08 March 2011

Liveblogging the Quarterly NDU Strategic Wargaming Roundtable

Feel free to post your thoughts and feedback in the comments, below. Thanks!

I'll be posting live updates once the fun starts in about 20 minutes or so...

OK - about to get started. Two presentations today:
1. LTC Mike Galope, of NDU, talking about Emerald Conflict, using the GEMSTONE system that I helped design last year.
2. Ralph Chatham, with the 2d discussion on some lessons learned in game design.

Introduction to Emerald Conflict/GEMSTONE
- Game was designed to support the International Combating Terrorism Fellows (ICTF) program at NDU
---- Program exists at the master's degree level, and is targeted at senior majors thru brigadier generals and their civilian counterparts
- Game was designed as a kick-off to get them excited about the program and the lessons to be learned and the mindset shift needed to move from the tactical/operation world to the strategic/resource/policy world

- Game was not intended to be a holistic model of COIN for any predictive/planning purposes, but to engender a shift in mindset

- Scenario was built using real-world facts, such as military ORBATs, resource allocations, etc.
- Scenario built around Colombia, circa 2010, and included Colombian military officers as SMEs for both friendly/OPFOR; focused on FARC as the only real insurgency for simplification
- Original plan was 4 months/turn to allow for time for economic decisions to take effect, with 11 turns planned to get thru 1 Presidential election cycle in Colombia
- Built the teams around the geographic divisions/departments and split btw urban/rural
- Used actual gov't reports for economic baselines and performance, access to services, infrastructure, etc
- 9 Player teams: 1 Government of Colombia (GoC), 8 military divisions (both geographic/military commands)
- GcC's real-world strategy of controlling & stabilizing, built around their concept of "fusion centers" which massed resources for effects
note - I'll try to get some of the slides to extract graphics for posting later
=== brief interlude of discussion on physical layout of rooms ===

Discussion of intel play in the game
- Tech intel: expensive, ramps up faster, but has a limit to payoff
- HUMINT: less expensive turn-to-turn but takes time to develop, higher ceiling

Resource allocations:
- Target development for local services/infrastructure, economic development, military aid
- Never enough money to go around
- Economic model would feed back into future budget availability. more econ development = more dollars later

Military allocations:
- Simplified military structures: army battalions, SF companies. Army battalions were role-based (light infantry, armor, mech cav, infrastructure protection, etc).
- Some units better at certain missions than others (graphic of capabilities linked here)
- Orders worksheets included updates on unit quality and condition, to reassign/reallocate units

Red Cards:
- Unexpected events that were injected by EXCON
- not game-changers, but "complications"
- tested the communications between divisions/GoC
- Required role-play responses
- affected strategic communication plan (STRATCOM)

- completely human-in-the-loop
- manually input the values from turn to turn
- based on how well messages - actions - outcomes meshed

Adjudication Model
- Let's face it, it's a really, really big structural equation model
- Output to players:
--- access to services
--- perception of security
--- economic performance
--> overall rolled up into Popular Support for the War (key metric)

Question about "Why did players get perfect info?"
-- in part b/c the game would've been that much harder to model. other reasons to follow in a later, more detailed article

Lesson learned w/ players:
Don't separate them from the technology (don't use "pucksters" to mediate their actions)
Give them the computers and let them run with it

Ralph Chatham talking about military training with a great Star Wars-inspired intro

DoD has training problems: force turnover, technology turnover, teaching out-of-date stuff as soon as we get it into the training pipeline, trying *not* to conduct a warfare experiment

The last "training revolution" was Navy Top Gun school during Vietnam, and later became the Army's NTC
-- create an artificial war with better-than-real opponents, that tailors the war to the needs of the unit being trained
-- no-holds-barred AARs with open dialogue

Why doesn't everyone do it?
-- doesn't scale well
-- can't do it often ($$$$)

Assumptions going into DARPA game-based training project
-- Cheap, fast, effective, trainer-less, universal
Almost all wrong
-- Content is expensive
-- Need NCO supervision
-- Soldiers have varying degrees of game literacy

Some successful games:
-- America's Army: built around weeding out people who would be disillusioned if they joined the 'real' Army, and had training rules built into it to ensure conformity to Army training/behavior standards
-- DARWARS Ambush: try to teach troops how to ID Ambush, react, recover, etc; built for the soldier, not the LTC; OPFOR would be the squad next door; 6 months to deploy it; built on Operation Flashpoint w/ user authoring tools
== some war stories about the lifecycle/development of DarWars Ambush ==

One of the keys out of DARWARS Ambush is that the soldier-created content of the game allowed it proliferate throughout the force for less than a $2mil investment

== note that this is the same lesson learned from FusionNet (as opposed to CIDNE): if you give the soldier the ability to get results back out of the system, he is much more apt to put something into it. This is true whether they're inputting game scenarios, IED contact reports, key leader engagements, or maintenance statuses; it's true whether it's a game, an intel system, or a log system. You give the soldier some value out of it, then they are more likely to contribute to it ==

Green & Bavelier, 2003 (Nature): action video games change the brain and actual change the performance of players on particular psychomotor skills
--- note that this has not yet been replicated, nor did hit have a great control group

Basak & Kramer et al (2008, 2009) tried similar approach on (a) older players, (b) non-action games, with similar results, but eventually tops out somewhere

Common thread of successful games: all the successes had adequate funding
-- AA $5mil or so
-- DARWARS Ambush! $1.5-$2mil
-- Tactical Iraqi $5mil
Other less successful ones didn't

== extended discussion of lessons learned: good/bad/ugly ==

note that w/ NDU's non-attribution policy, I'm not ID'ing any speakers other than the 2 presenters, and may not capture all the questions being asked or discussion points, if specific identities can be inferred from them

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
The following charts are mine, but are relevant to one of the questions, which is about how games, sims, exercises, all interact... (click to enlarge)

note that these images are copyright me
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

Part of the challenge in the adjudication is a timeline of when/how adjudication works in order to keep the game moving as the decisions are being made in order for the adjudication to not slow down the game.

= = = = = = = = = = = = = = = = = = = = = = = = = = = =

Questions to ask about game-based learning, w/ specific objectives:
1. What do you want?
2. Why do you want it?
3. Why don't you already have it?

= = = = = = = = = = = = = = = = = = = = = = = = = = = =

Reference to The Logic of Failure by Dietrich Dorner.

By: Brant


wargamer said...

I'm not following the graphics - specifically the arrow showing the Learning-Training axis.

Why does training move into the non-competitive quadrant?

Brant said...

Training is focused on skill rehearsal, rather than skill acquisition.

When training, the focus is on process more than outcome, so the competitive aspect is downplayed...