A work of speculative fiction. The Department of Strategic Continuity is not a real government agency, and this is not a government website. About this project
Technical reports, working papers, frameworks and policy documents released by the Department. All documents listed here are approved for public release.
This report is the first edition of the Civilizational Systems Outlook, the Department’s annual model-based assessment of systemic stress across critical global systems over a ten-year horizon. It was prepared by the Office of Strategic Risk Assessment with contributions from the Office of Advanced Artificial Intelligence, the Office of Long-Term Systems Research and the Continuity Infrastructure Administration. The assessment uses version 4.2 of the Integrated Civilizational Systems Model (ICSM), an eleven-module coupled simulation, run as a 40,000-member ensemble across 1,280 scenario configurations on the ALPHA-1 system at Continuity Research Facility Alpha. The Composite Systemic Stress Index (CSSI) is estimated at 54 for 2026, against approximately 31 in 2000. In 71 percent of weighted ensemble runs (90 percent interval 58–82 percent), the CSSI enters the systemic breakdown regime, defined as a CSSI of 70 or above sustained for 24 months, before 1 January 2035; the median entry year is 2033. The Office assesses this outcome as likely, with moderate confidence. The principal drivers are tight coupling between automated infrastructure control and energy systems, synchronized food-system shocks, concentrated supply chains, financial contagion and erosion of institutional capacity, with climate acting as a stress multiplier. The report sets out methods, assumptions, validation, the Model Review Panel’s two dissenting opinions and ten recommendations. It is a model-based assessment, not a prediction.
The Department makes commitments whose consequences extend over decades: research programs with phases running into the 2030s, facilities designed for a century of service, and archives intended to remain readable after the institutions that created them have changed beyond recognition. This working paper examines how such commitments can be made responsibly when the relevant futures cannot be assigned reliable probabilities. It distinguishes levels of uncertainty, describes robust decision making and adaptive pathways with worked examples drawn from long-duration archiving and facility power supply, and proposes a framework of signposts and pre-agreed triggers for revising plans as conditions change. It considers how long-horizon forecasts can be verified when their outcomes will not be observed for many years, recommending a registry of time-stamped projections scored on a fixed schedule. It sets out a graded posture for acting on projections of high consequence, distinguishing low-probability from high-probability cases and model-conditional from unconditional probability. Finally, it proposes nine design principles for institutions intended to persist across decades. The paper is offered for discussion and does not represent official Departmental policy.
The Neural Data Preservation Framework (NDPF) establishes requirements for the long-term preservation of data produced by the Neural Continuity Research Program, from raw imaging of nervous tissue to the parameters of computational models derived from it. Version 0.9 is a preliminary draft released for public comment. It defines four data classes — raw acquisition, reconstruction, model parameters, and provenance and custody records — and sets out 42 numbered requirements covering metadata and provenance, integrity and fixity checking, preservation formats, geographic and media replication, consent and custody, access control and review. The Framework requires at least three preservation copies on at least two media types, held at two or more facilities, with one copy held offline at Continuity Research Facility Charlie; working copies at Facility Alpha do not count toward this total. Human-derived data are subject to additional consent, custody and encryption requirements, and the Framework prohibits any representation that data preservation constitutes or enables the survival of a donor. The Framework does not authorize any acquisition or experiment. The public comment period closes on 15 December 2026, and Version 1.0 is planned for issue during fiscal year 2027 following adjudication of comments.
This assessment maps the dependencies among six critical global systems — energy, water, food, information, finance and computation — and examines how those dependencies shape the propagation of failure. It represents the six systems as a multilayer network of 1,326 function nodes across 24 world regions, with dependency strengths estimated from perturbation experiments in the Integrated Civilizational Systems Model, engineering data and structured elicitation. The assessment defines five dependency classes, reports the twelve strongest directed couplings, traces four cascade pathways that recur across perturbation experiments, and identifies nine concentration chokepoints whose loss would affect several systems at once. Network analysis indicates that the coupled system is markedly less robust than its layers considered separately, that this gap has widened since 2016, and that the computation layer has become the most central intermediary between the other five. The number of single nodes whose loss would remove more than 10 percent of another system’s function has risen from 11 to 19 over the decade. The assessment describes structure; it does not project future conditions. Its findings inform the probabilistic Civilizational Systems Outlook scheduled for October 2026.
This report sets out the technical requirements that a synaptic-resolution brain-mapping capability would have to meet to support the staged objectives of the Neural Continuity Research Program (NCRP). It reviews the principal imaging modalities — serial-section and block-face electron microscopy, expansion microscopy and X-ray methods — against the structural dimensions that must be resolved, and derives order-of-magnitude data volumes for reference specimens. At a reference sampling of about 500 cubic nanometers per voxel, one cubic millimeter of cortex yields approximately 2 petabytes of raw image data; a mouse brain approaches 1 exabyte; and a human brain of about 1.2 million cubic millimeters would yield on the order of 2 zettabytes. The report specifies sustained acquisition rates for each Program Phase, identifies automated segmentation accuracy and proofreading effort as the binding constraints on scale, and describes the molecular annotation needed to move beyond wiring diagrams. It stresses that a structural connectome is not a functional model and that no acquisition method can by itself establish emulation fidelity. Finally, it states twelve requirements for the acquisition-and-processing pipeline at Continuity Research Facility Alpha. No human whole-brain acquisition is proposed in any current Program Phase.
This report estimates the compute, memory, storage and electrical power that would be required to run computational models of a human-scale brain at each Emulation Fidelity Level (EFL) used by the Neural Continuity Research Program. Estimates are built from first principles — state variables, update rates and operations per update — and are presented as order-of-magnitude ranges. The ranges widen sharply with fidelity: a neuron-level connectome model (EFL-3) running in real time is estimated at 10¹⁶ to 10¹⁸ floating-point operations per second and 0.4 to 20 petabytes of memory, while synapse-level models with plasticity (EFL-4) span four orders of magnitude and molecular-state models (EFL-5) admit no meaningful upper bound. No estimate is possible for EFL-6. The estimates are compared with ALPHA-1, the primary system at Continuity Research Facility Alpha, and scaled to hypothetical facilities hosting many models. The report finds that memory capacity, rather than arithmetic throughput, is likely to be the binding constraint at EFL-3, and that energy requirements above EFL-3 are, at the upper end of present uncertainty, physically prohibitive. It examines power supply, cooling and siting implications for Facilities Alpha and Bravo and the sensitivity of all results to future gains in computing efficiency.
This report sets out the framework the Office of Advanced Artificial Intelligence uses to evaluate advanced computational systems, including large learned models and the automated decision systems built around them, that are deployed in, or proposed for, control and decision roles in critical infrastructure. Four sectors are in initial scope: electric grid operations, freight and logistics coordination, financial clearing and settlement, and communications network management. The framework assesses six dimensions: reliability under distribution shift, correlated failure across deployments, adversarial robustness, controllability and override, dependency concentration, and monitoring and observability. Evaluation depth is set by a system’s deployment role and the consequence class of the infrastructure it touches, producing four evaluation tiers that range from documentary screening to extended trials in sandboxed digital twins of infrastructure operations hosted at Continuity Research Facility Alpha. The report defines the metrics recorded for each dimension, the indicative acceptance criteria applied at each tier, and the reporting arrangements for system developers, infrastructure operators and the Department. It also describes how evaluation results are converted into parameters for the Computational Systems and AI module of the Integrated Civilizational Systems Model, so that system-level evidence informs the Department’s assessments of systemic risk. The framework is an evaluation method; it does not certify systems for deployment.
The Neural Continuity Research Program asks whether human cognition could eventually be represented and sustained on computational systems. Any such program must say what would count as success, and that requires an account of when, if ever, a computational model derived from a brain would stand in a relation of identity or continuity to the person whose brain it was. This working paper reviews the principal philosophical positions — psychological continuity, physical continuity and pattern views — and examines the duplication problem and the contrast between gradual and discontinuous transfer as thought experiments. It distinguishes measurable proxies, such as structural, dynamical and behavioral correspondence, from metaphysical questions that no present measurement can settle. The paper takes no position on whether an emulation would be conscious, and explains why that uncertainty must itself shape research practice. It proposes nine working criteria for research purposes only, together with moral-status precautions and principles of consent. The central recommendation is that Departmental documents make claims of correspondence, which can be tested, and refrain from claims of identity or continuity, which currently cannot. The paper is a contribution to discussion and does not represent Departmental policy.
This report assesses six classes of storage media against the Department’s requirement to retain selected digital and documentary holdings for at least one hundred years: linear magnetic tape (LTO), hard disk drives, flash storage, inorganic recordable optical discs of the M-DISC class, archival silver-gelatin microfilm, and laser-written silica-glass optical storage. For each medium it considers expected media life under controlled storage, environmental sensitivity, dependence on specialized reading equipment, storage density, qualitative cost categories, and survivability without power. Lifetime figures are presented as ranges and distinguish manufacturer ratings, accelerated-aging extrapolations and field experience, none of which can be verified directly over a century. The report concludes that no single medium meets the requirement alone. The limiting factor for most digital media is not the medium but the continued availability of devices able to read it. The Administration therefore adopts a layered approach: actively managed tape copies with scheduled fixity checking and migration; long-lived, low-maintenance copies on optical, microfilm and, on a pilot basis, glass media; a maintained reserve of reading equipment; and human-readable bootstrap documentation that allows encodings to be reconstructed without surviving software. The report specifies the environmental basis of Archive Level 1 at Continuity Research Facility Bravo and the unpowered survivability design of Continuity Research Facility Charlie.
This report documents version 4.0 of the Integrated Civilizational Systems Model (ICSM), the coupled simulation model maintained by the Office of Strategic Risk Assessment under the Integrated Systems Modeling Program, as baselined in April 2026. ICSM v4 represents eleven interacting modules, from energy and food systems to governance and demography, across 24 world regions at a monthly coupling interval. The report describes the module architecture, the coupler and the principal feedback loops; defines the Composite Systemic Stress Index (CSSI), the 0–100 headline metric used throughout the Department’s systemic risk work, together with its weights and stress bands; and summarizes the data classes, two-stage calibration and Bayesian history matching used to constrain 214 coupled parameters. Hindcast experiments covering the 2008 financial crisis, the 2010–2011 food price episode, the 2020–2022 pandemic and supply disruptions and a ten-year free run show positive probabilistic skill against a persistence reference, a systematic under-estimation of peak stress, and ensemble spread narrower than nominal. The report sets out the computational implementation, known limitations, configuration management arrangements and review by the Model Review Panel, and lists the v4.x revisions planned before the model is used for the Department’s first Civilizational Systems Outlook.
This Statement, issued by the Secretary of Strategic Continuity five weeks after the Department’s establishment, sets the Department’s research priorities for calendar year 2026. It builds on the findings of the Interagency Task Force on Civilizational Continuity, whose final report concluded that long-horizon systemic risks were not being assessed in an integrated way by any federal institution and that the means of preserving knowledge across severe disruption had received little sustained attention. The Statement establishes six priority areas: integrated assessment of systemic risk; evaluation of advanced computational systems in critical infrastructure; long-term forecasting and decision methods; continuity research infrastructure; long-duration information preservation; and neural continuity research, to be chartered as a dedicated program subject to independent ethics review. For each area it assigns a lead office and partner offices and sets objectives for 2026. It establishes four cross-cutting principles that apply to all Departmental research: research integrity, transparency, ethics review and public accountability. It directs lead offices to submit implementation plans within 60 days, sets quarterly and annual reporting requirements, and provides for review of the Statement in February 2027.
Identifiers are assigned sequentially across all document series within a calendar year. Numbers not listed are assigned to internal documents or documents pending release.
