irregular warfare project
Post on 29-Jan-2016
52 Views
Preview:
DESCRIPTION
TRANSCRIPT
Irregular Warfare Project
MCCDC Operations Analysis Division (OAD) January 2008
Purpose
Present Status of the MCCDC OAD Irregular Warfare (IW) Project
Project Goal: Develop a prototype methodology for analyzing a USMC IW problem in-house
Agenda
• IW Modeling Challenge
• Conceptual Model
• Scenario Background
• Data Acquisition
• How IW Data is used in our Model
IW Domain
The IW Modeling Challenge
Combat Model
Weapon Pk
Armor Thickness
Vehicle Speed
Lethality
Survivability
IW Model
Influence
Susceptibility
Information Ops
Attitude
Behavior
Population Response
Killer-Victim AdjudicationMilitary OR Analyst Comfort Zone
The Challenge: Different data, different algorithms, different MOEs
IW Modeling:Expectation Management
“Soft Sciences” typically have much lower statistical correlation than “Hard Sciences”– As a practical matter, for typical data found in the
social sciences, values of r2 as low as .25 are often considered useful. For data in the physical and medical sciences, r2 values of .60 or greater are often found; in fact, in some cases, r2 values greater than .90 can be found.*
* Statistics for Business and Economics by Anderson, Sweeney, and Williams
Modeling human behavior involves a higher level of uncertaintythan modeling traditional force-on-force combat
Modeling human behavior involves a higher level of uncertaintythan modeling traditional force-on-force combat
FARC Pro-FARC Neutral Pro-GoC GoC
Conceptual Model ofCivilian Population
Insurgency Behavior Orientation
CivilianPopulation
PopulationSegments
FARC = Revolutionary Armed Forces
of Colombia
GoC = Govt of Colombia
FARC
GoC
Conceptual Model ofCivilian Population
The desire to fulfill, alleviate, or eliminate perceived needs motivates behavioral
change
Different people in the same situation will not have the same perceived needs
Psychological Operations Doctrine
Narrative Paradigm
• People are essentially storytellers
• The world is a set of stories from which each individual chooses the ones that match his or her values
• Although people claim "good" reasons for their decisions, these reasons include history, culture, & perceptions about the status and character of the other people involved
Colombia Scenario•Background• MAGTF Mission:
• Refugee Camp Security• Humanitarian Assistance / Disaster Relief
• 2 Possible Courses of Action (COAs)
• Sea-Based• Shore-BasedProvide:
Joint “Cultural” Prep of the Operational Environment
Plausible Range of Civilian Population Behaviors
SME Interviews
• Selecting SMEs– 2 SMEs obtained via MCIA– SME credentials
• Analyst & cultural SME communication challenge– Analysts need numbers, e.g.,
probabilities, percentages– Cultural SMEs are non-quantitative
thinkers
Scenario Data
• Cultural data narrowly focused on this region• Data is not accurate for the rest of Colombia
Colombia “Operation Pacific Breeze” • Background• MAGTF Mission:
• Refugee Camp Security• Humanitarian Assistance / Disaster Relief
• 2 Possible Courses of Action (COAs)
• Sea-based• Shore-based
• Goal – Civilian Population Govt Support
Cultural Data Required
Step 0: Define population segments
Elicit data for each population segment• Prevalence of current behavior patterns• Perceived needs are affected based on three
factors (using Narrative Paradigm)1. Natural tendency of the population segment
• The population segment’s narrative with respect to the insurgency
2. Effect of current events on population segment (impact)• How the population segment reacts to a given COA
3. Effect of other population segments on a population segment (influence)• How the population segment reacts to the narratives
offered by other population segments
ColombiaPopulation Segments
• Illicit Organizations• Catholic Church• Police• Military• Displaced Persons• Urban Poor• Urban Middle Class• Old Money
Cultural Behavioral Data • Orientation (Initial, Tendency)• Impact Of MAGTF COAs• Influence Of Population Segment Interactions
Orientation Data
• Initial orientation– “How do the actions of this population
segment support the insurgency (FARC) or the Government of Colombia (GoC)?”
• Natural tendency of orientation– “Given no external influences, over time, how
would the actions of this population segment change to support the FARC or the GoC?”
– Captured as data for a Markov transition matrix
Example: Urban Poor
FARC Pro-FARC Neutral Pro-GoC GoCUrban Poor 5.8% 9.2% 65.7% 15.1% 4.2%
FARC Pro-FARC Neutral Pro-GoC GoCFARC 99.4% 0.3% 0.2% 0.0% 0.0%
Pro-FARC 0.1% 98.0% 1.3% 0.5% 0.0%Neutral 0.0% 0.4% 99.1% 0.5% 0.0%
Pro-GoC 0.0% 0.0% 1.2% 98.5% 0.3%GoC 0.0% 0.0% 0.1% 0.1% 99.8%
Initial Orientation
0%
20%
40%
60%
80%
100%
CatholicChurch
DisplacedPersons
IllicitOrganizations
Military Old Money Police Urban MiddleClass
Urban Poor
FARC Pro-FARC Neutral Pro-GoC GoC
Data Elicitation
• Charles Osgood’s Semantic Differential– Osgood’s method is a development of the Likert Scale in that
Osgood adds in three major factors or dimensions of judgment:
• EVALUATIVE (good - bad) • POTENCY (strong - weak) • ACTIVITY (active - passive)
– Semantic Differential is widely used in advertising and marketing research, including questionnaires, interviews and focus groups. The versatility of uses with bipolar adjectives and the simplicity of understanding them have made it ideal for consumer questionnaires and interviews.
– There are several large scale surveys done, providing data on EPA values for over 1000 different actions, emotions and people, led by David Heise, Department of Sociology, Indiana University
Translates SME words to a quantitative measure
Rolled up to a single parameter = E * sqrt (P2+A2)
Impact of COAsElicitation
• “What words would this population segment use to describe MAGTF ‘sea-based’ operations?”– ‘Positive words’ averaged to measure leaning more
towards GoC (right)– ‘Negative words’ averaged to measure leaning more
towards FARC (left)• “What words would this population segment use
to describe MAGTF ‘shore-based’ operations?”
Word Impact
agree with 1.05
authorize 1.21
believe 1.62
please 2.93
appreciate 3.24
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
Illicit Organizations Displaced Persons Urban Poor Catholic Church
Infl
uen
ce
Lef
t
--
Infl
uen
ce
Rig
ht
Shore - left Shore - right Sea - left Sea - right
• Left means the sea/shore base COA causes the actions of the population segment to lean towards the FARC• Right means the sea/shore base COA causes the actions of the population segment to lean towards the GoC
Impact of Shore/Sea Base
Impact of Shore Base
Example: Urban Poor
Multiply
Normalize
Left Right2.11 7.22
FARC Pro-FARC Neutral Pro-GoC GoCFARC 99.4% 0.3% 0.2% 0.0% 0.0%
Pro-FARC 0.1% 98.0% 1.3% 0.5% 0.0%Neutral 0.0% 0.4% 99.1% 0.5% 0.0%
Pro-GoC 0.0% 0.0% 1.2% 98.5% 0.3%GoC 0.0% 0.0% 0.1% 0.1% 99.8%
FARC Pro-FARC Neutral Pro-GoC GoCFARC 2.10 0.03 0.02 0.00 0.00
Pro-FARC 0.00 2.07 0.10 0.04 0.00Neutral 0.00 0.01 3.87 0.03 0.00
Pro-GoC 0.00 0.00 0.03 7.11 0.02GoC 0.00 0.00 0.00 0.00 7.20
FARC Pro-FARC Neutral Pro-GoC GoCFARC 97.8% 1.2% 0.8% 0.2% 0.0%
Pro-FARC 0.1% 93.8% 4.4% 1.7% 0.0%Neutral 0.0% 0.2% 98.9% 0.9% 0.0%
Pro-GoC 0.0% 0.0% 0.4% 99.4% 0.3%GoC 0.0% 0.0% 0.0% 0.0% 99.9%
Influence Elicitation
• Influence of other population segments – “What words would this population segment
use to describe another population segment?”
Word Influence
unsuccessful -2.41
impotent -1.53
cowardly -2.50
inexperienced -2.25
mediocre -1.47
average -0.14
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
Catholic Church Displaced Persons Illicit Organizations Urban Poor
Infl
uen
ce
Catholic Church Displaced Persons Illicit Organizations Urban Poor
Influence of other Segments
Influence of “x-axis” on “legend”
Multiply
Normalize
Influence of Population Interactions
Example: Urban Poor
Left Right0.05 0.16
FARC Pro-FARC Neutral Pro-GoC GoCFARC 99.4% 0.3% 0.2% 0.0% 0.0%
Pro-FARC 0.1% 98.0% 1.3% 0.5% 0.0%Neutral 0.0% 0.4% 99.1% 0.5% 0.0%
Pro-GoC 0.0% 0.0% 1.2% 98.5% 0.3%GoC 0.0% 0.0% 0.1% 0.1% 99.8%
FARC Pro-FARC Neutral Pro-GoC GoCFARC 0.05 0.00 0.00 0.00 0.00
Pro-FARC 0.00 0.05 0.00 0.00 0.00Neutral 0.00 0.00 0.09 0.00 0.00
Pro-GoC 0.00 0.00 0.00 0.16 0.00GoC 0.00 0.00 0.00 0.00 0.16
FARC Pro-FARC Neutral Pro-GoC GoCFARC 98.0% 1.1% 0.8% 0.2% 0.0%
Pro-FARC 0.1% 94.2% 4.1% 1.5% 0.0%Neutral 0.0% 0.2% 98.9% 0.9% 0.0%
Pro-GoC 0.0% 0.0% 0.4% 99.4% 0.3%GoC 0.0% 0.0% 0.0% 0.0% 99.9%
Quo Vadis?
• Run new version of Pythagoras with data on 8 population segments
• Perform sensitivity analysis on cultural data variables (using Design of Experiments)
• Solicit feedback from
cultural SMEs on
Pythagoras results
top related