Nationwide Diffusion of January 6–Type Political Violence: An Evidence-Based Risk Assessment

Full report

On this page

Executive summary

This report models the consequences conditional on an event resembling the January 6, 2021 U.S. Capitol attack spreading to multiple locations across the United States. It does not estimate the probability that such an event will occur, and it does not assume that ordinary political protesters, conservatives, liberals, gun owners, or any other broad constituency would participate in violence. The relevant scenario is a mixed mobilization containing peaceful demonstrators, a smaller violent subset, some organized militants, and opportunistic offenders, with repeated attempts to overwhelm or breach government or other protected sites.

January 6 provides an unusually useful U.S. baseline because its scale, criminal cases, law-enforcement injuries, intelligence failures and institutional response have been extensively documented. DOJ reported that approximately 140 police officers were assaulted, that roughly 99 defendants had been charged by January 2023 with using a dangerous weapon or causing serious bodily injury to an officer, and that by January 2025 more than 1,500 individuals had been charged. DOJ described the resulting investigation as one of the most complex and resource-intensive in its history.

January 6 should not, however, simply be multiplied by the number of states. Riot diffusion is nonlinear: research on the London and French riots finds that participation can propagate geographically through social-contagion mechanisms while policing, travel distance, local conditions and previous incidents affect where subsequent events appear. Online-network research likewise finds that protest participation can spread through network exposure rather than through purely independent individual decisions.

The central finding of this assessment is therefore:

Duration, repeated armed encounters and simultaneous violent hotspots matter more than the raw number of protesters.

A one- to three-day series of copycat breaches could produce thousands of injuries and arrests yet plausibly remain below several dozen fatalities. A multi-week, multi-state confrontation involving recurrent armed violence could plausibly produce tens to several hundred deaths. Sustained national political violence lasting months could move into a hundreds-to-low-thousands fatality range even without becoming a conventional civil war. If it crossed into a true insurgency with persistent territorial control, security-force fragmentation or systematic attacks on critical infrastructure, the consequences could exceed the high scenario substantially; that tail case is deliberately excluded from the principal estimates.

Modeled national consequences

These are scenario envelopes, not forecasts or casualty quotas. The ranges intentionally overlap with uncertainty rather than implying a deterministic relationship between crowd size and deaths. They combine the January 6 experience, urban-riot comparisons, academic diffusion literature, federal court capacity, infrastructure-dependency analysis and historical mass-mobilization cases.

The estimates imply several conclusions with substantially higher confidence than the exact numbers:

First, mass arrest and significant injury would be considerably more certain than mass death in the low and medium scenarios. January 6 produced roughly 140 assaulted police officers and ultimately more than 1,500 federal charges despite being concentrated around one federal complex and one day.

Second, zero fatalities remain possible in the low scenario. There is no historical rule that widespread political disorder necessarily kills people. Historical mass movements range from bloodless democratic transitions, such as Mongolia's 1990 revolution, to mass uprisings in which state repression produced substantial deaths, such as Tunisia, where Human Rights Watch documented 132 protesters killed during the core revolutionary period.

Third, the transition from episodic disorder to recurrent armed confrontation is the decisive inflection point. Once firearms are regularly discharged by participants or security forces at numerous independent incidents, a persistently zero-fatality outcome becomes increasingly implausible. This is an inference from the riot, mass-mobilization and armed-conflict evidence rather than a fixed casualty law.

Fourth, nationwide infrastructure collapse is not the base case, even in the high scenario. More likely are numerous local disruptions whose aggregate economic impact becomes large because electricity, communications, transportation, emergency services and other systems depend on one another. CISA explicitly treats infrastructure dependencies and interdependencies as a central resilience-planning problem.

Finally, legal capacity would become a serious constraint well before physical government capacity collapsed. Federal district courts recorded 79,129 criminal-defendant filings in FY2025. Tens or hundreds of thousands of unrest-related arrests would therefore require substantial reliance on state and local systems, prosecutorial triage, and years of litigation rather than straightforward federal prosecution of every participant.

Evidence base and modeling approach

The model uses January 6 as its primary U.S. benchmark, but deliberately does not treat it as an average riot.

The House Select Committee's final report and accompanying documentary collection provide the principal chronology and institutional account. DOJ court records provide the stronger quantitative basis for arrests, offenses and law-enforcement assaults. GAO provides independent evidence on intelligence collection, preparation and interagency information sharing.

GAO found that all ten federal agencies it examined identified potential threats of violence before January 6 and two—the FBI and Capitol Police—identified credible threats, yet relevant information was not always completely processed or shared. GAO subsequently reported substantial remedial action: by 2025, several recommendations concerning threat processing and sharing had been closed as implemented, including changes to DHS Intelligence and Analysis controls and FBI threat-intake procedures. Consequently, a hypothetical repeat should not simply assume that the exact January 6 intelligence failures would recur unchanged.

The FBI and DHS domestic-terrorism strategic assessments are also relevant because a nationwide scenario could contain both mass-crowd disorder and much smaller organized violent groups. The federal assessment framework explicitly addresses domestic-terrorism activity, data and recommendations; GAO separately identifies FBI and DHS as the principal federal entities responsible for domestic-terrorism investigation and threat-information functions.

Empirical anchors

The following cases are not treated as direct equivalents; they define boundaries on plausible outcomes.

The historical examples are important primarily because they reject a simple population-to-casualty multiplier. The same number of politically mobilized people can yield almost no deaths under disciplined nonviolence and restraint, or substantial fatalities where armed confrontation, repression, communal violence or prolonged disorder develops.

Diffusion model

Academic work on the 2011 London riots modeled riot participation as a spatial contagion process modified by distance and police deterrence. Research on the 2005 French riots similarly found epidemic-like temporal and geographic propagation, while online-network studies of protest recruitment found evidence consistent with complex social contagion through network exposure. These studies do not prove that a U.S. political disturbance would reproduce London, France or Spain, but they justify modeling copycat events and temporal clustering rather than assuming each city's behavior is independent.

The model therefore conceptualizes human harm as:

harm ≈ number of violent hotspot-days × population exposed × violence intensity × weapon lethality × duration/escalation multiplier

and infrastructure/economic harm as:

loss ≈ geographic economic exposure × disruption duration × infrastructure-dependency multiplier + direct security/property/restoration costs

Those expressions describe relationships rather than statistical equations; there is not enough empirical U.S. data to estimate reliable nationwide coefficients.

A hotspot-day means one locality experiencing significant political disorder for one day. Ten cities experiencing one violent day therefore constitute roughly ten hotspot-days, while twenty cities experiencing ten days each constitute roughly 200. This variable is more useful than total protest attendance because a million peaceful participants present far less casualty risk than a much smaller number repeatedly engaging in armed confrontations.

The model counts event-attributable fatalities during the modeled period, including direct violence and acute secondary deaths associated with fires or severe interruption of emergency services. It does not automatically count later deaths for which causal attribution is uncertain. “Injuries” refers primarily to physical injuries or significant medical encounters rather than every minor exposure to irritants or first-aid complaint.

Economic losses are short-run real-resource losses, including property damage, business interruption, public safety deployment, disrupted transportation, infrastructure restoration and lost economic activity. They exclude paper changes in stock-market capitalization and do not attempt to value long-term political uncertainty. BEA remains the authoritative source for current U.S. GDP measurement; its August 2026 estimate put real GDP growth at a 1.5 percent annualized rate in Q2 2026, underscoring that the scenario would strike a very large and diversified economy rather than an economy concentrated around a handful of cities.

Scenario definitions and quantified outcomes

Low scenario: distributed flashpoints

This scenario assumes five to fifteen substantial incidents over approximately one to three days. Total participation across peaceful and disorderly gatherings is approximately 50,000–150,000. Most participants remain unarmed and nonviolent; a small fringe attempts forced entry, assaults officers, commits property crime or carries weapons. Firearms may be present but are rarely discharged.

The geographic pattern resembles repeated January 6–type confrontations rather than an insurgency: state capitols, federal buildings or major civic areas may face demonstrations and attempted breaches, but there is no sustained territorial control, systematic sabotage campaign or organized national combat front.

Local police and state authorities retain command capacity. Some governors activate National Guard personnel under state-controlled authorities for perimeter security, logistics or public-safety support, but ordinary civilian law enforcement remains the dominant instrument.

Estimated consequence: 0–25 fatalities, 500–3,000 injuries, 3,000–15,000 arrests and less than roughly $8 billion in short-run national losses.

The lower end could plausibly include zero deaths, particularly if weapons remain holstered, crowds disperse quickly and counter-demonstrators are kept physically separated. January 6 itself illustrates how a violent political event can produce very large numbers of assaults and subsequent criminal cases without producing battlefield-level fatalities.

The main economic burden would likely come from emergency deployments, closures of government and commercial districts, damaged property, travel disruption, cancelled activity and subsequent security expenditures rather than national infrastructure failure. The 1992 Los Angeles disorder is a useful warning that even a single metropolitan disturbance can cause hundreds of millions of dollars of property and insured loss in historical dollars.

Medium scenario: coordinated multi-state disorder

This scenario assumes episodes in 20–40 states over two to six weeks, with approximately 250,000–1 million cumulative participants across demonstrations, counter-demonstrations and disorder.

A majority of participants still do not commit serious violence, but organized violent groups repeatedly appear within larger gatherings. Firearms are present more often and are occasionally discharged; serious assaults, arson, forced entry, property destruction and opportunistic looting occur at multiple hotspots. These are categories of conduct, not assumptions about specific methods or targets.

Events diffuse through a combination of planned demonstrations, perceived successful breaches, retaliatory mobilization, rumors and online amplification. GAO's January 6 work establishes that potentially violent intentions were visible in open-source and social-media information before the Capitol attack, while academic diffusion research provides a mechanism by which visible incidents can encourage temporally clustered copycats.

Police mutual-aid systems become stressed. National Guard deployments occur in multiple states. Federal law enforcement establishes national investigative coordination around federal offenses and organized violent networks.

Estimated consequence: 50–500 fatalities, 5,000–25,000 injuries, 25,000–100,000 arrests, 2,000–25,000 temporary displacements or evacuations, and $20–150 billion in short-run losses.

The range is deliberately wide. The 1992 Los Angeles disturbance alone produced 53 recorded fatalities, 2,383 injuries and roughly 17,000 arrests, demonstrating that one severely affected metropolitan area can reach the lower end of the medium scenario's national fatality count. Scaling that event mechanically across states would be inappropriate, however, because policing, density, weapons use and duration differ.

High scenario: sustained national political violence

The high scenario assumes two to six months of recurring serious disorder across more than 30 states and approximately 1–5 million cumulative participants. This does not mean millions of militants: the large majority may still participate peacefully or episodically. The distinction is that a persistent organized violent minority generates repeated lethal encounters and is reinforced by copycats, opportunistic offenders and counter-mobilization.

Local government becomes temporarily overwhelmed in some areas. Firearms are recurrently used in violent hotspots. Police and Guard forces experience fatigue and repeated deployments. Public events, government operations and portions of transport networks face recurring interruption.

This scenario remains below full civil war. It does not assume armies occupying large territories, widespread military defections, systematic destruction of the national electric system or continuous combat fronts. If any of those conditions appeared, the model's upper bounds would cease to be appropriate.

Estimated consequence: 500–5,000 fatalities, 30,000–150,000 injuries, 100,000–500,000 arrests, 25,000–250,000 people temporarily displaced and $150–750 billion in short-run losses.

The upper end requires multiple mutually reinforcing conditions: prolonged duration, recurrent firearm use, simultaneous hotspots, counter-mobilization, emergency-service strain and localized infrastructure failures. Without those conditions, outcomes would migrate toward the medium scenario.

A transition to actual insurgency would represent a separate tail case. Historical campaign datasets show that violent and nonviolent mass movements cover an enormous range of scale and duration, making it inappropriate to extrapolate conventional-war death rates into a scenario that has not yet become a war.

Sensitivity of the estimates

The sensitivity logic is supported more strongly than the numerical multipliers themselves: riot-diffusion research demonstrates contagion and policing effects; GAO documents the importance of information sharing; and CISA treats infrastructure interdependence as a mechanism through which localized disruption can produce wider consequences.

xychart-beta
    title "Illustrative midpoint of modeled short-run economic loss"
    x-axis ["Low", "Medium", "High"]
    y-axis "Billions of 2026 dollars" 0 --> 500
    bar [4.5, 85, 450]

The bars above are simply the arithmetic midpoints of the scenario ranges—$1–8B, $20–150B and $150–750B—and are not expected-value forecasts.

Escalation timeline and government response

A nationwide event would probably not begin as a nationwide event. The evidence on riot diffusion suggests that the more realistic pathway is a local shock followed by imitation, counter-mobilization and geographic clustering.

flowchart LR
    A["Hours 0–6<br/>Initial breach or violent confrontation"]
    B["Hours 6–24<br/>Images, rumors and narratives spread nationally"]
    C["Day 1–3<br/>Copycat demonstrations and counter-mobilization"]
    D{"Rapid containment?"}
    E["De-escalation<br/>arrests, evidence collection,<br/>localized recovery"]
    F["Day 3–14<br/>Repeated hotspots,<br/>mutual aid and Guard deployments"]
    G{"Violence declining?"}
    H["Weeks 2–6<br/>Investigations, prosecutions,<br/>gradual demobilization"]
    I["Weeks 2–8+<br/>Persistent armed encounters,<br/>infrastructure and economic strain"]
    J{"Civil law enforcement<br/>still effective?"}
    K["High scenario:<br/>sustained national political violence"]
    L["Tail risk outside model:<br/>insurgency / severe security fragmentation"]

    A --> B --> C --> D
    D -- Yes --> E
    D -- No --> F --> G
    G -- Yes --> H
    G -- No --> I --> J
    J -- Yes, but strained --> K
    J -- No in substantial areas --> L

GAO's January 6 review makes the first hours especially important. Agencies had identified potential threats in advance, but deficiencies in processing and sharing information reduced the usefulness of those warnings for operational preparation. GAO now reports that a number of those deficiencies have been addressed, which should improve—not guarantee—response to a future event.

Government decision path

The institutional response would begin with ordinary law enforcement and state emergency authorities rather than automatic deployment of federal troops.

flowchart TD
    A["Violent political disturbance"]
    B{"Local police can contain it?"}
    C["Local response<br/>arrest, crowd separation,<br/>investigation and medical response"]
    D["Mutual aid / state emergency coordination"]
    E{"State resources sufficient?"}
    F["Governor may activate National Guard<br/>under state-controlled authority"]
    G{"Federal crimes, federal sites,<br/>or multi-state coordination?"}
    H["FBI / DOJ / DHS coordination<br/>and federal protective-security response"]
    I{"Ordinary enforcement of federal law<br/>remains practicable?"}
    J["Continue civilian law-enforcement lead"]
    K["Presidential legal review of<br/>Insurrection Act authorities"]
    L{"Statutory conditions met<br/>and invocation chosen?"}
    M["Narrow federal military mission<br/>under applicable authority"]
    N["Restoration of civilian lead,<br/>demobilization and prosecution"]

    A --> B
    B -- Yes --> C
    B -- No --> D --> E
    E -- No --> F --> G
    E -- Yes --> G
    G -- Yes --> H --> I
    G -- No --> J
    I -- Yes --> J
    I -- No --> K --> L
    L -- No --> J
    L -- Yes --> M --> N
    C --> N
    J --> N

The Posse Comitatus Act, 18 U.S.C. §1385, generally prohibits using federal armed forces to execute domestic law except where the Constitution or an Act of Congress expressly authorizes it. The Insurrection Act, including 10 U.S.C. §252, supplies statutory authority under specified circumstances involving unlawful obstructions or rebellion that make enforcement of federal law through ordinary processes impracticable. Therefore, federal military involvement is legally possible in a severe national crisis but is neither automatic nor synonymous with a governor activating the National Guard under state authority.

This distinction materially affects casualty risk. A Guard presence used primarily to reinforce barriers, logistics and public safety is a different scenario from sustained armed confrontations involving federally controlled military forces. The latter should therefore be treated as an escalation indicator, not an assumption built into the low or medium case.

Cross-domain impact assessment

Human consequences

The greatest uncertainty in the casualty model is not crowd size; it is how often crowds transition into lethal encounters.

January 6 illustrates the asymmetry. DOJ counted about 140 assaulted police officers, yet the event did not become a sustained firefight. That produced a large injury and prosecution burden without battlefield-scale fatalities.

The 1992 Los Angeles unrest illustrates the next step upward: multi-day disorder dispersed across an urban area produced 53 deaths and 2,383 reported injuries.

Historical revolutions illustrate still wider variance. Mongolia's democratic transition is used here as a zero-bloodshed boundary case, whereas Tunisia demonstrates that mass political protest can produce triple-digit protester deaths when security-force violence becomes severe. NAVCO's broader campaign data reinforce that there is no universal “revolution casualty rate.”

For a nationwide U.S. crisis, the risk distribution would differ by participant:

Organized armed militants would have the highest fatality risk per participant if they engage in armed confrontations.

Police and National Guard personnel would experience substantial assault and injury exposure even before firearms became common, as the January 6 officer statistics demonstrate.

Unarmed protesters and bystanders could constitute a large share of overall injuries simply because they would greatly outnumber armed participants and share the same physical space during panic, crowd compression, projectiles, fires and police use of force.

Opportunistic criminals would contribute disproportionately to arrests and property losses if looting, arson or unrelated violence emerged around weakened public-order environments.

The high scenario also produces indirect harm. Repeated road closures, fire-service delays, hospital-access problems and interruption of essential services can cause injuries far from the original political confrontation. That is one reason infrastructure resilience matters to the casualty calculation, not merely the economic one. CISA's dependency framework is specifically intended to help planners anticipate how disruption in one infrastructure system affects others.

Infrastructure consequences

The most probable infrastructure consequence is fragmentation and localized interruption, not a Hollywood-style nationwide blackout.

In the low scenario, the dominant effects would be access restrictions, transit detours, government-building closures, telecommunications congestion and short local outages caused indirectly by disorder.

The medium scenario could produce repeated interruptions to transportation, public safety communications, commercial districts and utility restoration activities. The report models 100,000–2 million cumulative utility customers experiencing some significant unrest-associated service disruption, not two million simultaneously without power.

The high scenario assumes 1–10 million cumulative customers may experience a significant outage or essential-service disruption over months. Even there, the base case remains numerous regional incidents rather than collapse of the national grid.

This distinction follows CISA's resilience concept: infrastructure sectors are interdependent, so disruption can cascade, but dependency does not mean that every localized failure automatically becomes national systemic failure.

The highest-consequence sectors for defensive planning are public-safety communications, electricity, telecommunications, transportation and fuel logistics, emergency medicine, and water/wastewater services. The relevant policy objective is resilience and continuity—not publicizing specific facilities or technical vulnerabilities.

Legal and prosecutorial consequences

The legal burden could become enormous even at casualty levels far below civil war.

By January 2025 DOJ had charged more than 1,500 people in relation to January 6 and described the investigation as exceptionally complex and resource-intensive.

The entire federal district-court system received 79,129 criminal-defendant filings in FY2025.

For perspective, suppose only 10 percent of modeled unrest arrests generated federal prosecutions, with the remainder handled by state/local systems, declined, diverted or otherwise resolved:

The ten-percent assumption is illustrative rather than predictive. But the comparison demonstrates why the high scenario would require triage. Federal prosecutors would likely concentrate on serious violence, organized conspiracies, federal-property offenses and interstate conduct while state and local courts absorbed large volumes of assault, burglary, weapons, arson, trespass and public-order cases.

The burden would extend beyond prosecutors: public defenders, pretrial services, detention facilities, judges, probation offices, digital-forensics teams and appellate courts would all be affected. The federal Judiciary was already processing more than 79,000 criminal-defendant filings annually before adding any such extraordinary caseload.

Economic consequences

Economic costs would grow faster than property destruction alone.

In the low scenario, much of the $1–8 billion estimate consists of public-safety mobilization, overtime, court/investigative costs, closed commercial districts, event cancellations, travel interruptions and localized property damage.

The medium scenario's $20–150 billion range assumes that repeated disruptions begin affecting commuting, freight, tourism, retail activity and regional supply chains across numerous metropolitan areas for weeks.

The high scenario's $150–750 billion range assumes months of repeated disruption, higher insurance and security costs, significant interruptions to work and transportation, localized infrastructure restoration, lost investment and business activity, and sustained government response.

The 1992 Los Angeles event offers scale context: contemporaneous reporting documented $775 million in insurance losses, damage to roughly 1,700 businesses and more than $55 million in emergency-response costs from a single metropolitan crisis in 1992 dollars.

The ranges are intentionally broad because a national U.S. economy can reroute activity around individual cities. Geographic spread therefore matters less than whether unrest interferes repeatedly with major economic regions and interdependent transportation or communications services. BEA's national accounts underscore the scale and diversity of U.S. output, making a temporary shutdown of a few downtown districts categorically different from a months-long disruption across numerous states.

Communications and political consequences

January 6 also demonstrates why an information environment cannot be treated as separate from physical security. GAO found that agencies were already receiving open-source and social-media information indicating potential violence before January 6; the problem included incomplete processing and sharing of that information.

Academic protest and riot research suggests a plausible feedback cycle:

visible confrontation → online exposure → increased perceived participation/success → copycat mobilization → new confrontation → further online amplification.

Misinformation can intensify this process by falsely reporting deaths, attacks, government actions or victories. The model does not assume that any particular contemporary social platform dominates; platform ecosystems change too quickly for that assumption to be stable. Instead, communication speed and network connectivity are treated as general multipliers.

Politically, the low scenario would probably produce aggressive investigations, hardened security around government institutions, litigation over arrests and police tactics, and intense partisan disagreement over responsibility.

A medium scenario would create greater federal-state friction, competing narratives about legitimate protest versus political violence, pressure for broader emergency authorities and a heightened risk that otherwise lawful political organizations become stigmatized by association with violent factions.

A high scenario would pose a qualitatively more serious threat: persistent political violence could weaken confidence in elections, courts and law enforcement; normalize retaliatory mobilization; create major civil-liberties disputes; and force repeated decisions over the division of power among governors, federal law enforcement and the presidency. Those are analytical consequences rather than predictions about which party or ideology would benefit.

The most damaging political dynamic would be reciprocal escalation: one side interpreting every act of enforcement as repression and government officials interpreting increasingly broad populations as security threats. Historical campaign data show why distinguishing organized violence from peaceful mass participation matters: violent and nonviolent resistance have substantially different dynamics and should not be treated as one homogeneous phenomenon.

Mitigation priorities

The highest-value policies are those that reduce the number of violent hotspot-days while preserving lawful protest and civilian control.

GAO's most directly relevant lesson is that collecting warnings is insufficient if they are not processed and delivered to personnel who need them. Its January 6 review found precisely those failures; by 2025 it had closed several recommendations after FBI and DHS changes to threat intake, sharing and oversight. Maintaining those reforms should be considered a core preventive measure.

A second priority is avoiding indiscriminate escalation. The January 6 prosecutions themselves demonstrate that a large crowd can be legally differentiated: DOJ brought different charges depending on entry into restricted areas, assaults, dangerous weapons, property destruction and conspiracy rather than treating every person as committing the same offense.

A third priority is infrastructure continuity. CISA's infrastructure-resilience framework emphasizes identifying dependencies in advance. From a casualty-reduction standpoint, backup communications, restoration staffing and emergency-service continuity are more important than attempting to make every physical facility invulnerable.

A fourth is prosecutorial proportionality. A medium or high event would create far more potential cases than January 6. Prioritizing serious violence, organizers of criminal conduct and high-consequence offenses while handling lesser cases proportionately would preserve investigative resources and reduce court congestion. This recommendation is an analytical inference from the enormous January 6 investigative burden and the existing scale of annual federal criminal filings.

Finally, any transition to federal military involvement should be treated as an exceptional legal and political threshold. The Posse Comitatus Act's default limitation and Congress's statutory exceptions create a framework in which military support may become lawful under exceptional circumstances without making it the default public-order instrument.

Confidence, limitations, and prioritized sources

Confidence by impact category

The single biggest limitation is the absence of a modern U.S. example combining January 6–style political objectives, nationwide geographic spread, weeks or months of duration, substantial firearm prevalence and modern social-media coordination. That means the medium and especially high scenarios cannot honestly be produced by fitting a clean historical regression.

The second limitation is event definition. “Revolution,” “riot,” “insurrection,” “civil disorder,” “terrorism” and “protest” describe overlapping but legally and analytically different phenomena. The report therefore models observable behavior—violent hotspots, duration, weapons, geographic diffusion and government capacity—rather than assuming that a label determines casualty rates.

The third is that historical revolutionary data are most useful for bounding outcomes, not predicting U.S. casualties. Mongolia demonstrates that large political transformations can occur without bloodshed; Tunisia demonstrates that mass mobilization can incur significant fatalities under repression; NAVCO shows the diversity of violent and nonviolent campaigns. Estonia belongs in the low-casualty comparative family but is not assigned a numeric coefficient here because casualty definitions and event boundaries are insufficiently consistent for the precision required by this model.

The fourth is that January 6 response conditions have changed. GAO reports that FBI and DHS implemented several reforms to threat processing and information sharing after the attack. A model that simply reproduces January 6's pre-event intelligence failures in 2026 would therefore overstate one important vulnerability.

Prioritized evidence base

Bottom-line assessment: A nationwide repetition of January 6–type incidents would not automatically become a civil war and would not automatically produce thousands of deaths. Under a brief, decentralized copycat scenario, mass arrests, police/protester injuries, localized closures and multi-billion-dollar costs are more likely than mass fatalities. Once the crisis persists for weeks and recurrent armed confrontations occur across multiple states, the expected harm changes qualitatively: dozens to hundreds of deaths, thousands to tens of thousands of injuries, severe court overload and substantial regional economic disruption become credible planning assumptions. A months-long environment of recurring armed political violence could plausibly reach hundreds to several thousand deaths and hundreds of billions of dollars in losses, even while remaining below the threshold of conventional civil war.

The most consequential decision points are therefore not the total number of people who demonstrate. They are whether violent incidents reproduce faster than authorities can contain them, whether firearms are routinely used, whether opposing political groups begin retaliatory mobilization, whether public-safety institutions remain cohesive, and whether essential infrastructure and emergency services continue functioning. The historical record and the January 6 evidence both suggest that keeping those escalation mechanisms from reinforcing one another is the most effective way to keep a political crisis from becoming a national human-security disaster.