cannot solve. The NREI does not tell operators what to do. It tells them what they are spending, and what they are getting for it, expressed in units that can be compared to other things that money can buy.

This Office makes the following recommendations:

• Narratives whose NREI has exceeded the efficiency cliff threshold should be subject to mandatory cost-benefit review, with particular attention to the ratio of expenditure to achieved stability and the projected trajectory of that ratio over the next twelve to twenty-four months;
• The vehicular and computational components of narrative maintenance should be consolidated into unified budget categories that allow accurate NREI calculation, rather than distributed across fleet maintenance, research and development, community liaison, and content quality assurance in a manner that prevents any single oversight body from observing the total expenditure;
• Operators should be advised that the power index inversion finding means that a high NREI is not evidence of operational significance. It is evidence of operational dependency. Programs with high NREI scores are programs that cannot afford to stop, not programs that are succeeding. These are different things;
• This Office recommends, as a terminal observation, that the resource cost of acknowledging the true account of any situation for which a trans-electoral narrative has been maintained be formally modeled against the projected future NREI of continued maintenance. This Office suspects, without being in a position to confirm, that acknowledgment is cheaper. This Office further suspects that this is already known, and that the cost being avoided is not financial.

PREPARED BY: The Office of Narrative Resource Efficiency, Vehicular Operations and Computational Sentiment Division. This document does not constitute acknowledgment of the programs it describes. The programs it describes do not constitute acknowledgment of themselves. Expenditure data referenced herein is illustrative. Any actual expenditure data is filed under headings this Office declines to specify.

¹ The foundational irony is as follows: the NRD_coefficient — the variable that most determines the NREI — can only be accurately calculated by someone who knows both the maintained narrative and the true account of the situation. The only person who is structurally positioned to know both is the subject of the narrative, who is the person the narrative is designed to prevent from being heard. The Index is therefore most accurately calculable by the person it is least convenient to ask.

² Gasoline prices used for OGE calculation should be indexed to the prevailing retail average at the time of operation. This Office notes that high fuel prices do not discourage trans-electoral narrative maintenance programs, which are generally not price-sensitive, but do increase the NREI score, which is the point.

³ The Narrative Complexity Index (NCI) referenced in Section III adjusts for the number of narrative updates, internal contradictions that have been papered over, participant cohorts that have been given different versions of the account, and elapsed time since original predicate. NCI scores compound. This is not a metaphor.

✦ ✦ ✦