Home

Dr. Rita Bush, Intelligence Advanced Research Projects Agency, Reynard Proposers Days, IARPA-BAA-09-05 Overview , April 19, 2009. Unclassified.

Na

National Security Archive

May 23, 202620 min read

A 2009 IARPA briefing unveiled a bold plan to turn MMOGs into a new intelligence source, mapping avatars to real‑world demographics.

Source: Dr. Rita Bush, Intelligence Advanced Research Projects Agency, Reynard Proposers Days, IARPA-BAA-09-05 Overview , April 19, 2009. Unclassified. Date: Apr 19, 2009 Archive: Intelligence Advanced Research Projects Activity.


Editorial Analysis

Original analysis by the DriftSeas editorial desk. The complete primary-source document, transcribed from the National Security Archive scan, appears in full below.

The Birth of a Digital‑Intelligence Initiative

On April 19, 2009, Dr. Rita Bush, then director of the Intelligence Advanced Research Projects Activity (IARPA), briefed a select group of academic and industry researchers at a so‑called “Proposers Day” for the newly announced Reynard program. The slide deck, now declassified, was not a solicitation for proposals but a roadmap for a research agenda that sought to mine Massive Multiplayer Online Games (MMOGs) and other virtual worlds for clues about the real‑world (RW) characteristics of their users. The timing is crucial: the United States had just completed its 2008‑09 intelligence review, and the Office of the Director of National Intelligence (ODNI) was eager to explore unconventional data sources that could augment traditional human‑intelligence (HUMINT) and signals‑intelligence (SIGINT) streams.

The document emerged amid a broader push within the intelligence community to harness “big data” before the term became mainstream. Earlier that year, the Director of National Intelligence had asked IARPA to identify high‑risk, high‑payoff scientific problems that could give the United States a strategic edge. Reynard was the answer to a specific hypothesis: that patterns of behavior in persistent, avatar‑driven environments reflect underlying demographic, cultural, and even ideological traits of the players. The deck’s opening slide, a graph of the exponential growth of virtual worlds from the mid‑1990s onward, set the stage for a program that treated these platforms as a new kind of surveillance substrate.

From Hypothesis to Operational Concept

The Reynard hypothesis, as laid out in the presentation, rested on a simple premise: “Real World characteristics are reflected in Virtual World behavior.” The slides listed concrete examples—Russian‑flagged avatars in EVE Online being 90 % Russian in reality, or English‑speaking chat patterns correlating with users over 35 and possessing at least a high‑school education. These anecdotal data points were not presented as proof but as a justification for a systematic, quantitative research effort.

What the deck reveals between the lines is a clear intent to move beyond qualitative sociological studies that had dominated early virtual‑world research. The program called for “behavioral indicators” that could be measured at scale, such as guild size in World of Warcraft or in‑world economic transactions in Second Life. The inclusion of metrics like “customer spending” and “exchange rates” hints at an ambition to develop predictive models of economic status, perhaps to flag individuals of interest for further monitoring.

The list of RW attributes—gender, age, economic tier, education, occupation, ideology, geographic location, native language, culture—reads like a checklist for a modern demographic profiling system. Notably, the program encouraged researchers to propose additional attributes, indicating an openness to exploratory analytics that could uncover unforeseen correlations.

Actors, Ambitions, and the Limits of the Vision

Dr. Rita Bush, a career intelligence analyst with a background in both academic research and government contracting, was the public face of Reynard. Her role in the presentation—framing the program as a scientific challenge rather than a surveillance operation—was intentional. By positioning the effort as “research,” IARPA could tap into university expertise while sidestepping some of the legal and ethical scrutiny that direct intelligence collection would attract.

The audience at Proposers Day likely included computer scientists, data‑mining specialists, and behavioral psychologists, many of whom were already studying online communities. The deck’s emphasis on “eligibility” and “award information” suggests that IARPA planned to fund multiple, parallel projects, each tackling a slice of the overall question. This distributed approach mirrors the agency’s broader strategy of seeding high‑risk research across a wide academic base.

However, the document also betrays the program’s blind spots. While it acknowledges cultural variance—calling for studies of non‑U.S. players—it still frames the United States as the analytical reference point. The heuristics listed (e.g., flag icons, punctuation patterns) are simplistic and risk reinforcing stereotypes. Moreover, the slide on “Literal mapping … not always expected” hints at an awareness that virtual personas can diverge sharply from real identities, yet the program’s metrics appear to assume a high degree of correspondence.

Legacy and Contemporary Relevance

Reynard never matured into a full‑scale operational capability; the program was eventually folded into broader IARPA efforts on computational social science. Nevertheless, its DNA lives on in today’s intelligence focus on “social media mining,” “open‑source intelligence” (OSINT), and the use of machine‑learning models to infer demographic traits from online behavior. The deck presaged current debates over privacy, algorithmic bias, and the ethics of profiling users based on digital footprints.

In an era where avatars have migrated from niche MMOGs to mainstream platforms like Fortnite, Roblox, and the metaverse, the questions posed by Reynard are more pressing than ever. The declassified slide deck offers a rare glimpse into the early governmental attempt to formalize a science of virtual‑world analytics—a precursor to the data‑driven intelligence operations that now underpin national‑security decision‑making.

For historians of technology and security, the Reynard document is a touchstone: it captures a moment when the intelligence community first tried to codify the promise—and peril—of reading the real world through the pixelated lenses of virtual societies.


Page 1

UNCLASSIFIED

IARPA BE THE FUTURE

Reynard Proposers Day IARPA-BAA-09-05 Overview

Dr. Rita Bush Intelligence Advanced Research Projects Activity (IARPA) Office of the Director of National Intelligence (ODNI)

4/19/2009 UNCLASSIFIED 1

Page 2

UNCLASSIFIED

Disclaimer

This presentation is provided solely for information and planning purposes.

The Proposers’ Day Conference does not constitute a formal solicitation for proposals or proposal abstracts.

4/19/2009 UNCLASSIFIED 2

Page 3

UNCLASSIFIED

Today's Topics

  • Program Overview
  • Program Metrics and Milestones
  • Award Information
  • Eligibility Information
  • Application Review Information

4/19/2009 UNCLASSIFIED 3

Page 4

UNCLASSIFIED Reynard Program Proposers Day PROGRAM OVERVIEW 4/19/2009 UNCLASSIFIED 4

Page 5

UNCLASSIFIED

Introduction

Growth in Virtual Worlds by Year

350 300 250 200 150 100 50 0

Active VWs

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

New VWs Cumulative VWs

UNCLASSIFIED

Page 6

UNCLASSIFIED

Reynard Hypothesis

Starting from the premise that Real World (RW) characteristics are reflected in Virtual World (VW) behavior, the Reynard program seeks to determine behavioral indicators in the realm of Massive Multiplayer Online Games (MMOGs) and VWs that are predictive of RW characteristics of the users.

UNCLASSIFIED 6

Page 7

UNCLASSIFIED

Behavioral Indicators

  • Quantitative statements regarding RW characteristics of the user, based on VW cues in a given context

Context Quantitative Assessment In EVE® Online, across all servers, 90% of the users who display a Russian flag in their user profile are RW Russian VW Cue RW Characteristic

Example:

  • In Second Life™ , Just Leap In™, World of Warcraft™ and Warhammer™, across all servers, 80% of users who chat in punctuated, English sentences are 35 years or older and have at least a high-school education

4/19/2009 UNCLASSIFIED 7

Page 8

UNCLASSIFIED Reynard Virtual World Definition

  • “Virtual Worlds” or VW will be used to refer to both MMOGs and VWs
  • MMOGs
    • typically quest-oriented
  • VWs
    • primarily social

UNCLASSIFIED

Page 9

UNCLASSIFIED Reynard Virtual World Definition Has the following characteristics:

  • graphical landscape
  • avatars
  • persistent
  • massive
  • goals Not constrained by client hardware
  • PCs, consoles, mobile phones
  • Enables, or inhibits, expression of user behaviors UNCLASSIFIED
Page 10

UNCLASSIFIED Related Research Recent research indicates that RW behaviors do emerge in VWs (e.g., the norms for politeness) Much published research is qualitative Peer-reviewed, quantitative research just emerging Science Friday Di Rita Repine AFCEA Member Portis Netizen UNCLASSIFIED

Page 11

UNCLASSIFIED

Heuristics

Example heuristics:

  • Person is a RW adult or a RW child, based on complexity of their vocabulary
  • Person is a technologist or scientist in the RW, based on names of groups or guilds
  • Person is Russian, based on the national flag shown in their user profile

UNCLASSIFIED

Page 12

UNCLASSIFIED Example Research

Three example quantitative studies: ☐ Personal space use in Second Life™ ☐ World of Warcraft™ guild size ☐ Economic decision making in Second Life™

Customer Spending In-World 2/09 500,001 - 1,000,000 L$ 50,001 - 100,000 L$ 5,001 - 10,000 L$ 501 - 2,000 L$ Total Customer Spending.. 0 100000 200000 300000 400000 500000 http://secondlife.com/statistics/economy-data.php

4/19/2009 UNCLASSIFIED 12

Page 13

UNCLASSIFIED

Program Research Areas

RW Attributes of interest:

  • Gender
  • Approximate age (child/young adult/older)
  • Economic status (low/middle/upper)
  • Educational level (primary/secondary/college)
  • Occupation (management, technical, service, labor, educator, military, student)
  • Ideology or “world view” (e.g., pacifist/militancy, multilateralism/unilateralism, etc)
  • Degree of influence
  • Digital native vs digital immigrant
  • Approximate physical geographic location (e.g., North America, South America, Europe, Asia, Australia)
  • Native language (e.g., English, German, French, Chinese, Japanese, Korean, Arabic, Russian, other)
  • Culture (e.g., Western, Asian, Middle Eastern, Russian)

Researchers are encouraged to propose additional attributes

4/19/2009 UNCLASSIFIED 13

Page 14
UNCLASSIFIED

# Program Research Focus

- Literal mapping between RW and VW not always expected.
- Example: “Furries” in VW may never dress as animals in the RW!
- But: furries in VWs may have occupations in the creative arts

http://francisanderson.wordpress.com/2008/01/28/links-for-2008-01-29/

4/19/2009
UNCLASSIFIED
14
Page 15

UNCLASSIFIED Framework for Reynard Research Questions

Groups Money and Economy Group Formation and Dynamics

SCIENCE SIM

LindeX™ Market Data High/Low/Average Exchange Rates BUY L$ SELL L$

paper plane online society

Individuals Things Avatars Do Communication Avatars and Representation

MAKE FRIENDS JOIN

“Culture”

muxlim Enhancing the Muslim Lifestyle

4/19/2009 UNCLASSIFIED 15

Page 16

UNCLASSIFIED

Understanding Cultural Differences

  • Looking for study of non-U.S.-based players
  • Compare similarities & differences in motivation, usage patterns, and behaviors across multiple cultures.

www.time.com/time/photogallery/0,29307,1817271_1728550,00.html

Sydney Morning Herald www.smh.com.au/news/breaking/gamers-should-get-an-earful-of-this-gadget/2006/05/18/1147545447078.html

http://news.gotgame.com/tag/girl-gamers/

4/19/2009 UNCLASSIFIED 16

Page 17

UNCLASSIFIED

Specific Research Questions--Individuals

Avatars and Representation

  • What characteristics of an avatar may reveal something about the real-life person behind it?
  • What motivates avatar choice?
  • Can we predict when an individual might select certain characteristics over others?
  • Does avatar selection reveal something about an individual's real-life personal characteristics, attitudes, customs, groups, or culture?

Verbal Communication

  • Can we determine whether the person is a native-language speaker of the language (e.g., English) being used?
  • Is the use of language in text chat characterized by a "signature," similar to that found in other studies of authorship?
  • Can the path of "viral ideas" be traced through the chat history of the residents, as it apparently can in exclusively text-based media, which may then allow inferences about the degree of influence of the propagators of the ideas?

4/19/2009 UNCLASSIFIED 17

Page 18

UNCLASSIFIED

Specific Research Questions--Individuals

Nonverbal Communication:

  • When and why is nonverbal communication used?
  • Does the use of nonverbal communication cues in VWs vary from RW culture to RW culture?
  • What, if anything, might the use of VW nonverbal communication reveal about the RW person?

Things Avatars Do

  • To what extent might cues in activities provide indicators about the RW person?
  • Do individuals from different RW cultures choose to engage in different sorts of VW activities?

4/19/2009 UNCLASSIFIED 18

Page 19

UNCLASSIFIED

Specific Research Questions--Groups

  • Group Formation
    • What types of groups form, what are their characteristics, and what supports their continuance?
    • Does group membership reveal anything about the RW individuals who belong to them?

[Image containing text: FOOTBALL SUPERSTARS]

4/19/2009 UNCLASSIFIED 19

Page 20

UNCLASSIFIED

Specific Research Questions--Groups

Group Dynamics

  • Given the multi-national nature of some VWs, are individuals from different countries more likely to form relationships than in the RW?
  • Do individuals recognize others as from the same culture or different cultures?
  • Do VW residents establish more or less numerous relationships than in their RW relationships?
  • To what extent are VW relationships driven by RW relationships?

4/19/2009 UNCLASSIFIED 20

Page 21

UNCLASSIFIED

Specific Research Questions--Economics

☐ Economics

  • Do certain economic decisions or activities vary depending on the culture or nationality of the user?

☐ Money

  • Does handling of VW currency vary depending upon the RW culture of the user?
  • What inferences can be drawn about the RW individual based on their treatment of VW goods and currencies?

LindeX™ Market Data This page provides the most recent market data from the Linden™ Dollar Exchange. High/Low/Average Exchange Rates L$ / US$1.00 300 200 29 Sep 05 02 Mar 09 Volume L$ 60,000,000 0 29 Sep 05 02 Mar 09 Show: 7 days 14 days 30 days 90 days all BUY L$ SELL L$

4/19/2009 UNCLASSIFIED 21

Page 22

UNCLASSIFIED Other Research Proposal Considerations Demonstrate knowledge of VW and its suitability for the research Non-U.S.-based players – data sources Overall demographics of the VW Demographically heterogeneous set of VWs vs single demographically homogeneous VW. 4/19/2009 UNCLASSIFIED 22

Page 23

UNCLASSIFIED Other Research Proposal Considerations Minimum of 4 RW attributes Multiple hypotheses or approaches for deriving behavioral indicators KIWI HEROES HIPIHI Whyville ekoloko LINEAGE THE CHAOTIC THRONE MapleStory 4/19/2009 UNCLASSIFIED 23

Page 24

UNCLASSIFIED Out Of Scope

  • Large investment in construction of new VWs.
  • New VW software applications.
  • VW software architectures.
  • Development of new interface mechanisms, such as hardware peripherals that are alternatives to the keyboard and mouse.
  • Research on PC, console, or mobile platform improvements for VWs.
  • Enhancing the game or VW experience through use of artificial intelligence or improved graphics.
  • Study of the impact of the VW environment on RW attitudes, beliefs, and behaviors.

group membership attitude towards proposal civil or religious leader

4/19/2009 UNCLASSIFIED 24

Page 25

UNCLASSIFIED

Acceptable Research Methodologies

  • Some example research methods include:
    • Ethnographic methods, such as participant observation or interviews, in VWs
    • Experimental studies in VWs
    • Experimental studies in a Real-World laboratory environment with a stand-alone VW(s)
    • Analysis of log data from VWs
    • Large-scale survey research
    • Case Studies
  • Specify mitigation strategy for the weaknesses of chosen method
  • Validate the RW characteristics of subjects.

[Image: FSB RESEARCH CLICK HERE OFFICE HOURS EVERY WEDNESDAY 11 AM ET / 8 AM PT] taken by : Ahmad Hachem on Hyperborea (208, 85, 22) blog HUD!

4/19/2009 UNCLASSIFIED 25

Page 26

UNCLASSIFIED Other Research Method Considerations - 1 Must provide own virtual and/or real laboratory facilities Specify: Number of human subjects Number of unique characters and/or end-users Planned length of the data gathering period Quantitative research methods Qualitative methods may be used as an adjunct, to inform the direction of the quantitative research, or to help interpret the results of the quantitative research. Sample size Control for “experiment farming” 4/19/2009 UNCLASSIFIED 26

Page 27

UNCLASSIFIED

Other Research Method Considerations - 2

  • Technique(s) for safeguarding privacy
    • An important evaluation criteria in proposal review
    • Safeguard data sets from accidental release or malicious intrusions.
  • Must comply with all U.S. laws, End User License Agreements (EULA), Terms of Service (ToS), and Privacy Policies of the VW(s)
  • Must include appropriate techniques for safeguarding Personally Identifiable Information (PII), including avatar names

See the section on Civil Liberties and Privacy Protection Measures for more details

4/19/2009 UNCLASSIFIED 27

Page 28

UNCLASSIFIED Other Research Method Considerations - 3

  • IARPA will not receive any raw data
  • National Institute of Standards and Technology (NIST) is Reynard’s Validation Team.
    • Data set access may be provided to NIST as needed
  • NIST will retain data access no longer than 12 months after the conclusion of Period of Performance

4/19/2009 UNCLASSIFIED 28

Page 29

UNCLASSIFIED Civil Liberties and Privacy Protection Measures

  • Must address the following (as applicable):
    • Comply with federal policy for protection of human subjects in research
    • Receive approval from their Institutional Review Board
    • Obtain informed consent from subjects of active experimentation
    • Employ techniques to protect privacy and confidentiality
  • Compliance with the End User License Agreement (EULA), Terms of Service (ToS), and Code of Conduct (CoC)
  • Annual review by Reynard PM and ODNI Civil Liberties and Privacy Office

4/19/2009 UNCLASSIFIED 29

Page 30

UNCLASSIFIED Questions? 4/19/2009 UNCLASSIFIED 30

Page 31

UNCLASSIFIED Reynard Program Proposers Day Program Metrics & Milestones 4/19/2009 UNCLASSIFIED 31

Page 32

UNCLASSIFIED

Behavioral Indicators, again

  • Quantitative statements regarding RW characteristics of the user, based on VW cues in a given context

Context Quantitative Assessment In EVE® Online, across all servers, 90% of the users who display a Russian flag in their user profile are RW Russian VW Cue RW Characteristic

Example:

  • In Second Life™ , Just Leap In™, World of Warcraft™ and Warhammer™, across all servers, 80% of users who chat in punctuated, English sentences are 35 years or older and have at least a high-school education

4/19/2009 UNCLASSIFIED 32

Page 33
UNCLASSIFIED

# Research Program Metrics

* 3 years: 2 Phases of 18 months each
* Phase 1 and Phase 2 Metrics:

| RW Characteristic | Target Minimum Probability Value: Phase 1 | Target Minimum Probabillity Value: Phase 2 |
| :--- | :---: | :---: |
| Gender | 0.75 | 0.90 |
| Approximate Age (minor/young adult/older) | 0.70 | 0.85 |
| All other RW Characteristics | 0.60 | 0.75 |

Performers should also report on the confidence limits associated with the estimated probabilities

Proposers should suggest additional specific metrics that will enable the program manager to assess their progress

4/19/2009
UNCLASSIFIED
33
Page 34

UNCLASSIFIED

Reynard Research Milestones

Milestones Months After Kickoff
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 27 28 29 30 31 32 33 34 35 36
Phase 1
Phase 2
Phase 1 Kickoff
Site Visit #1
Pilot Test by Performer
Site Visit #2 Assess Progress
PI Meeting w/ Final BIs Reported
Phase 1 Final Report
Phase 2 Kickoff
Site Visit #3
Pilot Test by Performer
Site Visit #4 /w Prelim BIs Reported
PI Meeting w/ Final BIs Reported
Phase 2 Final Reports

The Reynard Program is envisioned to begin in the fourth quarter of Fiscal Year 2009 and end by 2012

4/19/2009 UNCLASSIFIED 34

Page 35

UNCLASSIFIED Questions? 4/19/2009 UNCLASSIFIED 35

Page 36

UNCLASSIFIED Reynard Program Proposers Day AWARD INFORMATION 4/19/2009 UNCLASSIFIED 36

Page 37

UNCLASSIFIED

Award Plan

  • 3-year Program starting FY4Q2009
    • Phase 1 - Base Period - 18 months
    • Phase 2 - Option Period - 18 months
  • Criteria for moving to Phase 2: sufficient progress in achieving Phase 1 metrics
  • Multiple awards anticipated, depending upon
    • quality of the proposals received
    • availability of funds

4/19/2009 UNCLASSIFIED 37

Page 38

UNCLASSIFIED Reynard Program Proposers Day ELIGIBILITY INFORMATION 4/19/2009 UNCLASSIFIED 38

Page 39

UNCLASSIFIED

Eligible Applicants

  • Collaborative efforts/teaming strongly encouraged
    • Content, communications, networking, and team formation - responsibility of proposers
  • Foreign organizations and/or individuals may participate
    • Must comply with Non-Disclosure Agreements, Security Regulations, Export Control Laws, etc, as appropriate

4/19/2009 UNCLASSIFIED 39

Page 40

UNCLASSIFIED

Ineligible Organizations

☐ Other Government Agencies, Federally Funded Research and Development Centers (FFRDCs), University Affiliated Research Centers (UARCs), and any organizations that have a special relationship with the Government, including access to privileged and/or proprietary information, or access to Government equipment or real property, are not eligible to submit proposals under this BAA or participate as team members under proposals submitted by eligible entities.

4/19/2009 UNCLASSIFIED 40

Page 41

UNCLASSIFIED Reynard Program Proposers Day APPLICATION REVIEW INFORMATION 4/19/2009 UNCLASSIFIED 41

Page 42

UNCLASSIFIED

Evaluation Criteria

Evaluation criteria in descending order of importance are:

  • Overall Scientific and Technical Merit
  • Effectiveness of Proposed Work Plan
  • Relevance to IARPA Mission and Reynard Program Goals
  • Relevant Experience and Expertise
  • Cost Realism

4/19/2009 UNCLASSIFIED 42

Page 43

UNCLASSIFIED

Meeting and Travel Requirements

  • Plan for 1) Program Kick-off meeting and 2) Program Review in each phase
    • in Metropolitan DC area
    • to facilitate an open exchange among all Program participants
  • Each meeting 2-3 days
  • Periodic site visits

4/19/2009 UNCLASSIFIED 43

Page 44

UNCLASSIFIED

Human Subjects

  • Must provide evidence of, or a plan for, review by an Institutional Review Board (IRB) upon final proposal submission to IARPA
  • Allot ample time to complete the approval process
  • No IARPA funding can be used towards human subjects research until ALL approvals are granted

4/19/2009 UNCLASSIFIED 44

Page 45

UNCLASSIFIED

Publication Approval

  • Pre-publication approval may be required for sensitive information
  • Any award may include a requirement for pre-pub review
  • Soft copy to:
    • the IARPA Reynard Program Manager and
    • the Contracting Officer Representative (COR)

4/19/2009 UNCLASSIFIED 45

Page 46

UNCLASSIFIED

Reporting

  • Minimum requirement: monthly technical and financial status reports
  • Final report

4/19/2009 UNCLASSIFIED 46

Page 47

UNCLASSIFIED

Point of Contact

Dr. Rita M. Bush Program Manager IARPA, Incisive Analysis Office Office of the Director of National Intelligence Intelligence Advanced Research Projects Activity Washington, DC 20511

Phone: 301-226-9100 Fax: 301-226-9137

Electronic mail: dni-iarpa-baa-09-05@ugov.gov (include IARPA-BAA-09-05 in the Subject Line)

Website: www.iarpa.gov

UNCLASSIFIED

Page 48

UNCLASSIFIED Thank You! Any Final Questions? 4/19/2009 UNCLASSIFIED 48

Page 49

Second Life™, SL™ and Lindex™ and are trademarks of Linden Labs World of Warcraft™ is a trademark of Blizzard Entertainment Just Leap in™ is a trademark of Leap In Entertainment WarHammer™ is a trademark of EA/Mythic Entertainment EVE® Online is a registered trademark of CCP hf. HIPIHI™ is a trademark of HIPIHI Co. Ltd. Kiwi Heroes™ is a trademark of Massively Me Inc. Lineage II® and The Chaotic Throne™ are trademarks of NCSoft Corp.

4/19/2009 UNCLASSIFIED 49

Page 50

4/19/2009 UNCLASSIFIED 50

Page 51

NATIONAL SECURITY ARCHIVE

National Security Archive, Suite 701, Gelman Library, The George Washington University, 2130 H Street, NW, Washington, D.C., 20037, Phone: 202/994-7000, Fax: 202/994-7005, nsarchiv@gwu.edu

Keywords

declassifiedNational Security Archive

Keep reading

More related articles from DriftSeas.