Systemic innovation for countering violent radicalization: Systems engineering in a policy context

This paper brings a systems engineering approach to policymaking in the context of violent radicalization. We test strategies to combat terrorism under the premise that violent radicalization is a complex system of social contagion resulting in terrorism. We built a simulation using DIME‐PMESII military standards to replicate a terror contagion occurring over 10 years in both physical and online environments under optimal, realistic, and worst‐case scenarios. We then tested antiterrorism, counterterrorism, and counter radicalization strategies as policy experiments in this simulation. These experiments identified four key dynamics relevant for developing policies to reduce terrorism. First, most well‐known policies are ineffective in containing terrorism driven by social contagion. Second, strategies generating backlash can become worse than doing nothing at all. Third, perceived grievance determines the carrying capacity of terrorism in a system, allowing disrupted networks to regenerate. Fourth, variable public support may result in sharp secondary waves of violence under certain contingencies. Experimenting with our model, we explore effective ways to address the violent radicalization problem.

a method similar to the Columbine school mass shooting of 1999.
Although it is unclear whether the perpetrator, who died in the attack, was replicating Columbine, there have been at least 30 perpetrators, some born after Columbine, specifically replicating that attack resulting in over 100 killed since 1999 2, pp. 6-7 including Sandy Hook in 2011 and Marjory Stoneman Douglas in 2018.
These incidents represent two different manifestations of the same phenomenon: terror contagion.A terror contagion is a social contagion of violent radicalization leading to mass violence terrorism.Perpetrators adopt a violent ideology and a specific means of conducting attacks by replicating previous incidents.Terror contagions spread through cultural scripts broadcast by the media after a high-fatality incident.
Cultural scripts act like programming on a sociotechnical system and in the case of a terror contagion they convey a radicalizing ideology and modus operandi of conducting terrorism.The great replacement terror contagion spreads an ideology of white nationalism and xenophobia against communities of color, immigrants, Jews, and Muslims.Policymakers and officers trying to reduce the violence of terrorism design policy systems to guide a sociotechnical systems selecting options among three broad policy areas: (1) antiterrorism, In the past, these activities were conducted separately from one another.More recent innovations have focused on "fusion" and crosscoordination.However, because these activities arise from fundamentally different domains-systems approaches often remain stove-piped in training, funding, and technology support development.In this paper, we are bridging this type of policymaking with the systems context.Systems engineering has made inroads both with policymakers through academic and government collaboration 3,4 and strategically as a community through the International Council on Systems Engineering's (INCOSE) Vision 2035. 5Through this vision and INCOSE's Future of Systems Engineering (FuSE) initiative 6 to implement this vision, there are specific goals to bring systems engineering into sociotechnical applications, including policy systems.
Policy life cycles and system approaches to policy are well-defined outside of systems engineering literature. 7,8Bringing these perspectives in would strengthen current systems engineering methodology.
Using systems engineering for policy has two important dimensions: (1)   evaluating what the policy life cycle is in the domain application and (2)   how is the policy represented as system architecture.In this study, we did this through system dynamics modeling as a tool for system analysis of the policy life cycles.We conducted simulation experiments for the three policies in isolation and then in combination.The life cycle of policy analysis is represented in these models.First, under hypothetical best-case scenarios, then adding realistic constraints.The realistic constraints included varying public support based on the perceived risk of terrorism, policy backlash, and a rising tide of potential perpetrators.If policies performed well under realistic constraints, we subjected them to a worst-case stress test.This stress test combined the most powerful form of contagion, realistic constraints, and a nonstate actor in a safe haven facilitating violent radicalization.
The motivation for this research is not simply to satisfy the desire of the systems engineering community to branch out into sociotechnical domains.Policymakers continue to struggle to identify effective approaches for combatting terrorism.The global war on terror has not successfully countered the violence and instability of violent radicals, whether acting as individuals or groups.In 2000, 247 attempted or completed terror attacks originated from all sources in the United States (US) and Western Europe (WEUR).In 2017, there were 306. 9 2000, a US State Department report on terrorism estimated the operational strength of Al-Qaeda at "several hundred to several thousand members. 10" Eighteen years later, a similar report estimated Al-Qaeda's worldwide strength at over 33,000.Simultaneously, an entirely new global terror organization, the Islamic State of Iraq & Syria (ISIS), had an estimated 15,000 members scattered throughout different countries. 11 this context, we ask the following research question: "How can a systems engineering approach to policymaking provide unique findings to drive systemic impact using countering violent radicalization leading to terrorism as a case?"We use ISO/IEC15288 12 as the standard lifecycle framework for using a systems engineering approach to sociotechnical systems, it is important to clarify how this standard is conceptualized in a sociotechnical context.In the life cycle management of sociotechnical systems, society is an enterprise where the emergent behavior or end-product is not necessarily technical.In this case, national security as an emergent property of society (enterprise system) is the goal of the sociotechnical system, where stakeholders interact in different life cycle stages of the development of artifacts to allow national security to emerge from the system.To allow for this emergence, policy systems as artifacts emerge from sociotechnical systems in order to change the system itself to guide (rather than control) the life cycle of the system toward a more desirable emergent system behavior (in this case, towards national security).
In the sections that follow, we begin with a background on this policymaking context and then describe the methods for addressing the research questions.We then present our results, which center around four key findings, and provide a discussion around these results, F I G U R E 1 Lifecycle of violent radicalization and matching policy responses.
including study limitations.Finally, we conclude with a summary and possibilities for future research.

BACKGROUND
What constitutes terrorism and the strategies to counter it shifted during the global war on terror. 13Previous statutory definitions classified certain acts as criminal rather than terrorism: such as mass shootings lacking a clear political motivation. 14Over time, however, academic consensus adopted broader definitions such as those made by the University of Maryland's Global Terrorism Database (GTD): under these definitions, terrorism is defined as an intentional incident involving violence or threat of violence against people or property, and perpetrators may not be state actors.The GTD also looks for evidence in two of the three remaining categories.First, the action occurred outside "legitimate warfare activities."Second, the act must advance political, religious, social, economic, or widespread change.pp. 10-11 " This broader definition would reclassify over 1/3rd of US mass shootings as terrorism rather than criminal violence 14 and also addresses sociotechnical dimensions.We adopt this broader definition of terrorism and use an equally broad definition of combating terrorism as all "actions, including antiterrorism and counterterrorism, taken to oppose terrorism throughout the competition continuum 15, p. 39 " which implicitly includes counter radicalization.
Policy systems guide sociotechnical systems toward a specific desired behavior, and in defining policy options in this context, we graphically locate their intervention focus on a notional continuum of the lifecycle of violent radicalization in Figure 1.Along the top of

National security policy framework on terrorism
p. 30 The strategy "necessitates addressing the root causes of radicalization," mentioning antiterrorism, counterterrorism, and counter radicalization efforts, including those directed at "violent extremist content online. . .pp. 30-31 " The NSS reflects earlier language found in the 2021 National Strategy for Countering Domestic Terrorism, which articulates "strategic pillars" and corresponding strategic goals to both understand better and then pursue antiterrorism, counter terrorism, and counter radicalization efforts in Strategic Goals 1.1, 2.1, 2.2, and 3.1. 17,6][17][18][19][20][21][22][23][24][25] Although many of these goals are not new, Strategic Goal 3.3 clearly articulates a new risk of radicalized extremism infiltrating the Federal government, especially in positions holding clearance or national trust. 17,p. 26 From these directives, national security institutions responded by incorporating similar language and goals in their most recent strategic guidance.pp. 15-16 However, referencing the 2021 countering terrorism national strategy, the Secretary called for the Department of Defense (DoD) to "eradicate all forms of extremism in our ranks 19, p. 21 " under workforce goals.p. 8  threat assessments, and red flag laws.We first discuss these options for background in the sections that follow before relating these policies to testable parameters in the simulation in Table 3.

2.2
Policies in governed and ungoverned spaces

Antiterrorism policies
Antiterrorism is the "defensive measures used to reduce the vulnerability of individuals and property to terrorist acts. 22,p. 173, p. 17 " These measures aim to reduce the impact of a terrorist incident that has already begun. 23,24In Figure 1

Counterterrorism policies
Counterterrorism is "activities and operations taken to neutralize terrorists and their organizations and networks to render them incapable" of committing terrorist acts. 3,p. 5222, p. 52 We locate these interventions in Figure 1.Once a radical has activated, and they are engaged in active planning to commit a terror incident, but before they have begun the attack, known as "out the door. 9,p. 10 " US counterterrorism policy in the 21st Century focused on disrupting foreign safe havens.Efforts began along the so-called twotrack approach, 26 including strategic regime change to deny foreign terrorists a safe haven and breaking the connective links between global terror networks like al-Qaeda and indigenous local networks, 27, p. 243,27, p. 295,28, p. 8 It shifted to a three-track approach in 2008: depose, suppress, and retard, 27, p. 326 reflecting the perception that the risk of violent terrorism originated abroad in foreign countries.
Increasingly, however, the concern has shifted to digital safe havens consisting of websites, content platforms, or social media 21, p. 11 and violent radicalization occurring in the domestic space.

Counter radicalization policies
Counter radicalization efforts target populations of high-risk people and their communities.These policies seek to break the cycle before radicals initiate terrorism.It overlaps with counterterrorism but differs in that it aims to cause the perpetrator to abandon their radical ideology or shift it from violent to nonviolent.
Counter radicalization is another tool to reduce violence associated with terrorism.Past successful strategies in this category include the reduction of the grievance that motivates the perpetrators, undermining the previous social identity of the high-risk population, creating a new higher-order social identity that trumped the in-group, as well as "proportional. . .pp. 150-153 In a meta-analysis of 22 studies on mass shootings, nearly half of the studies (46.5%) suggested counter radicalization strategies, including community advocacy, to reduce radicalization's root cause factors, such as grievance. 30They also found high support for deradicalization efforts, including individual, family, peer-group, school, and societal interventions. 30other systemic review in cooperation with the Department of Homeland Security examined 19 studies from 2000 to 2018 examining counternarrative strategies, including stereotype-challenging, inoculation, alternative accounts, prosocial and moral "exemplars 31 ." However, few studies showed an impact on the primary outcomes, such as intent to act violently.The review showed small effects on secondary outcomes at "targeting realistic threat perceptions, in-group favoritism, and out-group hostility" but showed no effect in reducing "symbolic threat perceptions or implicit bias. 31"

Focused deterrence
Research differentiates mass killings in the United States by type, with the three most frequent and lethal being "family," arising from intimate partner violence (IPV), "felony" that accompanies other parallel criminal activity or gang motivations, and "public" which involve an individual trying to kill as many people as possible in a public place. 32 Associated Press (AP)/Northeastern/USA Today online database between 2006 and 2023 includes 477 mass killing incidents, which combined resulted in 2533 fatalities. 33Analyzing that dataset by type of mass killings indicated family accounted for 54% of the incidents and 46% of the fatalities; felony for 24% of the incidents and 20% of the fatalities; and public for 22% of the incidents and 33% of the fatalities. 34Although public mass killings dominate media focus, a policy portfolio known as focused deterrence, or pulling levers, targets IPV and criminal gang violence, produced promising results in reducing both individual homicides and mass killings of those types.
This policy approach spans antiterrorism and counterterrorism measures.It does this by identifying risk factors of the targeted criminal behavior through hyperdimensional analysis that identifies a very small group with the highest crime risk of conducting a crime.This highrisk population receives interventions focusing on services, support, and deterrence-communication rather than incarceration. 35p. 238 When applying the method to domestic violence, one study found a reduction of injuries by 29%, from 66.8% to 47.3%. 36

Terror contagion containment
Another policy portfolio approach treats violent radicalization as a form of social contagion and targets public mass killings that focused deterrence does not.Though there is no empirical evidence of using social contagion policies to reduce terrorism, there is a solid foundation for countering other forms of social contagion that lead to harmful effects.One of the most well-studied social contagion effects occurs after a celebrity suicide spawns imitation suicides, known as the Werther effect. 37The social contagion is activated when mass media reporting on a celebrity suicide conveys both ideation and modus operandi of the celebrity's death.A high-risk segment of the overall population who recognize the celebrity and view themselves as selfsimilar to the first suicide are at the greatest risk of imitating both the attempt and specific method of suicide the celebrity used. 38Based on research that began in the 1970's 37 and continued through the 1980's 39 first the Centers for Disease Control (CDC) in 1989 40 and later the World Health Organization (WHO) in 2000 41 released media guidance on how to report on celebrity or sensationalized suicide.
Over the decades hundreds of studies have both demonstrated and improved our understanding of the Werther effect 42,43 confirming it's not limited to the United States, 44 . 45The effect varies by suicide but a general increase of 13% for suicides overall and a specific increase in suicides of the methods depicted by 30% can occur when media depictions do not follow the best practice guidelines. 43Researchers are now even able to forecast suicide clusters before they happen if fictional portrayals of suicide do not take into account the Werther effect as was the case in Netflix's 13 "Reasons Why" television series. 46cent research suggested a terrorism contagion hypothesis as a model for understanding violent radicalization as a social contagion similar to the Werther effect. 21We developed a system model of radicalization, including numerous simultaneous potential causes.Then, through a process of evaluating top-down and bottom-up causes, we narrowed those down to seven "root causes" identified as a terror contagion hypothesis depicted in Figure 2  Although these results are preliminary, the potential is significant.
In the AP/Northeastern/USA Today mass killing database, 33 an analysis identified that 57% of all public mass killing fatalities, or 848 deaths, arose from suspected terror contagion incidents.Of those suspected terror contagion incidents, 66% or 484 fatalities from 2006 to 2022 are from the top five suspected terror contagions. 34Though this analysis only covers the US, where researchers more carefully track mass killings due to their frequency, the percentage of fatalities attributable to terror contagions is likely to be higher in other countries where mass killings are less frequent.A separate dataset of over 250 suspected contagion incidents identified occurrences around the world of the contagions mentioned above, including Austria, Brazil, Canada, Finland, France, Germany, Italy, Mexico, Netherlands, New Zealand, Norway, Poland, Portugal, Russia, Spain, the United Kingdom as well as in the United States. 34Although this does not discount the possibility of variations between countries based on development status, and the data is likely skewed to countries that can access cultural scripts in English, it does indicate that these contagions are global.
p. 12 This style of intervention focuses on cutting the positive feedback loop by interrupting the spread of recognizable cultural scripts at the location indicated in the CLD in

Realistic policy constraints
None of these policies operate in ideal circumstances.Two of the many realistic constraints these policies face are important to mention.
The first realistic constraint deals with changes in the size of the high-risk population.Best-case scenarios assume a stable population.Scenarios of rising tides and backlash represent exogenous and endogenous influences.As a result of different conditions, both influences increase the high-risk population over time.
A rising tide is any exogenous factor causing the high-risk population to increase over time.pp. 4-5 Rising tides occur when exogenous forces amplify these factors across broad swaths of the population.pp. 99-100 Local or foreign conflicts providing sources of outrage and ideology can also create rising tides, 29, pp. 153-154 as can other changes that lead to increasing internet use.These exogenous factors are outside a policymaker's control, distinguishing rising tide from backlash scenarios.Backlash occurs when an intervention generates Type I errors (false positives) that target high-risk populations who have not radicalized or activated yet, or worse, the larger population within which a small high-risk population occupies.Where these Type I errors generate more harm, such as incarceration or curtailment of rights, than is perceived to reduce violence, the intervention can increase perceived grievance.Historically, backlash scenarios occur when violence reduction programs use tools of "eradication or forceful repression," which only worked against small communities with high risks of failure 50, p. 214 -targeting larger populations with the same methods increased sustained violence.Another realistic constraint is when a policy becomes a victim to its own success.In the wake of completed terror attacks, a high perception of risk encourages Federal agency, local law enforcement, and public support to policies.But if those policies prove successful in reducing terrorism frequently-the support may erode.In public health policy, this dynamic is known as the Health Belief Model (HBM).The HBM associates public risk perceptions with risk-prevention measures they may take or support. 51HBM was used to forecast, evaluate, and review the impact of population level adherence to, and policy enforcement of, public health policies concerning the COVID-19 pandemic. 52,53First, society/individuals acts with more vigilance and attention if there is a societal belief that terrorism will affect them.Second, as the perceived severity of a negative outcome increases, the more motivation to act to counter or mitigate it.Third, the individual must perceive the targeted behavior as protecting from the negative outcome.Finally, if individuals perceive strong barriers to adopting a model, they are less likely to adopt it. 51HBM concepts allowed early forecasting of second and subsequent waves due to "premature relaxation of interventions 52, p. 1321 ", which later materialized with a second/third wave significantly worse than the first. 54p. 1146 The study found adherence to measures reduced over time across regions and income levels, but the pace of reduction varied by policy.p. 1155 We extend the concept of an HBM to model societal-level reactions to terrorism.In our simulations, HBM models varying societal support and adherence to policies that occur outside of a backlash effect.Crucially it can dynamically represent how a successful effort at countering terrorism reduces perception of risk, which erodes public compliance to policies, thus counterintuitively raising the risk of terrorism over time.

METHOD
Our method of evaluating policies involves an existing peer-reviewed model, the terror contagion simulation.This system dynamics simulation was designed to study the terror contagion hypothesis and meets DIME-PMESII military standards for simulations.The original publication includes complete model documentation and discussion. 47In  To create baseline simulations for policy comparison, we leveraged five base runs.We list the key parameter values in Table 1.
We used common values for most parameters, including a starting high-risk population of 600, an out-of-door incident success rate near the global average of 80%, 55 and 10 fatalities per completed template method incident.We varied the pathway to violence success rate.This rate is the chance of an intervention stopping an activated perpetrator during their planning stage before they go "out the door" (OTD) to start an attack.We selected this one for variation because it represents a template method factor, allowing us to hold high-risk population values such as perception of near suffering, personal resonance to near suffering, normal activation, and normal abandonment constant.Additionally, the other template method factor, OTD success rate, does not vary substantially in our terrorism research, with a global average approaching 80%. 55th this simulation, we could recreate the growth reference modes we identified from historical time series data in previous research of over 4600 incidents from the GTD, 9,55 presented in Figure 5.The radicalized and proceeding on a pathway to violence that ends when they initiate an incident.
The simulation continuously calculates changes in radicalization at the individual and societal levels.We model terror incidents within the simulation as discrete events using five stochastic random number generators (RNG).To ensure any results we see are not the outcome of a particular RNG sequence, we vary the RNGs of terror incidents to create 1000 permutations of each baseline run (see simulation Supplementary materials for more discussion on using this stochastically discrete formulation).In Equilibrium (EQ), there is no seed event.No violent ideology or template method spreads across the high-risk population.Although there may be normal crime levels, there are no terror incidents.
In the Failure to Grow (F2G), there is a range of zero to 32 incidents with an average of 3.2.F2G reflects terror contagions with a seed event that failed to provoke a strong contagion.Reasons may include a small susceptible high-risk population with self-similarity to the perpetrator.Alternatively, cultural scripts of the ideology and template may be too hard to understand to replicate.Known as low coherence or cohesion. 21,[26] In Struggle to Grow (S2G), there is a range of zero to 105 inci- We also base our criteria for evaluating policy effectiveness on

First finding: Popular policies in isolation are not effective even under hypothetical best-case conditions due to feedback effects and a system memory of radicalizing scripts acting like a viral reservoir
The first finding of our experiments is that popular policies most frequently raised by the public and politicians after a mass violence incident are ineffective at containing or reducing contagions.pp. 17-21 The results in Table 4 indicate at what levels of reduction these policies shift CONT behavior to a lesser S2G, F2G, down to EQ or ideally below EQ.
Unfortunately, these experiments are not promising for two reasons.p. 36 Each mass violent incident expels countless cultural scripts to reach high-risk populations who may become "infected" through radicalization.In this analogy, cultural scripts conveying template ideology and template methods that remain accessible within the system memory of the high-risk population are a reservoir for the virus (explained further in our fourth finding below.)The high-risk population leverages this cultural script memory reservoir, especially on template method, to learn, adapt, and innovate more successful and lethal attacks.The thresholds to create and sustain contagions are low.
p. 27 Even if years pass between completed inci-dents, the reservoir of cultural scripts can still radicalize the high-risk population.p. 22 Even if the pathway to violence took 24 months of planning, if there was a high enough fatality rate resulting from the planning, the contagion incident could spawn enough new cultural scripts to refill the reservoir.We confirmed this experimental result by reviewing the historical record of suspected contagions.Varies the 25% threshold necessary to shift from CONT to S2G style contagions. 63p. 153 However, this is only half the necessary fatality reduction to shift from a CONT to an S2G behavior of 25%-34%.Because media exposure is driven by fatality counts, reducing injuries, though beneficial, does not affect contagion behavior materially.Furthermore, the same research found that among all mass murders, only 20%-58% use LCMs, 64, p. 151 and template cultural scripts allow future perpetrators to adapt and innovate by creating an ad-hoc LCM, for example.Finally, the bivariate analysis of comparative fatality rates did not account for circumstantial factors, including: "the intentions, motives, mental state, and skill of the shooter(s); the nature of the circumstances surrounding the shooting (e.g., offender and victim relationships); the type of location where the shooting occurred (e.g., whether it was indoors or outdoors, the type of venue, and how confined potential victims were); the number of people present who could have been shot deliberately or incidentally; the characteristics and health of potential victims; the number of shooters; p. 155 " Yet these circumstantial factors are exactly what the cultural scripts of the template method conveyed allowing perpetrators to learn from past incidents.
p. 238, 36 Finally, these synthetic proposition tests were made under bestcase conditions and not subjected to real-world constraints as described below, where presumably, benefits would be even lower and unforeseen consequences might be higher.

Second finding: Even effective policies under realistic constraints, including backlash effects, can struggle to improve outcomes
The second main finding of our policy experiments results from policies that show effectiveness in best-case conditions but whose results erode under realistic constraints.Table 5 lists three possible realistic constraints identified in the literature.
Two common realistic constraints are the backlash effect and rising tide scenarios.Backlash effects occur when an intervention policy designed to target a very small high-risk population of ∼600 people creates increased perceived grievances in the larger population, which then becomes high-risk.Rising tide scenarios are when some exogenous socioeconomic, cultural, or circumstantial factor causes a rising increase in the pool of high-risk individuals.Under either scenario, a RAMP function represents the growth by increasing Additions to Undecided by 2% beginning at month 12 and continuing for five years.
Over this time, Additions to undecided/month will increase from 10 people/month to 20 people/month, which will hold constant at that rate.
The variable societal response represents the HBM dynamic identified in the literature review.It dynamically alters the strength of certain policies that require public support based on the public's perception of risk.We model HBM through a Variable Societal Memory that increases after each completed terror contagion incident of mass violence and fades over time absent reinforcement.Although highly abstracted, we feel this sufficiently represents the HBM dynamics for a model of understanding.
We explore how realistic constraints impact otherwise effective policies using a focused deterrence strategy modeled as combining antiterrorism and counterterrorism methods.Based on findings in the literature, we used a 20% reduction applied to CONT base run starting parameters of the two success rates and perception of near suffering, as shown in Table 6.We then experimented using these values for focused deterrence under both best case and a backlash scenario.
We compare the results of a focused deterrence effort under bestcase conditions versus a backlash scenario in Figure 6.
The behavior modes in Figure 6 show that a focused deterrence policy initially decreases the activated population in response to the seed event and, under best-case conditions, results in a reduced contagion effect.However, it rises at the end of the ten years, and if a backlash is sparked (#4), it becomes a better-before-worse effect.An initial decline in the CONT behavior makes it look like focused deterrence measures are working.However, the increasing additions to the high-risk population from the backlash resulted in more becoming radicalized and activated after a time delay.If the simulation continued, it would be far worse over time, settling into a new higher equilibrium.

Third finding: Perceived grievance determines the carrying capacity of terrorism in a system. The carrying capacity can restore or renew temporarily disrupted radicalization networks
How perceived grievances operate as a carrying capacity of the system of terrorism is an important concept to understand for crafting effec-tive policies to combat terrorism.The perceived grievance is the source of violent radicalization and can restore violent radicalization networks disrupted by policy interventions as long as it exists.
pp. 118-119 From ecosystems to complex systems, latent carrying capacities are the support structures from which activities arise and create policy resistance to reducing those activities.p. 1 Experiments into disrupting safe havens illustrate this finding.
p. 5816 We are not modeling the disruption itself, which differs based on the type of safe haven.For example, serial raiding 67, p. 37 may represent strikes or punitive expeditions against nonstate actors exploiting a physically ungoverned space to disrupt these safe havens.In a digital environment, serial raiding may instead represent periodic deplatforming, closure of websites, or even mass-banning of accounts on mainstream platforms such as Twitter.However, our research focus is not on the implementation aspect of policies but on the impact those policies have on countering violent radicalization leading to terrorism.In this sense, even though disruption techniques differ from physical to online safe havens, the dynamic that arises from A key insight here is the relationship between long-term grievance and the casting capacity of cultural scripts.In the simulation, the safe haven casting capacity of nonstate actors develops to meet the demand of the grievance via an implicit adjustment generic structure. 68Over time, the value of grievance in Figure 8 generates a desire to create TA B L E 1 0 Number of contagion incidents of safe haven (SH) intervention after 1000 permutations.A "-> " indicates a strengthening of a contagion from its initial baseline behavior mode.

Policy
Intervention in ungoverned space ungoverned space grievance.
the infrastructure necessary for the casting capacity of nonstate actors at the same scale as the grievance.In goal-seeking behavior, casting capacity increases or lowers to the level of grievance.
Continuing the analogy, counterterrorism policies like serial raiding to disrupt casting capacity are like pulling weeds in a garden without cutting off the water supply or attacking the roots.The weeds will grow back.
We can observe the pulling-weeds effect in Figure 9, first for the S2G Serial Raiding of Safe Haven policy which aims to disrupt the casting capacity.Even though serial raids periodically eliminate the casting capacity to zero-it always grows back to the level of carrying capacity of grievance in Figure 8.The coupling generates the oscillating pattern of the Activated Population in Figure 7. Periodic raids disrupt casting capacity, but that capacity will return as long as the underlying grievance remains.When counterterrorism serial raiding creates backlash effects, the grievance increases in Figure 5, and the cast- Intervention in ungoverned space casting capacity.
ing capacity in Figure 6 grows to increasingly higher levels, becoming worse than doing nothing at all.
A contemporary example of challenges pulling weeds emerged during the course of submitting this article.The utility of the long-standing Israeli policy of "mowing the lawn" through serial military raids of Hamas forces in Gaza 69,70 came under question after the Hamas invasion into Israel on 7 October, 2023. 71 addition to the pulling weeds effect, the second challenge of serial raiding into safe havens is the time it takes to implement and decisively eliminate casting capacity.In that time, a contagion can take hold in the governed space and become self-perpetuating.In Figure 8, grievance reduction reaches its maximum decline after month 60 and cuts grievances by ∼75%, resulting in a correspondingly low casting capacity in Figure 9.By then, however, repeated contagion incidents and mass media reporting have filled the domestic system memory with a reservoir of cultural scripts.The casting capacity of nonstate actors in the safe haven is no longer needed to sustain the contagion (see Fourth Finding below).

Fourth finding: Public support for interventions varies based on perceived risk.
Effective policies reduce the perceived risk causing reduced public support and compliance.This can lead to subsequent waves of violence emerging via a latent support from the system memory, or viral reservoir, of radicalizing cultural scripts The final key finding in our experimental research is how the public perceives the risk of terrorism and how support may decline and undermine successful policies as risk declines.We illustrate this counterintuitive effect by exploring the experimental results of a novel policy.The terror contagion containment experiment combined three distinct interventions.First, reduce the mass media effect after each terrorist incident to limit the spread of cultural scripts relating to template ideology and template method.This would build upon existing media guidelines 40,41 for reporting on celebrity suicides to reduce negative Werther effects that spread scripts while increasing positive Papageno effects that diminish the spread of radicalizing content 48  Preliminary experiments under best-case conditions against a normal contagion proved promising, so we tested against a worst-case scenario that consisted of a strong contagion (CONT+) supported by a nonstate actor in the safe haven while a rising tide occurred.We also activated the Societal Variable Rate (SVR), which is our construction of HBM dynamic modeling the rise and fall of public support and adherence to policies relative to their perception of risk.These two experiments show the strong effect of the contagion containment approach on eliminating terror contagions, as displayed in Table 11.Visually examining worst-case scenarios helps explain this effect.
Without intervention, the baseline of this worst-case scenario demonstrates a rise to a sustained higher equilibrium of violence in Figure 10 (Top #1) below.
Introducing a terror contagion containment policy against this worst-case scenario leads to an initial decline in violence in Figure 10 (Top #2).However, a sharp second wave reaches a higher peak of Activated high-risk than the first one under worst-case CONT+ conditions, which is why numerically, overall violence levels are similar.This effect of a second wave significantly worse than the first is behaviorally similar to subsequent COVID-19 waves predicted and later confirmed by the HBM of variable public adherence Figure 10 (Bottom), even though time scales of manifestation differ. 54A third intervention demonstrates mitigating this second wave of violence by adding focused deterrence and grievance reduction in Figure 10 (Top #3) in combination with contagion containments.
The dynamics that drive these results are important for policy research and are shown in Figure 11.The broadcast memory is the "viral reservoir," or system memory, of cultural scripts in the high-risk population conveying ideology and method.Failure to drain this reservoir or memory is what stymies most popularly debated policies in earlier experiments.However, when new media guidelines on terrorist incident reporting are implemented in month 24, the level of these cultural scripts declines.
Furthermore, the lower level of cultural scripts in the broadcast memory has reduced template attractiveness due to   Only using focused deterrence and grievance reduction performs much worse, creating another higher equilibrium of violence Figure 13 (#2).Table 12 provides the numerical results of the experiments.A <symbol in a column indicates a weakening from a stronger baseline to a weaker behavior mode, from CONT+ to CONT for Policy # 9.
These results emphasize the key role contagion containment plays as the primary means to target the positive feedback loops of social contagion.

Summary of policy evaluations
We summarize the numerical results of our policy experiments in We locate these portfolio policy evaluations on the lifecycle radicalization diagram in Figure 14.We did not find any single policies that worked in isolation and marked them in red as "NA."

Implications
In World War II, analysts and strategic planners identified the German military-industrial complex as a critical strategic target to disable the capability of the Nazis to fight.Planners could confidently target manufacturing and transportation infrastructure with strategic bombing and see the outcome of those efforts downstream within a military supply chain.Without that material and logistics support, the war in Europe could not continue.
The key implication of this paper is to suggest to policymakers a potential "industrial complex" for violent radicalization leading to terrorism.

Limitations
Limitations accompany these findings.Our dataset of over 4600 terror incidents from which we drew the historical reference modes and generic parameter values for the simulation only covers Western Europe and the United States from 1995 to 2018. 55A separate dataset of over 250 suspected terror contagion incidents has a broader representation of 17 countries, including six not in the above data set, such as Austria, Brazil, Canada, Mexico, Poland, and Russia. 34 Excluded from the scope is terrorism conducted in a conflict zone or counterinsurgency, which we have researched elsewhere. 73,74Data from other regions or time periods may create additional behavior modes.
As a model for understanding, we do not calibrate our simulation to a specific violent ideology or high-risk population.We conceptualized many parameters as their influence on a percentage of the population, which can obscure the causal mechanism of how they create the effect modeled.For example, when Template Self-Similarity = 0.75, then 75% of the high-risk population will see themselves in that Template.Still, the specifics of how that occurs are left to further research.
Additionally, the simulation is in an early stage of development.The effectiveness of certain policies like focused deterrence or a terror contagion containment strategy may change on further revision to the simulation.Also, these are novel approaches to combating terrorism.
Finally, we may not have fully modeled unforeseen challenges in implementation and the impact those implementation challenges have on the effectiveness of policies.

CONCLUSIONS
Our goal in this paper was to show how systems engineering can be used in policymaking by conducting simulated experiments into antiterrorism, counterterrorism, and counter radicalization policies for combating terrorism under the premise that violent radicalization operates as a social contagion known as a terror contagion.An important part of this is to show how the analysis of policy life cycles from established methods not commonly applied in systems engineering environments (i.e., system dynamics to evaluate sociotechnical systems and policy life cycles) could be brought into a systems engineering context-using a domain that is ubiquitous in systems engineering literature, defense.The first key finding was that terror contagions are driven by positive feedback loops, like a viral infection.The more highrisk population activated to commit acts of terrorism, the larger the number of cultural scripts operating to radicalize others.A system memory, like a viral reservoir, retains cultural scripts conveying violent ideology and methods of mass violence.The reservoir can then reinfect high-risk populations years later.These two factors combine so that once a terror contagion passes a threshold, it gains a high fault tolerance, and years can pass between subsequently completed attacks.
The second finding was that even when policies are effective in the best case, if they spark backlash effects by being too broadly targeted, they can generate perceived grievance within the high-risk population or the larger population within which the high-risk population exists.
This backlash effect can cause a steady increase in the high-risk population, which undermines the policy's effectiveness over time.The third finding shows how this perceived grievance is the carrying capacity of the system of a terror contagion serving as both the source of a terror contagion and renewing or restoring violent radicalization networks disrupted by periodic interventions.Our fourth finding shows how even a policy that overcomes the previous three challenges can struggle over the long run because as terrorist violence reduces, public perception of risk reduces, and public support for successful intervention declines.
Though early-stage, these experiments in policy research help understand the opportunities and challenges of combatting terror contagions.Future research would advance the simulation from its current form by creating simulations calibrated to specific violent ideologies and corresponding high-risk populations in different geographic regions.Simulation improvements could build upon our existing profiles and data set of 4500 incidents leveraging GTD data. 55Incorporating clustering, hyperdimensional, or machine learning techniques could improve profiles and support calibration. 75Additional calibrations could incorporate natural language processing of social media to identify specific cultural scripts used in a profile.
Finally, within our stated limitations, we note that focused deter- counter radicalization efforts.Each targets a different phase of violent radicalization leading to terrorist violence.Antiterrorism and counterterrorism target already violently radicalized perpetrators in motion toward a terrorist incident.Counter radicalization seeks to remove the motivation to conduct attacks altogether, either by treating violent radicalization as a social contagion or reducing the grievances that give rise to radicalization.Policymakers may also choose between intervening directly against those likely to become violently radicalized or against nonstate actors operating from safe havens to facilitate radicalization.Safe havens can exist in foreign physical spaces requiring military interventions or online digital spaces suggesting account purges, deplatforming, and website bans as policy options.Unlike foreign physical spaces, interventions into online spaces may come from either state or private commercial actors attempting to enforce content policies.

Figure 1
Figure 1 is a time scale indicating the duration of time typically seen for each phase of the violent radicalization lifecycle.Below we summarize the current national security policy framework as well as providing some examples of each specific policy option.
" This network of strategic guidance creates demand signals throughout Federal Departments, and their subordinate Services and Agencies.Policymakers have a variety of options to respond to these demand signals and sociotechnical systems engineering including simulations, can help evaluate and deploy solutions among these options.Policies can target governed or ungoverned spaces and mix and match across the three categories including enhanced Law Enforcement Organization (LEO) and bystander response training, hardening facilities, gun control laws, community awareness and advocacy, counter terrorism investigations, , we locate these measures after the planning stage and once a violent radical has gone 'out the door' , a term for describing the start of an actual terror attack.As the definition of terrorism broadens, research into reducing the impact of mass shootings now falls into the realm of antiterrorism.A comprehensive review of mass shooting research of 73 papers, most published between 2014 and 2019, identified 40 papers making a policy suggestion.Most dealt with antiterrorism proposals, including professional training for mass shooting response (60%), bystandertraining for mass shooting response (20%), and gun-control laws to mitigate the lethality of mass shootings (42.5%).25 in the form of a causal loop diagram (CLD): ". . .for the purpose of replicating itself, a terror contagion exists in the form of a template ideology and method suitable for social contagion [Figure 2A].This combined template exploits a contextual circumstance of existing grievances and moral outrage well suited for radicalizing [Figure 2B].The contagion's cultural scripts communicating template information find their way to a high-risk population filtered by similarity, notoriety, and coherence biases as well as already being susceptible to violent radicalization [Figure 2C].The templates initiate a radicalization process which is the immediate cause for activating an existing mammalian adaptation for predatory violence [Figure 2D].Following the template method is the physical cause of predatory mass violence terrorism [Figure 2E].If following the template method generates enough fatalities, the media will disseminate cultural scripts communicating the template ideology and method [Figure 2F] to the extent of media reach [Figure 2G].The spread of cultural scripts sets conditions for subsequent replication of individuals adopting the template ideology and pursuing the template method, allowing the terror contagion to become self-perpetuating [Figure 2A] [21, p. 34]."Research building confidence in this hypothesis began with an exploratory DIME-PMESII simulation.DIME-PMESII are a broad category of simulations useful for modeling sociotechnical systems as they explore how diplomatic, information, military, and economic policies can interact with political, military, economic, social, information, and infrastructure outcomes.In this context "law enforcement" can replace "military."Our terror contagion DIME-PMESII simulation, the same used in this paper, identified that all causes of the dynamic hypothesis finding had to be simultaneously present in combination, or no growth mode would occur in terrorism.Falsifying this, all other propositions listed could alter the behavior mode's shape but not eliminate it based on its absence.47,pp. 22-23 Empirical research then used the simulation to recreate multiple historical contagions from starting conditions, Columbine-style school shootings, VA-Tech-style school shootings, and Great White Replacement Theory with acceptable behavior mode accuracy.34Another prediction of the hypothesis was that a successful template method could crossover between violent ideologies given temporal proximity, using the case of ISIS vehicular attacks in Europe leading to the Finsbury Bridge vehicular ramming and then the Charlottesville protest ramming all in 2017.The Counter Extremism Project has since identified nearly 40 vehicular ramming incidents in the May-September 2020 time period, confirming the use first as a right-wing and then spreading to left-wing template methods for attacking crowds of protestors (Vehicles as Weapons of Terror, no date)

Figure 3 . 48 F I G U R E 3
Figure 3.The purpose of contagion containment is to reduce the influence of cultural scripts spread by the media after each terrorist act of mass violence.This effort to reduce social contagion spread is based on significant prior research on how to contain Werther celebrity suicide contagions through similar measures.48

F I G U R E 4 7 Figure 4
Figure4depicts an aggregate view of the Terror Contagion Simulation "core model," which contains five system levels: incidents (1), agents (2), networks & actors (3), system of spaces (4), and system of systems(5).Each level represents one layer of the system's structure within which key dynamics occur.The arrows in Figure4represent the upward and downward causation of these causal influences crossing between system layers.Excluded from this depiction for clarity are modules containing model documentation, model values, and testing structure.

5
behavior modes show the Activated Population of the simulation, representing the number of high-risk people who are both violently TA B L E 2 Contagion incidents in base runs across 1000 permutations.Simulated base runs of growth modes of terror contagion.
dents and, on average, 27.7 over the 10-year simulation.An S2G is the abstracted representation of a brief terror contagion that fails to sustain over time, such as the wave of vehicular terrorism by ISIS supporters in Europe in 2017.57In the contagion baseline (CONT), contagion incidents jump to a range of 57-372 with an average of 259.8 incidents over ten years.The CONT represents recognizable terror contagions initiated by singular seed events that have served as patterns for replication.Representative seed events for CONT style contagions include the Columbine attack for school shooters in 1999, 2 the Norway attack of 2011 for Great Replacement Theory white nationalists, 58, pp. 17-18and the Isla Vista attack for so-called incels in 2014.59A strong contagion (CONT+) is the baseline behavior when conditions are ideal for a terror contagion.CONT+ is significantly worse than CONT, with a range of 469-806 incidents and an average of 704.Incidents.Over time, these levels of sustained violence indicate mass instability and may represent a nascent insurgency or violent separatist movement.

Figure 5 and
Figure 5 and Table2.Through visual inspection,60 a policy able to adjust the behavior mode in the activated population down to lower baselines shows merit, reducing the severity of the contagion.A policy that drives the terror contagion behavior down to or below the EQ mode demonstrates an ability to reduce mass violence terrorism to at or below normal criminal violence.Numerical values bolster visual inspections by demonstrating a net reduction in mean contagion incidents over 1000 permutations.We use <or -> in table results to indicate when a behavior mode has strengthened (->) or weakened (< -) from its initial baseline behavior mode.A policy that changes the numerical value of the contagion incidents but fails to adjust the behavior mode is not considered useful.

9
Parameter values for safe haven intervention tests." in either form provides the key insight into carrying capacity.In a series of experiments outlined in Table8, we first show how a struggling domestic contagion (S2G) increases to a Contagion (CONT) behavior by introducing nonstate actor (NSA) cultural script casting from a safe haven.Then, we subject the safe haven to two policy experiments.The first is serial raids,67, p. 37  such as military operations in a foreign country, de-platforming, or periodic account purges for a digital safe haven.Second, counter radicalization interventions in the safe haven focus on grievance reduction, which could range from aid programs, additional support services, or other information operations efforts to address the underlying grievance.We assume that targeting the grievance provokes no backlash.We use the same parameter values for backlash as in Realistic Constraints Table5, while Table9 detailsparameter adjustments for serial raiding and counter radicalization policies.These experiments highlight the difficulty of dealing with safe havens, as shown graphically in Figure 7.Because of the facilitated radicalization from the safe haven, the contagion no longer requires a seed event and initiates earlier.Although behavior modes vary, they all fall within the visual depiction of a CONT behavior, confirmed by the numerical end values of these scenarios presented in Table 10.A "-> " symbol in a column indicates strengthening a baseline behavior mode to the next behavior mode.All policies reflect starting as S2G, boosted to CONT by nonstate actors casting cultural scripts.None of the interventions could reduce the behavior mode to S2G or lower.Table 10 (#3) shows an S2G strengthened into a CONT behavior with a mean of 184 contagion incidents and a range of 157-210 F I G U R E 7 Behavior modes of safe haven intervention scenarios.contagion incidents absent intervention.Even though the mean performance varies by intervention, all still fall within that range.Serial raiding Table 10 (#4) shows almost no change to the mean number of contagions or range, while an intervention provoking a backlash Table 10 (#5) is worse than doing nothing.Only an intervention focusing on grievance reduction reduced the mean and slightly reduced the range of contagion incidents, but not enough to shift it back to an S2G behavior mode.
modified for use in terror contagions.Second, template attractiveness is weakened by flooding confusing or contrary cultural scripts to obfuscate self-similarity and reduce cohesion, known as counter-reification.Reduced self-similarity means the highrisk population has difficulty seeing themselves in the perpetrator.Reduced cohesion makes it difficult for high-risk populations to fully understand the template ideology and template method from cultural scripts.Finally, we employ failure notoriety to counteract the notoriety of completed events.Failure notoriety are media broadcasts of failed incidents instead of just completed ones.Creating and broadcasting cultural scripts of failures, arrests, or other negative outcomes diminishes perceived notoriety and increases abandonment among the radicalized high-risk population every time an incident fails.
The < -symbol indicates a weakening from the baseline starting behavior mode to a lower mode.Two arrows indicate a shift of two positions, from CONT+ to S2G for policy #7.A contagion containment intervention Table 11 (#7) is the most effective intervention we have tested, even with the realistic constraint of a variable societal response.However, in a worst-case stress test, the intervention struggles to alter violence (#8) significantly.

F I G U R E 1 0
Top. Contagion containment intervention on Activated Population.Bottom COVID-19 subsequent wave effects, US MAR 2020 to FEB 2021.

F I G U R E 1 2 F I G U R E 1 3
Figure 13 compares all the combinations of three policies (#1) with only focused deterrence and grievance reduction combined (#2).

F I G U R E 1 4
Developing and deploying these integrated approaches of technology, policy, and regulation successfully is a challenge systems engineering is uniquely equipped to contribute.The evaluation of policy life cycles and using a systems approach to evaluate policy in other domains are more established than in systems engineering.However, a Lifecycle of violent radicalization with policies plotted.disconnect often exists between these approaches and system conceptualization established in standards used in practice, such as ISO/IEC 15288.It is critical to link theory, and practice of the real-world impact of bringing these outside policy system analysis perspectives into systems engineering is critical.Systems engineering is embedded in the US DoD and other government agencies in the US that have policy proponency to combat terrorism, reduce extremism, and counter violent radicalization.The practice of systems engineering outside of an academic context means that the societal transfer of academic work can make impactful changes in improving national security.
Differences in national economic, individual socioeconomic, and media consumption patterns differ across this space, and any policy work would require local consideration and customization.Although violent radicalization can fuel both affective (e.g., opportunistic and unplanned) and predatory (e.g., targeted and planned) violence, "virtually all acts of terrorism are predatory (instrumental) violence.72,p. 10  " Affective violence arising from radicalization is out of the scope of our research.Policies not useful for reducing terrorism may still prove useful for countering other forms of violence.Finally, although we simulate policies on countering the influence of nonstate actors at an aggregate level, we are focused on combating terrorism in domestic nonconflict zones.
rence and terror contagion containment have significant empirical evidentiary support in adjacent domains.Numerous ongoing pilots of focused deterrence show its benefit against criminal gang violence and domestic partner violence.There are several decades of research on countering the Werther effect of celebrity suicide contagion.Although applying these methods to combating terror contagions will require work, it is not starting from scratch.These results should encourage additional research under the premise that violent radicalization is a form of social contagion, either through the terror contagion hypothesis or other alternatives.Developing strategies along this line to counter violent radicalization may open new effective opportunities for reducing terrorist violence.
50, p. 214More recent contemporary examples of backlash effects generating perceived grievance within larger populations that a high-risk population may exist within include stop and frisk procedures in New York City and broad targeting of Muslims in the wake of 9/11.Ultimately, backlash at the societal level may lead to erosion of public support and abandonment of existing measures or resistance to new measures deemed controversial, such as gun control laws in the United States.
overview, the terror Contagion Simulation Models a single abstracted and generic high-risk population and its internal network dynamics within larger societal dynamics.At the beginning of the simulation, this high-risk population supports no violent ideology nor employs a template method of mass violence.If they commit violence, it is indistinguishable from normal crime.The simulation initiates with a seed event.The seed event is a single contagion incident, a terrorist attack 55 template methods (set at .8 and ten respectively in this paper), factors related to the high-risk population, and the extent to which this violent ideology is, or is not, supported by nonstate actors operating in a safe haven.Profiles also contain policy response options activated as switches to test policy responses against a specific violent ideology.In this paper, the profiles are generic, using average values determined from prior research across a continuum of terrorist behavior identified in both the US and WEUR.55System dynamics concerns itself with determining cause and effect (Forrester, 2003, p. 342) through understanding feedback in complex systems (Richardson, 1985, p. 1).Five characteristics define system dynamics modeling from other forms.First, models are based on a causal structure with feedback so that a change to a single variable causes a chain reaction that eventually influences the first variable.Positive (reinforcing) feedback loops push a system in one accelerating direction, leading to exponential growth.Negative (balancing) feedback loops balance the systems towards equilibrium.The interaction of these loops determines how its output behavior unfolds over time.Second accumulation and delays in these feedback loops are foundational, represented by four building blocks.Stocks accumulate over a delay.Flows determine the rate of change in stocks.Auxiliaries are stock-flow structures changing so fast that we model them as variables.Constants are stock-flow structures that change so slowly relative to the time boundary of the model that we model them as fixed.Feedback loops must include at least one stock that accumulates over a delay to avoid and accumulations represented in equations through feedback, creating the changing behavior of a system over time.The focus on equations and continuous time connects calculus philosophy to system dynamics practice.Fifth, the analysis focuses on feedback dynamics.The calculus equations integrated over continuous time create dynamics reducible to the causal feedback loops that generated them.This way,

Table 2
displays the statistical means and ranges of base runs numbered contagion incidents.Although any incident can broadcast a template ideology or employ a template method, only when they occur together is that counted as a contagion incident.

TA B L E 3
Single policy examples.Reduction requirements by parameter to adjust behavior mode.

Table 3
, we list a variety of popular policies grouped by the policy type they fall within, what parameters in our model are changed, and the real-world examples of these policies.Our simulation allows us to test multiple different policies using one aggregate parameter.For example, "Pathway to Violence Success Rate" is the probability that a perpetrator completes all preparations and begins the incident.Therefore, anything preventing the perpetrator from ultimately conducting an incident, including counter terrorism investigations, threat assessment interventions, and red flag laws removing access to firearms before the incident, reduces this probability.
Realistic constraints on policies.
a small margin within which to contain it and how these challenges persist over time.Containing the contagion means overpowering the positive feedback loops and depleting the cultural script reservoir that serves to reinfect a high-risk population years later.For example, antiterrorism gun-control measures designed to reduce fatalities have historically generated mixed results beneath TA B L E 5

TA B L E 6
Parameter values for focused deterrence (FD) test.Number of contagion incidents of focused deterrence (FD) intervention after 1000 permutations.
F I G U R E 6Policy results of focused deterrence (FD) policy intervention including backlash scenario.The results in Table7(#1) demonstrate that a focused deterrence effort can reduce the mean number of contagion incidents over 1000 permutations shifting CONT mode towards an area where CONT and

TA B L E 1 1
Number of contagion incidents of contagion containment interventions after 1000 permutations.A " < -" indicates a weakening of contagion from its initial baseline behavior mode.
F I G U R E 1 1 Contagion containment intervention high-risk broadcast memory.
Number of contagion incidents of combined policy in worst case scenario after 1000 permutations.A " < -" indicates a weakening of a contagion from its initial baseline behavior mode.

Table 13 .
Note that the < -symbol indicates the weakening of a violent behavior mode from the baseline of the policy intervention.The - average contagion incidents of scenario #3, where a nonstate actor operating from a safe haven bolsters an S2G.Also, policies #7-#10 begin with a stronger baseline behavior of a strong contagion (CONT+) than the others.Focused deterrence (#1) reduces the mean of contagion incidents in a (CONT) by 68%.Still, its effectiveness is cut in half if it provokes a backlash (#2).A nonstate actor operating from a safe haven can boost the contagion incidents of an S2G by over five-fold and shift it to a CONT (#3).Moreover, interventions at that point are ineffective.Either barely reducing the contagion incidents (#4 and #6) or making matters worse if a backlash is provoked (#5).Contagion containment (#7) is the most effective, reducing a strong contagion (CONT+) by 98% to one that fails to grow (F2G).Even under a worst-case scenario, with both rising tide and an active nonstate actor in the safe haven , Summary policy portfolio results.A "-> " indicates a strengthening of a contagion; while a " < -" indicates a weakening of a contagion from its baseline behavior mode.
34rgeting this industrial complex, however, poses key challenges.Unlike a physical infrastructure, an underlying societal grievance sustains violent radicalization.Radicalizing cultural scripts creates a system memory like a viral reservoir from which it can TA B L E 1 3ities between 2006 and 2022 in the United States, of which 66% of those fatalities are attributable to five suspected contagions.34Second,we suspect a positive feedback loop drives this effect in a generic structure not tied to a specific region, culture, religion, or ideology.If we can cut the loop, we can dramatically reduce violence by leveraging compounding feedback in the system to work with us rather than against us.Cutting those feedback loops does not mean starting from scratch, as the methods to do this already exist.Werther celebrity suicide condetect and react to cultural scripts via algorithms at scale.