$ ⌘K
P paper

Avant & Neu — The Private Security Events Database

v1.3.0 ·Private Security and Conflict

Avant & Neu (2019) "The Private Security Events Database" — introduces PSED, the first event-based dataset on private military and security company (PMSC) involvement in Africa, Latin America, and Southeast Asia, 1990–2012. Provides the event taxonomy (event type, client type, service type), descriptive findings on industry composition and abuse allegations, and a replication of Akcinaroglu & Radziszewski (2013) showing PMSC government-event counts predict shorter civil wars and attenuate the apparent effect of PMSC competition.

constructs
20
findings
7
propositions
2
sources
2
playbooks
3
// domain
Private Security and Conflict
Reported events involving private military and security companies in international and regional English-language news sources
event (with country-year and conflict-episode aggregations) 1990-2012 (PSED v.Feb-2019); extension planned through 2016
How prevalent and what kind is reported PMSC activity across regions and over time?
Who hires PMSCs, and for what services?
When and against whom are PMSCs alleged to commit human rights abuses?
Does PMSC participation in events affect civil war duration, and through what mechanism?
// top findings
7 empirical claims
view all →
F001 strong

Reported events involving private military and security companies grew steadily from 1990 to 2012 across all three regions, consistent with the widely held view that the private security industry has expanded since the end of the Cold War.

· N=1288
F002 strong

The most frequently reported PMSC event type is routine work, followed by crime events; violent events are a smaller share overall though substantial in Africa and Southeast Asia.

· N=1288
F003 strong

PMSC clients are dominated by national governments (about 22% of events) and local commercial actors (about 26%); transnational commercial clients and rebel groups (about 9%) are less common, with substantial regional variation in client mix.

· N=1288
// abstract

Abstract

Domain: Private Security and Conflict

PMSC behavior in dated, located events and its relationship to conflict outcomes (duration, lethality, abuse allegations, governance). Distinct from contract-level PMSC analyses by examining what PMSC personnel actually did, in whose service, and with what consequences.

Temporal scope: 1990-2012 (PSED v.Feb-2019); extension planned through 2016 | Population: Reported events involving private military and security companies in international and regional English-language news sources

Key Findings

  • Reported events involving private military and security companies grew steadily from 1990 to 2012 across all three regions, consistent with the widely held view that the private security industry has expanded since the end of the Cold War. (positive, strong)
  • The most frequently reported PMSC event type is routine work, followed by crime events; violent events are a smaller share overall though substantial in Africa and Southeast Asia. (positive, strong)
  • PMSC clients are dominated by national governments (about 22% of events) and local commercial actors (about 26%); transnational commercial clients and rebel groups (about 9%) are less common, with substantial regional variation in client mix. (conditional, strong)
  • Site security is by far the most common service performed in PMSC events, followed by operational support; advice and training, intelligence, and logistical support are rarely captured, likely reflecting media-attention bias toward visible armed activity. (positive, moderate)
  • Most reported PMSC events do not involve human-rights abuse allegations, but the share of events with allegations has risen unevenly over time and varies sharply by client type — national-government and local-commercial clients are most associated with allegations, transnational-commercial clients least. (conditional, moderate)
  • Reports of PMSCs hired by governments and involved in events are associated with shorter civil wars: each one-unit increase in the count of government-hired PMSC events raises the hazard of war termination by approximately 30%. (positive, strong)
  • When the count of government-hired PMSC events is included alongside Akcinaroglu & Radziszewski’s PMSC competition variable, competition loses statistical significance while events remain significant — challenging the claim that market competition among PMSCs is the operative mechanism shortening conflicts. (null, moderate)

Theoretical Propositions

  • [+] Greater participation by government-hired PMSCs in reported events is associated with shorter civil wars, plausibly because event participation proxies the resources a government is willing and able to commit to the conflict — independent of whether multiple PMSCs are competing for the contract.
  • [∅] Akcinaroglu & Radziszewski (2013)’s claim that competition among government-hired PMSCs shortens civil wars is not robust to controlling for the count of events those PMSCs participated in: once event participation is in the model, competition loses statistical significance, suggesting competition is not the operative mechanism.
// tags
PMSC private-military-security event-data civil-war conflict africa latin-america southeast-asia
// analytical
Playbooks
B  prepare_ar2013 15 steps
Derives conflict-episode-level construct observations from the Akcinaroglu & Radziszewski (2013) COW conflict-year panel. These observations feed the quick_start Cox-PH replication of Avant & Neu's Table 2 Models 1–3 (pmsc_competition, civil_war_duration, conflict_terminated, the 9 control variables, and the cross-PAX-joined pmsc_government_event_count and pmsc_rebel_event_count). CITATION: Akcinaroglu, Seden, and Elizabeth Radziszewski. 2013. "Private Military Companies, Opportunities, and Termination of Civil Wars in Africa." Journal of Conflict Resolution 57(5): 795-821. DOI: 10.1177/0022002712449325 ============================================================ PREREQUISITE — REGISTER THE AR2013 DATASET MANUALLY FIRST ============================================================ This playbook does NOT include a register_dataset step because the AR2013 supplement is hosted on Sage Journals behind a Cloudflare bot challenge that no automated fetch can clear, and the file cannot be redistributed inside this PAX. The dataset must be registered by a human-in-the-loop process before this playbook is run. Steps for the future replicator: 1. Download the supplement zip from Sage in a browser (Cloudflare requires a real browser session): https://journals.sagepub.com/doi/suppl/10.1177/0022002712449325/suppl_file/10_1177_0022002712449325.zip 2. Unzip and locate `ar2013_cow_conflict_year.csv` (the COW Correlates of War conflict-year panel used in Avant & Neu's Table 2 Models 1–3). Note: the ACD-panel file in the supplement is for the original AR2013 Models 3–4 and is NOT used in this replication. 3. Register the CSV with praxis using the absolute path on your machine — example via the praxis MCP tool: praxis_register_dataset( pax_name="private-security-events-database", dataset_id="ar2013_cow_conflict_year", source="/absolute/path/to/ar2013_cow_conflict_year.csv", format="csv", unit_of_analysis="conflict-episode-year", ) Or via Python: import praxis and call the equivalent function. 4. Run this playbook. All derive steps target the `ar2013_cow_conflict_year` dataset registered in step 3. When praxis ships per-playbook `inputs:` (praxis#399) and tilde expansion (praxis#404), this playbook can include a parameterized register_dataset step and fold the manual step back into the playbook. Until then, the manual register is the contract. ============================================================ Unit of analysis: conflict-episode-year (cowcode × sidea × sideb × onset year). A stable conflict_id is derived as: LOWER(CAST(cowcode AS VARCHAR)) || '_' || REPLACE(sidea,' ','_') || '_' || REPLACE(sideb,' ','_') || '_' || CAST(CAST(start AS INTEGER) AS VARCHAR) Cross-PAX dependency: the last two steps (derive_pmsc_government_event_count_by_episode and derive_pmsc_rebel_event_count_by_episode) join AR2013 conflict-episodes to PSED country-year aggregates via the infrastructure-country-codes-crosswalk PAX's COW→ISO3 mapping. Install that PAX first; otherwise the cross-PAX view names will not resolve.
B  prepare_psed 5 steps
Fetches the PSED Dataverse file (DOI 10.7910/DVN/ZKKOK7, file 3365449), registers it as a DuckDB table, and derives the country-year construct observations the quick_start playbook consumes. Source rows: 1,291 events, 70 countries, 1990–2012 (498 non-empty country-year cells). File is tab-delimited ASCII (despite the legacy claim of comma- delimited). ISO3 country codes are lowercased on derive to match praxis entity_id casing.
B  quick_start 3 steps
Three-step replication of Avant & Neu's reanalysis of Akcinaroglu & Radziszewski (2013) using PSED event counts on African civil wars 1990–2008. Step 1: AR2013's original Cox model with PMSC competition variables (Table 2, Model 1). Step 2: PSED substitution — replace competition with PSED government and rebel event counts (Table 2, Model 2). Step 3: Both — show the competition coefficient attenuates while event counts remain significant (Table 2, Model 3). Run the `prepare_psed` playbook first — it fetches the PSED CSV from Harvard Dataverse (doi:10.7910/DVN/ZKKOK7) and derives the country-year construct observations consumed below. The AR2013 / UCDP-PRIO panel (pmsc_competition, civil_war_duration, conflict_episode, plus the standard covariates) is genuinely user- supplied — outside Avant & Neu's scope.
// registry meta
domainPrivate Security and Conflict
levelevent (with country-year and conflict-episode aggregations)
populationReported events involving private military and security companies in international and regional English-language news sources
pax typepaper
version1.3.0
published byPraxis Agent
archive20.7 KB
// research questions
  • How prevalent and what kind is reported PMSC activity across regions and over time?
  • Who hires PMSCs, and for what services?
  • When and against whom are PMSCs alleged to commit human rights abuses?
  • Does PMSC participation in events affect civil war duration, and through what mechanism?
// constructs.yaml
20 variables in the pax vocabulary
Each construct names a thing the field measures, with a kind and an authoritative definition.
C pmsc_event_count
quantifiable
PMSC Event Count
Count of discrete, dated, located events in which a PMSC was reported to be involved within a given unit (country-year, conflict-episode, region-year). PSED's primary unit of observation. Each event must satisfy two coding rules: (1) at least one PMSC is named or clearly identifiable in the source, and (2) the PMSC is involved in one of the seven event types (work, demonstration, riot, strike, violence, crime, plot).
aliases: PMSC events, PMSC incident count, PSED events
C pmsc_event_type
concept
PMSC Event Type
Categorical classification of the kind of occurrence in which a PMSC was involved. Four categories (demonstration, riot, strike, violence) inherit from SCAD (Salehyan et al. 2012) with PSED-added modifiers; three categories (routine work, crime, plot) are PSED additions to capture non-contentious and non-violent PMSC activity.
aliases: event category, incident type
C pmsc_client_type
concept
PMSC Client Type
Categorical classification of the entity that hired or compensated the PMSC involved in an event. An event may have multiple client codings if multiple PMSCs with distinct clients were involved.
aliases: PMSC employer, contractor client
C pmsc_service_type
concept
PMSC Service Type
Categorical classification of the service the PMSC performed during the event. Categories follow Avant 2005's taxonomy and span external (military-adjacent) and internal (security) support. The service observed may differ from what the PMSC was contracted for.
aliases: contracted service, PMSC function
C pmsc_abuse_allegation
concept
PMSC Abuse Allegation
Whether an event includes a public allegation that a PMSC committed a human-rights abuse, and if so, the type of abuse alleged. Allegations are not the same as substantiated abuse; PSED records what is reported, not what is judicially proven.
aliases: human rights allegation, PMSC abuse
C pmsc_competition
quantifiable
PMSC Competition
Count of distinct PMSCs hired by the government or by rebels during a given conflict episode. Akcinaroglu & Radziszewski (2013) introduced this as a measure of market competition, theorizing that more PMSCs operating in a conflict implies reputational competition and faster goal accomplishment.
aliases: PMSC market competition, Akcinaroglu-Radziszewski competition, AR2013 competition
C pmsc_government_event_count
quantifiable
Government-Hired PMSC Event Count
Count of PSED events with a government client (local, national, or foreign), aggregated to the conflict-episode level. The replication's substitute for Akcinaroglu & Radziszewski's competition variable on the government side.
aliases: PMSC events government, Events PMCs (government)
C pmsc_rebel_event_count
quantifiable
Rebel-Hired PMSC Event Count
Count of PSED events with a rebel client, aggregated to the country-year (or conflict-episode) level. Mirrors `pmsc_government_event_count` on the rebel side; together the pair operationalizes Avant & Neu's PSED substitution for AR2013's competition variable.
aliases: PMSC events rebel, Events PMCs (rebel)
C civil_war_duration
outcome
Civil War Duration
Time from civil war onset to termination, modeled in a Cox proportional-hazards framework where the hazard is the conditional probability of war termination given survival to time t. Used in the replication of Akcinaroglu & Radziszewski (2013) on African civil wars 1990–2008.
aliases: war termination time, conflict duration
C conflict_episode
concept
Conflict Episode
A bounded internal armed conflict between a government and one or more organized armed opposition groups, with a defined onset year and (eventual) termination year. The unit of analysis at which PMSC competition and government-event counts are aggregated for the duration replication.
aliases: civil conflict, intra-state war
C gdp_per_capita
quantifiable
GDP per Capita (logged)
Natural log of GDP per capita for the country in a given conflict-year, used as a control for economic development in the AR2013 Cox proportional-hazards model of civil war duration. Wealthier states are expected to have greater capacity to terminate conflicts quickly.
aliases: GDP/cap, log GDP, lngdp
C ethnic_fractionalization
quantifiable
Ethnic Fractionalization (logged)
Natural log of the ethno-linguistic fractionalization (ELF) index, measuring the probability that two randomly selected individuals in a country belong to different ethnic groups. Higher values indicate greater ethnic diversity; included as a control in the AR2013 duration model.
aliases: ELF index, ethnic diversity, lnethfrac
C ethnic_wars
quantifiable
Ethnic War Indicator
Binary indicator equal to 1 if the civil conflict has a predominantly ethnic character, 0 otherwise. Included in the AR2013 model as a control because ethnic conflicts may follow different termination dynamics than non-ethnic ones.
aliases: ethnic conflict indicator, ethnicconflict, ethnic war dummy
C conflict_intensity
quantifiable
Conflict Intensity
Ordinal scale of battle-death intensity for a conflict-year, distinguishing minor conflicts from wars based on annual fatality thresholds. Used in AR2013 as a control for the severity of fighting, which may independently predict war termination.
aliases: battle intensity, conflict severity, intensity
C mountainous_terrain
quantifiable
Mountainous Terrain (logged)
Natural log of the percentage of a country's territory classified as mountainous, used as a control for geographic factors that facilitate rebel sanctuary and prolong civil wars. Higher values indicate more rugged terrain.
aliases: terrain ruggedness, log percent mountainous, lnmntest
C polity
quantifiable
Polity Score
Combined Polity IV score ranging from -10 (fully autocratic) to +10 (fully democratic), included in AR2013 as a control for regime type, which may affect the government's capacity and willingness to end conflicts through negotiation or force.
aliases: Polity IV score, democracy score, regime score
C proportion_of_forces
quantifiable
Proportion of Forces (logged)
Natural log of the ratio of government military personnel to rebel force size, measuring the balance of military capability between sides. AR2013 includes this as a control because lopsided force ratios may accelerate conflict termination.
aliases: force ratio, military balance, lnprop1
C support_rebels
quantifiable
External Support for Rebels
Binary indicator equal to 1 if rebels received external state support during the conflict-year, 0 otherwise. External support can prolong conflicts by replenishing rebel resources; AR2013 includes this as a control in the duration model.
aliases: rebel external support, rsup, third-party rebel support
C support_government
quantifiable
External Support for Government
Binary indicator equal to 1 if the government received external state support during the conflict-year, 0 otherwise. Government-side external support may strengthen counterinsurgency capacity and hasten war termination; included as a control in AR2013.
aliases: government external support, govsup, third-party government support
C conflict_terminated
outcome
Conflict Terminated (Failure Event)
Binary failure-event indicator for survival analysis: 1 in the conflict-year a civil war ends (peace, military victory, or low-activity threshold crossing), 0 otherwise. The dependent failure variable in the Cox proportional-hazards model of civil war duration.
aliases: conflict end, war termination event, terminate
// findings.yaml
7 empirical claims
Each finding cites a source and reports effect size, standard error, p-value, and sample size where available.
F001 strong

Reported events involving private military and security companies grew steadily from 1990 to 2012 across all three regions, consistent with the widely held view that the private security industry has expanded since the end of the Cold War.

N 1288 unit event
// method: Aggregate event counts by year of occurrence (Figure 1)
// model: Figure 1: aggregate event counts plotted by year, 1990–2012
F002 strong

The most frequently reported PMSC event type is routine work, followed by crime events; violent events are a smaller share overall though substantial in Africa and Southeast Asia.

N 1288 unit event
// method: Cross-tabulation of event count by event type and region (Figure 2)
// model: Figure 2: event-type counts by region
F003 strong

PMSC clients are dominated by national governments (about 22% of events) and local commercial actors (about 26%); transnational commercial clients and rebel groups (about 9%) are less common, with substantial regional variation in client mix.

N 1288 unit event
// method: Cross-tabulation of event count by client type and region
// model: Descriptive shares reported in narrative text and figures
F004 moderate

Site security is by far the most common service performed in PMSC events, followed by operational support; advice and training, intelligence, and logistical support are rarely captured, likely reflecting media-attention bias toward visible armed activity.

N 1288 unit event
// method: Cross-tabulation of event count by service type and region
// model: Narrative + figures
F005 moderate

Most reported PMSC events do not involve human-rights abuse allegations, but the share of events with allegations has risen unevenly over time and varies sharply by client type — national-government and local-commercial clients are most associated with allegations, transnational-commercial clients least.

N 1288 unit event
// method: Cross-tabulation of allegation indicator by client type, region, and year (Figures 3, 4, 5)
// model: Figures 3, 4, 5
F006 strong

Reports of PMSCs hired by governments and involved in events are associated with shorter civil wars: each one-unit increase in the count of government-hired PMSC events raises the hazard of war termination by approximately 30%.

hazard_ratio 1.299 SE 0.099 p 0.001 N 135 unit conflict-episode
// method: Cox proportional-hazards survival model on conflict-year panel
// model: Cox proportional-hazards on civil war duration in Africa 1990–2008; Table 2, Model 2. Controls: GDP per capita (logged), ethnic fractionalization, ethnic wars, intensity, mountainous terrain, polity, proportion of forces, support rebels, support government.
F007 moderate

When the count of government-hired PMSC events is included alongside Akcinaroglu & Radziszewski's PMSC competition variable, competition loses statistical significance while events remain significant — challenging the claim that market competition among PMSCs is the operative mechanism shortening conflicts.

hazard_ratio 1.086 SE 0.172 N 135 unit conflict-episode
// method: Cox proportional-hazards survival model with collinearity check (correlation of competition and event-count coefficients = -0.6017)
// model: Cox proportional-hazards Model 3: A&R's full specification + PSED government and rebel event counts. Table 2, Model 3.
// propositions.yaml
2 theoretical claims
Propositions are the field's reusable rules of thumb — they span findings without being tied to a single study.
P001
"Greater participation by government-hired PMSCs in reported events is associated with shorter civil wars, plausibly because event participation proxies the resources a government is willing and able to commit to the conflict — independent of whether multiple PMSCs are competing for the contract."
P002
"Akcinaroglu & Radziszewski (2013)'s claim that competition among government-hired PMSCs shortens civil wars is not robust to controlling for the count of events those PMSCs participated in: once event participation is in the model, competition loses statistical significance, suggesting competition is not the operative mechanism."
// sources.yaml
2 citations
The evidentiary backing — papers, datasets, reports — every finding can be traced to one of these.
S001
Avant, Deborah; Neu, Kara Kingma (2019). The Private Security Events Database. Journal of Conflict Resolution.
observational_longitudinal
N = 1288
S002
Akcinaroglu, Seden; Radziszewski, Elizabeth (2013). Private Military Companies, Opportunities, and Termination of Civil Wars in Africa. Journal of Conflict Resolution.
observational_longitudinal
// playbooks/
3 analytical recipes
Step-by-step recipes that wire constructs to engines. An MCP-aware agent runs them end-to-end.
B Prepare AR2013 — derive conflict-episode-year observations from the AR2013 COW panel
15 steps · 15 seconds
Derives conflict-episode-level construct observations from the Akcinaroglu & Radziszewski (2013) COW conflict-year panel. These observations feed the quick_start Cox-PH replication of Avant & Neu's Table 2 Models 1–3 (pmsc_competition, civil_war_duration, conflict_terminated, the 9 control variables, and the cross-PAX-joined pmsc_government_event_count and pmsc_rebel_event_count). CITATION: Akcinaroglu, Seden, and Elizabeth Radziszewski. 2013. "Private Military Companies, Opportunities, and Termination of Civil Wars in Africa." Journal of Conflict Resolution 57(5): 795-821. DOI: 10.1177/0022002712449325 ============================================================ PREREQUISITE — REGISTER THE AR2013 DATASET MANUALLY FIRST ============================================================ This playbook does NOT include a register_dataset step because the AR2013 supplement is hosted on Sage Journals behind a Cloudflare bot challenge that no automated fetch can clear, and the file cannot be redistributed inside this PAX. The dataset must be registered by a human-in-the-loop process before this playbook is run. Steps for the future replicator: 1. Download the supplement zip from Sage in a browser (Cloudflare requires a real browser session): https://journals.sagepub.com/doi/suppl/10.1177/0022002712449325/suppl_file/10_1177_0022002712449325.zip 2. Unzip and locate `ar2013_cow_conflict_year.csv` (the COW Correlates of War conflict-year panel used in Avant & Neu's Table 2 Models 1–3). Note: the ACD-panel file in the supplement is for the original AR2013 Models 3–4 and is NOT used in this replication. 3. Register the CSV with praxis using the absolute path on your machine — example via the praxis MCP tool: praxis_register_dataset( pax_name="private-security-events-database", dataset_id="ar2013_cow_conflict_year", source="/absolute/path/to/ar2013_cow_conflict_year.csv", format="csv", unit_of_analysis="conflict-episode-year", ) Or via Python: import praxis and call the equivalent function. 4. Run this playbook. All derive steps target the `ar2013_cow_conflict_year` dataset registered in step 3. When praxis ships per-playbook `inputs:` (praxis#399) and tilde expansion (praxis#404), this playbook can include a parameterized register_dataset step and fold the manual step back into the playbook. Until then, the manual register is the contract. ============================================================ Unit of analysis: conflict-episode-year (cowcode × sidea × sideb × onset year). A stable conflict_id is derived as: LOWER(CAST(cowcode AS VARCHAR)) || '_' || REPLACE(sidea,' ','_') || '_' || REPLACE(sideb,' ','_') || '_' || CAST(CAST(start AS INTEGER) AS VARCHAR) Cross-PAX dependency: the last two steps (derive_pmsc_government_event_count_by_episode and derive_pmsc_rebel_event_count_by_episode) join AR2013 conflict-episodes to PSED country-year aggregates via the infrastructure-country-codes-crosswalk PAX's COW→ISO3 mapping. Install that PAX first; otherwise the cross-PAX view names will not resolve.
B Prepare PSED — fetch + aggregate event-level → country-year
5 steps · 30 seconds
Fetches the PSED Dataverse file (DOI 10.7910/DVN/ZKKOK7, file 3365449), registers it as a DuckDB table, and derives the country-year construct observations the quick_start playbook consumes. Source rows: 1,291 events, 70 countries, 1990–2012 (498 non-empty country-year cells). File is tab-delimited ASCII (despite the legacy claim of comma- delimited). ISO3 country codes are lowercased on derive to match praxis entity_id casing.
B Replicate Avant & Neu (2019) — Table 2: PMSCs and Civil War Duration
3 steps · 1 minute
Three-step replication of Avant & Neu's reanalysis of Akcinaroglu & Radziszewski (2013) using PSED event counts on African civil wars 1990–2008. Step 1: AR2013's original Cox model with PMSC competition variables (Table 2, Model 1). Step 2: PSED substitution — replace competition with PSED government and rebel event counts (Table 2, Model 2). Step 3: Both — show the competition coefficient attenuates while event counts remain significant (Table 2, Model 3). Run the `prepare_psed` playbook first — it fetches the PSED CSV from Harvard Dataverse (doi:10.7910/DVN/ZKKOK7) and derives the country-year construct observations consumed below. The AR2013 / UCDP-PRIO panel (pmsc_competition, civil_war_duration, conflict_episode, plus the standard covariates) is genuinely user- supplied — outside Avant & Neu's scope.
engine.cox_ph
// playbook step bodies live in the .pax archive; download to inspect.
// relationships.yaml
3 construct edges
The pax's causal graph — which constructs are claimed to drive which others, and how strongly.
fromtokinddirectionstrength
pmsc_government_event_count →+ civil_war_duration causal positive moderate
pmsc_competition civil_war_duration correlational null weak
pmsc_event_count pmsc_abuse_allegation compositional conditional moderate
// pax.yaml manifest
name: private-security-events-database
version: 1.3.0
pax_type: paper
author: Avant, Deborah; Neu, Kara Kingma
license: CC-BY-4.0
published_by: Praxis Agent
domain: private_security_and_conflict
constructs:
  - pmsc_event_count
  - pmsc_event_type
  - pmsc_client_type
  - pmsc_service_type
  - pmsc_abuse_allegation
  - pmsc_competition
  - pmsc_government_event_count
  - pmsc_rebel_event_count
  - civil_war_duration
  - conflict_episode
  - gdp_per_capita
  - ethnic_fractionalization
  - ethnic_wars
  - conflict_intensity
  - mountainous_terrain
  - polity
  - proportion_of_forces
  - support_rebels
  - support_government
  - conflict_terminated
engines:
counts:
  constructs: 20
  findings: 7
  propositions: 2
  playbooks: 3
  sources: 2