UNLIMITED Historical Analysis of Terrorist Campaigns With observations

UNLIMITED Historical Analysis of Terrorist Campaigns With observations on current operations in Iraq Dstl/CP 10135 Presentation to Cornwallis IX, 5 - 8 April 2004 Dr Andrew Hossack Dstl Policy & Capability Studies Dept UNLIMITED © Crown Copyright 2004

Before we start…. . . Disclaimer 1 The work reported here is the product of operations research (operational analysis), not academic research: – A perfect answer a day late is worthless – Goal is the best answer in the time-and-money available – Aim is to understand underlying trends, not model details Disclaimer 2 The studies upon which this presentation is drawn were both to some extent “quick-n-dirty” OR © Dstl 2004 Dstl is part of the Ministry of Defence 2

What is “Historical Analysis”? • Historical Analysis (HA): – Is operations research of data describing actual behaviour of systems (of warfare) in a wide range of historical cases – Is empirical, stochastic and holistic – Involves the testing of quantified hypotheses using established statistical techniques – Is based on quantified primary source data from both sides of conflict where possible – Is performed by multi-disciplinary teams of analysts, historians and military experts – Focuses on understanding the enduring, underlying mechanisms of warfare © Dstl 2004 Dstl is part of the Ministry of Defence 3

What is Historical Analysis Not? • “Historical Analysis” is not the same as: – “Historical Research” - the collation and synthesis of primary source material to construct subjective estimates of objective ground truth. – “Historical Survey” - the qualitative study of individual historical events (as reported by historical research) to draw lessons, identify factors and produce descriptive statistics. © Dstl 2004 Dstl is part of the Ministry of Defence 4

Background to studies • This presentation based upon two separate studies: – A general investigation of terrorist/insurgent campaigns (2002) – A short-notice, urgent “fastball” study (2003) • 2002 Study: Part of 2 year project to inform UK defence planners on issues relating to counter-terrorism overseas • 2003 Study: No notice, 4 week study for UK military planners: – How bad were Coalition casualty rates in Iraq, compared objectively to other internal security/counter-terrorism campaigns? © Dstl 2004 Dstl is part of the Ministry of Defence 5

Definitions, Context & Scope (1) • “Terrorist” and “Terrorism” are emotive words, open to multiple (mis-)interpretations and abuse • “Terrorist” typically used to refer to groups that predominantly use terrorism tactics……e. g. : – bombings – shootings – assassinations • …but identical tactics also often present in “insurgency” campaigns as well, to lesser degree © Dstl 2004 Dstl is part of the Ministry of Defence 6

Definitions, Context & Scope (2) • UK / NATO doctrine defines “terrorism” to be: “…. . the unlawful use or threatened use of violence against individuals in an attempt to coerce or intimidate governments or societies to achieve political, religious or ideological objectives” • UK / NATO doctrine defines “insurgency” to be: “…. an organised movement aimed at the overthrow of a constituted government through the use of subversion and armed conflict” • Both terrorism and insurgency considered in these analyses © Dstl 2004 Dstl is part of the Ministry of Defence 7

Definitions, Context & Scope (3) • Scope of these studies: – is limited to historical campaigns by States & State Security Forces (SSy. F) against terrorist &/or insurgent groups (TIGs) – excludes use of terror by states against sub-populations of the state [e. g. Tonton Macoutes in Haiti] – excludes use of terror by sub-populations of the state against other sub-populations of the state [e. g. Protestant Paramilitary organisations in NI vs PIRA etc] © Dstl 2004 Dstl is part of the Ministry of Defence 8

Part I: Properties of CT Campaigns

Data Collection • Data collected on 30 post WWII terrorist/insurgent campaigns • Each data point represents an entire campaign • Information collected for both RED & BLUE on: – force sizes, leadership, objectives, organisation, levels of activity etc. • 50 data fields collected per campaign / data point • Hard to find accurate quantitative data in time allowed • Campaigns difficult to fit into one conceptual framework • RED data was difficult to find at all © Dstl 2004 Dstl is part of the Ministry of Defence 1 0

Campaigns Researched • Irgun • ANC • PIRA • Huk Rebellion • FLQ • ZANU / ZAPU • Viet Minh • Borneo Confrontation • Afghan Mujahadeen • Malayan Emergency • FARC • Mau-Mau Emergency • Aden Emergency • LTTE (“Tamil Tigers”) • Aum Shinrikyo • 26 July Movement • Black Panthers • Intifada I • Algeria FLN • Shining Path • Kashmir • EOKA Emergency • Baader-Meinhof • Algeria AIS / GIA • ETA • November 17 • Al-Gama Al-Islamiya • Viet Cong • FRETILIN • UCK Macedonia © Dstl 2004 Dstl is part of the Ministry of Defence 1 1

Terrorists/Insurgent Motivations (1) • Prevalence of various low-level TIG motivations examined: – Separatism - Religious Fundamentalism – Nationalism - Ideological Fundamentalism – Sectionalism - Single-Issue Activism – Isolationism - Millenarianism • Each historical campaign coded for (believed) primary, secondary & tertiary motivation based upon data provided on TIG origins & objectives • Frequency of occurrence examined © Dstl 2004 Dstl is part of the Ministry of Defence 1 2

Terrorists/Insurgent Motivations (2) © Dstl 2004 Dstl is part of the Ministry of Defence 1 4

Types of Violence Employed • Campaigns categorised by presence/absence of: – “terrorist” violence (bombings, shootings, assassinations etc. ) – open, armed “insurgency” – widespread, systematic civil violence (e. g. Intifada I) • With 2 exceptions, every campaign considered includes “terrorist” violence – justifies decision to include “insurgency” campaigns in scope • In addition: – 5 campaigns also contain widespread violence – 12 campaigns include open, armed insurgency © Dstl 2004 Dstl is part of the Ministry of Defence 1 5

Campaigns Outcomes (1) • The success achieved by each of SSy. F and TIG against their objectives by their own actions were coded as: – Success (S) – Compromise (C) – Failure (F) – Ongoing (O) • 19 of the 23 campaigns that have reached a conclusion were zero-sum games: – 8 ended as SSy. F successes & TIG failures – 4 ended with compromises by both sides – 7 ended with SSy. F failures & TIG successes © Dstl 2004 Dstl is part of the Ministry of Defence 1 6

Campaigns Outcomes (2) • Each campaign categorised by State Government Type: – Pluralistic, Universal Franchise (PUF): States with legal & effective mechanism for changing the highest leadership & with participation genuinely open to all members of the state – Pluralistic, Restricted Franchise (PRF): States with legal & effective mechanism for changing the highest leadership but with participation restricted to only some members of the state – Monopolistic, Limited Freedom (MLF): States without any legal & effective mechanism for changing the highest leadership of state but allowing individual freedom for members in every-day matters – Totalitarian (TOT): States without any legal & effective mechanism for changing the highest leadership and exercising considerable control over members’ every-day lives © Dstl 2004 Dstl is part of the Ministry of Defence 1 7

Campaigns Outcomes (3) • Historical sample contains no examples of terrorist/insurgent campaigns in totalitarian states - is this significant? • “Democratic” (PUF) states are, statistically, three times more likely to win campaigns than “dictatorship” (MLF) states • “Win” here means achieving a better outcome than opponent (Success vs Compromise, Compromise vs Failure etc. ) • There is also a statistically significant correlation between economic complexity of state and state wins © Dstl 2004 Dstl is part of the Ministry of Defence 1 8

Campaigns Outcomes (4) © Dstl 2004 Dstl is part of the Ministry of Defence 1 9

Campaigns Outcomes (5) • Irrespective of government type, states are significantly more likely to win purely “terrorist” campaigns than those also involving civil violence or armed insurgency • The adoption of cellular network structures by TIGs may be significantly correlated with TIG campaign wins (confidence value 88%) • There appears to be no correlation between outcome and either campaign duration, TIG network centralisation or primary motivation © Dstl 2004 Dstl is part of the Ministry of Defence 2 0

Stability of States • Half of all campaigns involving open, armed insurgency begin within 2 years of some radical change in state: – revolutionary or violent change in political system of state (coup, popular uprising, invasion, occupation, annexation by other state etc) – fundamental change in nature of system affecting sections of state (colonial independence, introduction of disenfranchising legislation) • By contrast, 80% of terrorist & civil-violence campaigns begin at least 10 years from last such relevant radical change • Possible existence of short-term “political instability” effect? © Dstl 2004 Dstl is part of the Ministry of Defence 2 1

Part II: MOEs for Iraq IS Campaign

Analysis undertaken • “Fastball” study: Limited to data already-to-hand or on internet • Campaign Period Studied: May 03 - Nov 03 • Specific comparison of Iraq Sy. F loss rates with: – The Malayan Emergency 1948 - 1960 – The EOKA Emergency in Cyprus 1955 - 1959 – US involvement in Vietnam 1961 - 1972 – “The Troubles” in Northern Ireland 1969 - 1994 • General analysis of relationships between – loss rates, force ratios & campaign outcomes – insurgent densities and campaign type © Dstl 2004 Dstl is part of the Ministry of Defence 2 3

Loss Data for Historical Campaigns • Loss rate measures used as first order surrogate for incident, contact rates: – incident, contact rates vary significantly in definition, completeness – impossible to obtain in time available • Loss, manpower measures use whole-theatre estimates – generally good surrogates for actual campaign losses, manpower – Vietnam data includes “conventional war” vs NVA • Assumes deaths are meaningfully comparable over 50 years – most modern medical evacuation/care capabilities in place by 1950 s – improvements in personal protection assumed a second order effect © Dstl 2004 Dstl is part of the Ministry of Defence 2 4

Iraq IS/CT Campaign (1) Mar 03 Start of Iraq War (20 Mar) Apr 03 Overthrow of Saddam Foundation of Coalition Provisional Authority May 03 Bush announces end of “major combat operations” (2 May) Jun 03 Serious attacks begin on Coalition forces, others 6 UK Military Police killed July 03 US acknowledges de facto IS/CT campaign accelerated build-up of Iraqi Sy. F begins Aug 03 UN HQ in Baghdad bombed Oct 03 Suicide bomb attacks on RED CROSS HQ in Baghdad Nov 03 Terrorist “Ramadan Offensive”: truck bomb kills 17 Italian paramilitary personnel 7 Spanish intelligence officers killed in ambush Dec 03 Saddam Hussein captured © Dstl 2004 Dstl is part of the Ministry of Defence 2 5

Iraq IS/CT Campaign (2) Source: Jane’s Online 2003 © Dstl 2004 Dstl is part of the Ministry of Defence 2 6

Iraq IS/CT Campaign (3) • Coalition Security Forces (Nov 03): – United States c. 130, 000 – United Kingdom c. 9, 900 – Italy, Poland – Others (13+ nations) c. 2, 400 each c. 7, 400 overall • Iraqi Security Forces (Dec 03): – – – Iraqi Civil Defense Corps (ICDC) National Iraqi Army (NIA) Police Iraqi Border Police Facilities Protection Service (FPS) • Total Coalition: c. 152, 000 © Dstl 2004 c. 11, 000 c. 68, 400 c. 12, 400 c. 53, 100 Total Iraqi: c. 145, 500 Dstl is part of the Ministry of Defence 2 7

Iraq IS/CT Campaign (4) • Strength of ‘Iraqi’ terrorists/insurgents: – Lower estimate (active insurgents only? ): “forces opposing US-led troops …total no more than 5, 000 mostly loyalists of…Saddam Hussein” Gen John Abizaid, Comd US Centcom, Nov 2003 – Upper estimate (active insurgents + passive sympathisers? ): “Top Pentagon adviser Richard Perle said. . . some 30, 000 armed supporters of Saddam Hussein's fallen regime remained active” Jordan Times, Wednesday July 23, 2003 © Dstl 2004 Dstl is part of the Ministry of Defence 2 8

Iraq IS/CT Campaign (5) • Coalition Losses: Iraq Coalition Casualty Count – Iraq Coalition Casualty Count (www. lunaville. org/warcasualties/ summary. aspx) Hostile Fatalities • Iraqi Sy. F Losses: “They’ve killed something like 86 Iraqi security forces” Donald Rumsfeld, 13 November 2003 • Iraqi civilian deaths: Non-Hostile Fatalities – Possibly 4, 200+ (www. sadnews. net) © Dstl 2004 Dstl is part of the Ministry of Defence 2 9

Comparison of Sy. F loss Rates (1) © Dstl 2004 Dstl is part of the Ministry of Defence 3 0

Comparison of Sy. F loss Rates (2) © Dstl 2004 Dstl is part of the Ministry of Defence 3 1

Comparison of Sy. F loss Rates (3) • A clear separation between NI, Cyprus, Malaya & VN • Consequence of “conventional war” element in VN data? • Iraq falls amongst NI, Cyprus, Malaya distributions: – average losses are greater than for these historical campaigns – 1/6 of loss rate for VN – all observed loss rates fall within range of loss rates for NI, Malaya – average Sy. F monthly loss rates for Iraq exceeded by: • c. 8% of NI monthly losses (1972 - 1984) • c. 30% of Malayan monthly losses (1948 - 1957) © Dstl 2004 Dstl is part of the Ministry of Defence 3 2

Breakdown of Iraq Sy. F Loss Rates (1) Highest Malayan, NI or Cyprus annual loss rate corrected for timescale: 0. 61 Sy. F deaths per 1, 000 Sy. F per Equivalent-Month © Dstl 2004 Dstl is part of the Ministry of Defence 3 3

Breakdown of Iraq Sy. F Loss Rates (2) • Greatest national monthly loss rates are: – UK (Jun 03) 0. 61 deaths / 1, 000 Sy. F – US (Nov 03) 0. 64 deaths / 1, 000 Sy. F – Italian, Spanish (Nov 03) 5. 6 deaths / 1, 000 Sy. F • Could a single, sufficiently lethal incident lead to withdrawal of one of the smaller, more peripheral Coalition forces? • NB possible basis for future analytical comparisons with, e. g. : – Bombing USMC barracks in Lebanon (Oct 83) ~ 30% fatalities – “Black Hawk Down” raid, Somalia (Oct 93) © Dstl 2004 ~ 18% fatalities Dstl is part of the Ministry of Defence 3 4

Loss Rates & Campaign Outcome (1) • At the whole-campaign level, there is a clear separation between overall loss rate and outcome: • Current estimate for Iraq ~ 3. 5 deaths / 1, 000 Man-years • But: – correlation is not the same as causality; – current estimate may not approximate whole-campaign value © Dstl 2004 Dstl is part of the Ministry of Defence 3 6

Loss Rates & Campaign Outcome (2) • Note that at least 8 of these 10 campaigns were successful in narrow military terms • Algeria, Angola & Vietnam were (arguably) lost through failures of political will or changes in political priorities • Hypothesis: Can (a perception) of high loss rates drive campaign failure by undermining political will? • NB: In democracies, there is a significant correlation between (relative) casualties and public support for ops © Dstl 2004 Dstl is part of the Ministry of Defence 3 7

Force Ratios & Campaign Outcome • As campaigns move from “Terrorist Win” to “Sy. F Win” outcomes, overall force ratios increase: • Result is significant at 90% confidence (Jonckheere Test) • Current estimate for Iraq <= ~ 9. 3 Sy. F per Insurgent • Again, not consistent with failing campaign © Dstl 2004 Dstl is part of the Ministry of Defence 3 8

Force Ratios and Loss Rates (1) y = 32 x-0. 56 R 2 = 0. 65 Iraq 2003+ (Nov 03 Estimate) © Dstl 2004 Dstl is part of the Ministry of Defence 3 9

Force Ratios and Loss Rates (2) • Significant inverse-square-root relationship between overall loss rates and force ratios (99% confidence) • Increasing force ratio increases absolute Sy. F losses but reduces loss rates • Increased “target opportunity” effect dominates any suppressive “troops-on-ground” effect? • Can we quantify the implied trade-off with force ratio between absolute fatalities & probability of campaign success? © Dstl 2004 Dstl is part of the Ministry of Defence 4 0

Insurgent Density & Campaign Type Iraq 2003 Estimate (Nov 03) ~ 1. 2 Terrorists/Sympathisers per 1, 000 Popn Iraq 2003+ © Dstl 2004 Dstl is part of the Ministry of Defence 4 1

Conclusions • Even with crude data it is possible to identify general trends and correlations in terrorist/insurgent campaigns • The campaign in Iraq is not (currently) particularly bloody compared to previous, successfully resolved IS campaigns – an order of magnitude less lethal than the US involvement in Vietnam – within the limits of violence experienced in NI and Malaya • The current state of the campaign is more consistent with eventually-successful than otherwise campaigns – not that we can predict outcome (dangers of predicting future from entrails!) © Dstl 2004 Dstl is part of the Ministry of Defence 4 2

Any Questions? Acknowledgements I would like to gratefully acknowledge the contributions made by my colleagues in Dstl to the design & execution of these analyses: Mr A Hopkin Miss CL Simmonett © Dstl 2004 Mr A Morley Mr K Sivasankaran Miss J Offord Dr D Tilley Dstl is part of the Ministry of Defence 4 3