Boot's most distinctive contribution to the study of insurgency is methodological: he built a database of 443 insurgencies since 1775 and coded each for outcome, duration, ideology type, external support, and other variables. The database is the empirical spine of Invisible Armies — it allows him to make statistical claims about insurgency patterns rather than just reasoning from selected case studies. Every major lesson in the "Implications" section is grounded in what the database shows.
This page is a reference artifact summarizing the database's headline findings. For the causal analysis behind each finding, see the individual concept pages.1
Overall insurgent success rate (1775–2013): 25.5% Overall incumbent/counterinsurgent success rate: 63.6% Draws/stalemates: 10.8%
Post-1945 insurgent success rate: 40.3% Post-1945 incumbent success rate: 50.8%
The post-1945 shift is the most significant pattern in the data. Insurgencies became substantially more likely to succeed after World War II — not because guerrilla tactics improved but because the international legitimacy environment changed (decolonization norms, UN system, international media) and because public opinion in democratic counterinsurgent states became a more decisive variable.1
Average duration of insurgency since 1775: approximately 10 years Average duration post-1945: approximately 14 years
The longer post-1945 average reflects both the more complex political environments of decolonization conflicts and the fact that external support — which became more available post-1945 — prolonged insurgencies that would otherwise have collapsed.1
Boot identifies several variables with the strongest correlation with insurgency outcomes:
Highest correlation with insurgent success:
Highest correlation with counterinsurgent success:
Variables with weaker-than-expected correlation:
| Era | Insurgent Success Rate | Notes |
|---|---|---|
| 1775–1850 | ~20% | Colonial-era; European powers at peak capacity |
| 1850–1914 | ~15% | Imperial expansion; counterinsurgent technological advantage maximized |
| 1914–1945 | ~25% | WWI/WWII resistance movements; mixed outcomes |
| 1945–1975 | ~45% | Decolonization peak; highest insurgent success era |
| 1975–2013 | ~35% | Post-decolonization; partial return to counterinsurgent advantage |
The decolonization peak (1945–1975) explains much of the romantic narrative around guerrilla warfare's effectiveness. This was a historically anomalous period when colonial powers were simultaneously militarily superior and politically unable to absorb the legitimacy costs of continued colonial rule. The lesson generalized from this era ("guerrillas always win") is historically specific.1
Boot's data does not provide precise numbers by ideology type, but his qualitative analysis identifies patterns:
Boot notes a pattern in his most recent data: the era of successful insurgencies may be entering decline. Post-Cold War, with the Soviet patron no longer available, and with democratic counterinsurgents developing better population-centric doctrine, the insurgent success rate has declined from the 1945–1975 peak.
This claim is highly contested and the most temporally sensitive in Boot's database. The Arab Spring, the Syrian civil war, ISIS's territorial state, the Taliban's return to power in Afghanistan — all post-date the 2012/2013 writing of Invisible Armies and suggest the "insurgency declining" trend was premature.
Boot is transparent about the database's limitations:
Coding decisions are contestable: What counts as an "insurgency"? What counts as "success"? The border between a successful insurgency and a civil war, between a failed insurgency and an ongoing one, involves judgment calls that other scholars might make differently.
Selection bias in historical record: Insurgencies that left significant historical records (primarily those in conflicts that Western powers fought or documented) are overrepresented. Many pre-20th-century insurgencies in Africa, Asia, and the Americas that were suppressed quickly and left minimal documentation are absent.
The 1775 start date: Boot begins in 1775 because data quality improves significantly there. The long history of pre-1775 insurgency is addressed in qualitative chapters but not in the database, which limits the statistical generalizability of pre-modern claims.1
Propaganda Model — Worthy vs. Unworthy Victims (Cross-Domain): The database's finding that international legitimacy environment drives post-1945 success rates connects to Chomsky and Herman's analysis of how media systems construct worthy vs. unworthy victims. Insurgencies that fit the dominant legitimacy narrative (anti-colonial, nationalist) get "worthy" media coverage that amplifies their external support; insurgencies that don't fit get "unworthy" coverage that doesn't. The database pattern that nationalist insurgencies succeed more often than ideological ones is partly a media effect — nationalist insurgencies are better at generating the "worthy victim" framing.
Arthashastra — State Enterprises (History): Arthashastra — State Enterprises — Kautilya's analysis of the state's economic organization as a resource mobilization system maps onto Boot's external support variable. States that can efficiently extract resources and convert them into military capacity (patron states) can fund insurgencies at scale. States with weak resource extraction capacity cannot. The database finding that external support is the highest-correlation variable with insurgent success is partly a story about state economic capacity on the patron side.
The Sharpest Implication The database's most counterintuitive finding is the weakness of the body-count correlation with outcome. Military establishments throughout history have measured success through enemy attrition — if we kill enough of them, they'll stop fighting. Boot's data shows this logic is empirically weak in counterinsurgency: the kill ratio predicts outcome less than legitimacy variables, external support, and doctrine type. If the kill metric doesn't predict success, optimizing for it is optimizing the wrong variable. The entire incentive structure of conventional military performance measurement (promotion based on operational success, measured through enemy kills and area cleared) may be systematically selecting for the wrong outcomes in COIN contexts.
Generative Questions