Friendly fire, historical analysis, and why database size counts

Paul R. Syms, Dstl LBSD Fort Halstead 25 ISMOR Bishops Waltham, 2629 August 2008

Dstl/CP29251

Contents
Introduction: friendly fire on the modern battlefield Implementing the catalogue
how does historical analysis compare with other OA methods?

Some statistics from the catalogue Incident frequency on three recent operations Value and impact of HA for friendly fire studies Database size, and HA project management Questions
2
26 August 2008 Dstl 2008 Dstl is part of the Ministry of Defence

UK UNCLASSIFIED

Introduction
Friendly fire, amicicide, amicide or fratricide
the problem of attacking ones own side in battle

Historically accounted for 1020% of battle casualties Greater impact per casualty than those from enemy fire
loss of morale loss of operational effectiveness

Awareness raised by events in Iraq, 1991 onwards

solution needed, but problem was not well understood

3
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Examples of incidents
Firing at returning patrols
e.g. death of Stonewall Jackson at Chancellorsville in 1863

Bombarding own troops with artillery

many incidents, particularly in WW1

Mis-directed air attacks on own ground forces

St. L (1944), Iraq (1991, 2003), Afghanistan (2001, 2007)

Mis-identification of friends in the contact battle

all conflicts WW1, WW2, Vietnam, Falklands, Iraq

Also ground-to-air in fact, all environments

4
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Vietnam, 3-Aug-1967

5
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Need for a solution

Three potential solutions:
improve training, tactics, techniques and procedures (TTPs) improve situational awareness (SA) improve target identification using technology, e.g. a TID

Question: which is most cost-effective?

depends on frequency of events, causes, impact, environment

1990s attempts to simulate battles with fratricide

battle modelling not best suited to problem revealed lack of understanding and dearth of data
6
26 August 2008 Dstl 2008 Dstl is part of the Ministry of Defence

UK UNCLASSIFIED

Implementing the catalogue

Development of catalogue
1996: need to understand the phenomenon
idea of listing events proposed by Di Wardleworth catalogue started as an unofficial project events gathered by those interested, in DERA/Dstl and outside

First working paper published 2004

listed 1318 incidents, including 1238 post-1900

Compilation of events ongoing now around 2600 DG(S&A) sponsored archiving and validation in 200607
8
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Plusses and minuses of HA

Historical analysis is based on ground truth
albeit a partial and biased sample models and simulations are based on our imperfect understanding

Totally uncontrollable
the only analysis tools are retrospective statistics whereas a combat model is a battle in a goldfish bowl

Protracted data-gathering phase should apply to both They are complementary techniques
e.g. catalogue has informed Dstls INCIDER simulation
9
26 August 2008 Dstl 2008 Dstl is part of the Ministry of Defence

UK UNCLASSIFIED

Sources for catalogue

Problem: events scattered throughout military literature
sources have taken months or years to find

Monographs on fratricide, e.g.:

Shrader (1982), other UK and US studies Percin (1921), Kemp (1995), Garrison (1999)

General books on military history

typically yielding 2 to 6 events per book

Military obituaries and web pages Over 300 sources searched

10
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Information recorded per event

When? Where? What type of event?
environment (LL, AL etc.), actual event or a near miss etc. circumstances: day or night, in combat or not

Short factual statement of what happened Cause(s) of the incident 9 categories plus unknown Casualties (K&W) and materiel losses
and did the victim return fire on the initiator?

Full source information, including a confidence rating

11
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Some statistics from the catalogue

Statistics from catalogue 1

Occurrence of 2526 events post-1800 by decade
600

500

1130

No. of incidents

400

300

200

100

0 1800 1810 1820 1830 1840 1850 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

Decade starting ...

13
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Statistics from catalogue 2

Breakdown of 20th. century events by environment
Airland
4% 5% 1 9%

Seasea Sealand
9%

Seaair Landsea Landland Landair

9% 50% 1 % 2% 1 %

Airsea Airland Airair

Landland

32% were cross-environment events

14
26 August 2008 Dstl 2008 Dstl is part of the Ministry of Defence

UK UNCLASSIFIED

Statistics from catalogue 3

Breakdown of 20th. century events by causes
45% 40%

% incidents attributable

35% 30% 25% 20% 15% 10% 5% 0%

% C2

% Comms

% Nav

% IFF

% ID

% Mvt

% Wpns

% Delib

% Accidt

% unkn

Causes

15
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Incident frequency on recent operations

Falklands, 1982 1
UK land units involved in Operation Corporate:
20 18 16 14 No. of units 12 10 8 6 4 2 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 Days from start Infantry Armour CAS Aviation Artillery

242 unitdays

17
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Falklands, 1982 2
Land amicides on Op. Corporate including near-misses:
6

20 incidents
5

No. of incidents

San Carlos Goose Green

= 0.08/unit-day
Dismount Armour CAS Aviation Artillery

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 Days from start

18
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Iraq, 1991 1
UK land units involved in Operation Granby:
20 18 16 14 No. of units 12 10 8 6 4 2 0 1 2 3 Days from start 4 5 Infantry Armour CAS Aviation Artillery

87 unit-days

19
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Iraq, 1991 2
Land amicides on Op. Granby including near-misses:
20 18 16 14 No. of incidents 12 10 8 6 4 2 0 1 2 3 Days from start 4 5

20 UK incidents = 0.23/unit-day
Dismount Armour CAS Aviation Artillery US (all)

20
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Iraq, 2003 1
UK land units involved in Operation Telic 1:
20 18 16 14 No. of units 12 10 8 6 4 2 0 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 Days from 20 March 41 Infantry Armour CAS Aviation Artillery

637 unit-days

21
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Iraq, 2003 2
Land amicides on Op. Telic 1 including near-misses:
12

46 UK incidents
10

= 0.07/unit-day
Dismount Armour CAS Aviation Artillery US (all)

No. of incidents

0 10 13 16 19 22 25 28 31 34 37 Days from 20 March 40 1 4 7

22
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Conclusions on incident frequency

Incident rates are similar when normalized
from 0.05 to 0.25 per unit-day marked effects of operational tempo

Normalized rates very similar to Normandy, 1944

at around 0.1 incidents per unit-day

Thus more dependent on human factors than technology

reinforces validity of HA for these research purposes

23
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

What value has HA added to friendly fire studies?

Value of catalogue
Has compiled a representative sample of events
statistically robust for analysis purposes

Has provided understanding of patterns in incidents

e.g. commonly poor C3 followed by poor physical ID emphasizing multiple causes per event Reasons (2000) Swiss cheese model of civil accidents

Highlighted frequency of cross-environment events

alerted procurement to need for cross-boundary solutions

25
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Impact of project
Better understanding of event frequency and causality Informed building of INCIDER* decision-making model Input to BoI studies for CID procurement
quantifying relative importance of each environment in terms of casualties and maintaining operational effectiveness

Input to NAO and PAC reports on CID

2002, 2006 and 2007

Ongoing input to Parliamentary Questions on CID

* Integrative Combat Identification Entity Relationship model; see Dean et al. (2005)

26
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Database size, and HA project management

Database size
Can it ever be definitive?
in very limited areas, maybe; in general, no but it can provide a representative sample of events large enough to be statistically robust

Does it need to be this large? Yes projects always require a subset of the data
but which subset cannot be predicted selection can quickly reduce from 2600 to (say) 35 events

28
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Database size
Projects must select data from one I prepared earlier Quantity has a quality all of its own.
Josef Stalin

Is this requirement compatible with modern PM practice?

29
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

HA and project management

HA database compilation is a slow-burn activity
has taken 12 years to compile 2600 amicide events not particularly intensive, thus low cost per year

Issue: no overall problem owner

it is a resource that has been used by about 20 projects to date hard to assign to any one budget should HA databases be treated as facilities?

Data usually required at short notice

insufficient elapsed time for bespoke data gathering by projects
30
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

What HA databases are needed?

Potentially, one for each OA study area
with content guided by past study questions e.g. for AFV vulnerability, a database of losses, by cause etc.

They rely on analysts interested in history

should OA organizations recruit more history graduates?

OA culture must value data as much as it values models

or more so good data may feed many different models!

31
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Some sources
DAVIS P.K. & BLUMENTHAL D. (1991) The Base of Sand Problem: A White Paper on the State of Military Combat Modeling RAND Note N-3148-OSD/DARPA DEAN D.F., HYND K., MISTRY B., VINCENT A. & SYMS P.R. (2005) A new technique to address CID and IFF studies Paper for 22 ISMOR, Southampton, September 2005; Dstl/CP16723 GADSDEN J.A. & OUTTERIDGE C.C. (2006) What Value Analysis? The Historical Record of Fratricide Paper for 23 ISMOR, Southampton, August 2006; Dstl/CP21027 GARRISON W.B. (1999) Friendly Fire in the Civil War Rutledge Hill Press, Nashville TN: 229 pp. KEMP P. (1995) Friend or Foe: Friendly Fire at Sea 1939-1945 Leo Cooper, London: 198 pp. NAO (2002) Combat Identification Comptroller and Auditor Generals Report HC 661, 7 March 2002 NAO (2006) Progress in Combat Identification Comptroller and Auditor Generals Report HC 936, 3 March 2006 PAC (2007) Progress in Combat Identification House of Commons Public Accounts Committee Report HC 486, 23 April 2007 PERCIN Gn. A. (1921) Le Massacre de Notre Infanterie 1914-1918 Michel Albin, Paris: 301 pp. REASON J. (2000) Human error: models and management British Medical J. 320: 768770 SHRADER C.R. (1982) Amicicide: The Problem of Friendly Fire in Modern War US Command & General Staff College Fort Leavenworth KS Combat Studies Institute Research Survey No. 1

32
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence

Acknowledgements
The following contributed events to the catalogue:
33
26 August 2008 Dstl 2008

Nigel Beer James Bleach Ian Budd Dominic Davies Ian Dewar Julie Gadsden Bernard Garaty Ian Gardner Paul Glover Philip Hardy Chris Hobson

Fred Hood Richard Hurley Nigel Johnson Zoe Lench Alexandra Lewis Larry Lewis Geoffrey Muir Gordon Pattison Bryan Perrett Michael Phipp Bob Prescott

Kirk Ramsay Lisa Scandling John Salt * Digby Smith Mark Taylor George Tomlin Peter Trevett Geoffrey Vesy-Holt Mattias Wallen Di Wardleworth **
* Co-author of 2004 paper ** Proposed catalogue in 1996
Dstl is part of the Ministry of Defence

UK UNCLASSIFIED

Questions?
Always keen to hear of amicide events for catalogue compilation of V2.0 is ongoing please e-mail me on prsyms@dstl.gov.uk

34
26 August 2008 Dstl 2008

UK UNCLASSIFIED

Dstl is part of the Ministry of Defence