Conflict simulation, peacebuilding, and development

Comments wanted: Draft CGSC “stability operations” simulation requirements

The US Army Command and General Staff College is currently developing its ideas and requirements for a stability operations simulation that would be used in professional military education at the CGSC and elsewhere. They’re also crowd-sourcing ideas and feedback—and so they’ve asked for your help, via PAXsims. There is a summary of the challenge below, and two attached documents to look over (here and here).

The US military defines stability operations as “various military missions, tasks, and activities conducted outside the United States in coordination with other instruments of national power to maintain or reestablish a safe and secure environment, provide essential governmental services, emergency infrastructure reconstruction, and humanitarian relief.” This might involve foreign humanitarian assistance and disaster response, peace operations, counterinsurgency, or combinations of these—usually undertaken in fragile and conflict-affected states. (For more detail, have a look at the US Army field manual on the subject, FM 3-07.)

For those of you who aren’t used to the jargon of the military and the military simulation community some of the material attached below will be unfamiliar. Don’t worry about that, however—the core question here is really one of “what do simulation users need to learn about stability operations, and how might a simulation best teach them that?” Folks who work in the humanitarian and development communities, or who work on the politics and economics of fragile and conflict-affected states, may have especially valuable “outside” perspectives to offer.

If you do have comments, ideas, or suggestions, please post them here in the comments section.

* * *

Attached below is a very rough draft of requirements for a stability operations simulation intended to support staff exercises at the US Army Command and General Staff College. We’re looking for comments, and we’re interested in any simulations that might already fit these.

Overview: This document outlines required functional capabilities and training effects for a Stability Operations simulation enabling student staff exercises at echelons from battalion through brigade. There is no requirement to interface with the Live, Virtual, Constructive – Integrating Architecture (LVC-IA). At this time, the only Mission Command System that needs to be populated is Command Post of the Future (CPoF). There is no requirement to federate with other simulations. Stimulating additional Mission Command systems, federating with other simulations, and working with LVC-IA is acceptable if and only if there is no additional workload or cost associated with the capability and those systems are not required for fully capable operation. This document will assist the Material Developer to better understand the required functional capabilities and training effects to be included in the Simulation.
Description: The purpose of this Use Case is to provide requirements for a simulation to support competitive-play low-overhead educational staff-centric stability operations exercises conducted at battalion through brigade level by Professional Military Education (PME) students acting as commanders and key staff officers. This simulation is focused on Stability Operations and enables experiential educational environments. It adjudicates the results of student staff planning and decisions, requiring students to adapt their plans to an evolving situation. This is not a predictive simulation. It is intended to produce generally plausible outcomes whose dilemmas will drive student learning.

More detailed technical specifications can be found in this enclosure.

This is not a formal statement of requirements, nor is it a solicitation for bids. There will be a long and difficult road between these documents and spending money (and us getting the simulation we need). Eventually, the final version of these documents will go to the National Simulation Center (NSC). Assuming it makes it through a Requirements Board process, the NSC will turn them over to PEO-STRI, who will contract out to have it made. There are no guarantees that the process will go through all those steps.

However, we’d like to have the best thinking on this we can, in hopes of getting the best product at the far end should we get there.

A few other notes, framing what this is supposed to be:

  • As noted above, this is not a predictive simulation. It is intended to produce generally plausible outcomes whose dilemmas will drive student learning.
  • There will not be a full staff, let alone all the supporting & subordinate staffs. Thus, the simulation has to produce data directly into a student-useful format. Correlating spot reports isn’t a useful employment of student’s time; analyzing the meaning of the summary of activity reports is. This allows for a lot of abstraction in the simulation.

Last but not least, apologies for the format. Yes, there is a lot of overlap between the primary document and Enclosure 1. This may give you a window into the “wonderful” world of requirements writing, though.

James Sterrett
Deputy Chief, Simulations
Digital Leader Development Center
US Army Command and General Staff College

39 responses to “Comments wanted: Draft CGSC “stability operations” simulation requirements

  1. Rex Brynen 17/09/2012 at 6:16 pm

    Everyone: Keep the comments coming! For those who are new to the discussion, you might also want to have a look at these follow-up posts:

    Post any comments on the CGSC proposal here, though.

    Grant: Using a MMORPG-type game system to leverage the participation of the “crowd” is an interesting one. We’ve had some prior discussion of this on PAXsims, stimulated by an excellent article on the topic by David Earnest:

  2. Grant Martin 17/09/2012 at 5:40 pm

    Why not create a massive on-line game that would be interesting enough to attract a necessary number of people, a framework that mimics a COIN environment, and a tool that allows students to link into it as some kind of admin type that allows them to make policy decisions that drive certain aspects of the game and then see the results of their actions on a “crowd” scale? Imagine- and I’ve never played one of these games so this idea might be needing some tweaking- a massive on-line game with a 20’s style or drug-war style environment or Blood Diamond, etc.. The goal is to gain followers, traits, and money and … whatever drives people to play these games. The game is always going and can even be self-moderated by users so the military- outside of initial development costs and future updates- wouldn’t keep paying for much. The only connection to CGSC is that students in certain classes will have- through their instructors and initial protocols- certain privileges that will allow them to come into the game as a certain kind of “control”. They could be allowed to make tweaks based on their position- a government official, a police chief, a military commander, etc. If they are playing a staff- then they are only the military commander- and their injects will be confined to those that naturally a military commander and force of whatever force their unit will represent would have in this environment. The environment would be constantly morphing, it would be impossible to predict any of the various outcomes of any of their decisions, but one could learn how complex adaptive environments behave- which is what I’d think we’d want staffs to learn about. We might even identify flawed assumptions in our doctrine- which is what I’d think would make most simulations bad: we’ll simply build a sim based on our understanding of the world- not what it might really be.

    Grant Martin
    MAJ, US Army
    Fort Bragg, NC
    Special Warfare Center and School

  3. Mark Monday 16/09/2012 at 8:49 pm

    The link is fine with me. (Frankly, I didn’t know it was still up anywhere on the Internet.)

    If you don’t think it applies, that’s fine, but I just wanted to get my word in that any game — to be realistic — must contain elements of half a dozen types of insurgency and the player must then be able to move (actually be forced to move) from one to type to the other. The American revolution was a combination of non-violent protest, regional coup, terrorism, guerrilla warfare and civil war. Almost all insurgencies have varied elements but there is a generalized failure to understand the moving parts of the insurgency machine.

    Have a good day.

    Mark Monday
    Lead OSINT Instructor
    US Army Intelligence Center of Excellence
    Ft. Huachuca, AZ

  4. Rex Brynen 16/09/2012 at 7:57 pm

    Mark, since most of your post was a cut-and-paste from elsewhere, I’ve replaced it with a link. It also wasn’t clear how much of it pertained to the CGSC simulation requirements under discussion here.

  5. Mark Monday 16/09/2012 at 6:58 pm

    Insurgency–What’s in a Name?

    Insurgency–WhIn my view, the key to building a realistic game that teaches requires acknowledgement of the various forms insurgency takes and a way, witin the game, to move up and back on the violence level between the various forms. Insurgency, as Mao postulated, is not the end in itself but is a step to the end result. The description below may be helpful:

    What’s in a Name? An Integrated Look at Non-violence, Terrorism, Guerrilla Warfare, Revolution, Civil War and Coups

  6. Gabriel Tobias 14/09/2012 at 9:58 am

    It seems I am more than a bit late to the party!

    The previous posts here are both lively and interesting, and well beyond my acronym-comprehension level. It occurs to me that the human/AI debate goes to the core purpose of the exercise: is the goal to simulate the process of ‘stability building’, or is it to help participants find a way to a successful conclusion of that process? These two goals are not mutually exclusive, but I think there is a difference and that should guide the luddite vs. nintendo choice. If the goal is to simulate the process (and provide a chance for productive reflection), then I think the best — not to mention most cost-effective — choice is a fully human-based simulation. As the former British foreign secretary during the 2006 civil war in Brynania, I can attest to the capacity of multiple role players using email and message boards to create a very real environment of stress and uncertainty. This type of simulation allows a great degree of flexibility, and having students play roles other than the military (I would say that there definitely needs to be more than 1 Red player and several ‘Purple’ players — potentially turning Red or Blue depending on the situation) would be very helpful in giving students a better understanding of those actors.

    That being said, if the goal is to find real solutions I think a simple virtual simulation with a well-crafted incentive structure for other actors would do the trick. The incentive structure would be the trickiest part, and require some fiddling. You could have several groups of students play the game at the same time, all as military. I think the best thing would be to have a set of smaller games with specific goals (ie restoring electrical power to a groups of villages) where the outcomes are interrelated — you do a good bringing power back without pissing people off, the next time you come back to those villages things are easier (and the converse).

    Just a few thoughts. These kind of simulations can be incredibly revealing and useful, and I am glad to see the CGSC getting involved!

  7. Daniel Oakes 14/09/2012 at 3:49 am

    This article was cross-posted to a forum I enjoy, and I’m coming at this project as someone who is interested in ‘entertainment’ strategy computer modeling – and how the resulting product could influence wider adoption of process-modeling in society.

    My understanding of the tone of the article was a full on multi-actor simulation, with multiple under-the-hood varible calulations with an incremental change effect – modeling historical processes where user actions can produce ahistorical results.

    Suggested link –
    And suggested comparison game –

    But upon reading the comments section, it appears that you favor more of a closed world approach with more of pre-determined ‘action cards’ being used on a turn basis, then have the environment processing all decisions, including ‘random’ events, for the next turn.

    As such I suggest the game ‘Fate of the World’ – the game itself suffers from severe environmental idealogical blindness, but the underlying decision tree mechanics could provide an important blueprint for your modeling.

    Both programs allow for multi-month/year modeling showing how minor events can have generational effects – and each session play-through lasts only hours before an easy reset or continuation from saved scenario points.

    Paradox fans in particular already write AARs to describe their user campaigns on the company forums, so there is a pre-existing cultural mentality amongst the developer team which could translate well to to a transfer of ideas/simulation engine.

    Good luck on the project,
    D. Oakes

  8. Ronald Skip Cole 22/07/2012 at 10:19 am

    “But if this ever changing world in which we live in
    Makes you give in and cry ” — Sir Paul.

    Given how fast things can change, and how entire populations are still being transformed due to in sudden influx of modern things like cell phones, satellites, ‘banking,’ etc. Maybe it is better to take an entirely different approach.

    Have the students themselves design or help tweak the design of the simulation. Modern research shows it can be more effective than putting them into someone else’s role-play. Which, of course, is especially true if the role-play is behind the times.

    From the abstract of that paper, “With regards to this last, we describe two experiments we’ve conducted, assigning students to design and author simulations, rather than participate in them as role-players. Amongst other benefits of the design method, we found that designers showed higher levels of concept learning and motivation than did role-players.”

    As an added bonus, having the students work on the design of the simulation may well fit into modern budgets better.

  9. Graham Longley-Brown 16/07/2012 at 10:32 am

    Hi Bill,

    Like so many other topics this has all sorts of ‘shades of grey’. The distinction between training and education is inevitably blurry so I would never die in a ditch over it. They are points on the same continuum. However, it is something that we need to be cognisant of, and I’ll give some examples to illustrate this. As ever, I’m approaching this from a purely military perspective, so can’t speak for other areas.

    First, though, it is worth reiterating the point that if the likes of Gen Graham say it matters then I am inclined to believe him. Apart from being an ex-Director of our Defence Academy, he is a very bright guy. His assertion carries weight. Although I wouldn’t put myself anywhere near his league I also speak from experience: at one end of the continuum I’ve trained guys due to deploy on ops in a matter of days; at the other end I’ve delivered educational exercises introducing junior majors to working in a Combined Joint Task Force HQ when many of them couldn’t even spell CJTFHQ.

    Here are some examples where consideration (or the lack of) of the distinction between training and education made a real difference to an exercise. Like so much I write the problems appear obvious in retrospect.

    Assessing the level of pressure on players. I ‘inherited’ an educational staff college exercise where the only gauging of pressure on players was done retrospectively by assessing staff work they produced at the end of the planning process. This might suffice in a training environment. In an educational environment where the students were being taught the planning process we had to introduce regular feedback mechanisms to assess, for example, whether Excon was giving the students too much information to assimilate. If the level of pressure was too high the students learned nothing. Obvious. But we had to re-design the exercise processes to build in these feedback mechanisms.

    The Nintendo effect. As previously discussed, give people a real-time simulation and they become fixated with it, immediately diving into their comfort zone of fighting the current (tactical) battle. In a training environment learning to resist this is often part of the training. In an educational context it is often necessary to absolutely preclude the possibility by, for example, providing only time-stepped Common Operational Picture (COP) updates. This is often seen as anathema to trainers, who think that the simulation(s) must look and feel as close as possible to the real C2 systems people will use; not necessarily so in an educational environment.

    Instructor adjudication of outcomes. This same real-time versus time-stepped COP updates has a huge impact on the ability of instructors to adjudicate simulation outcomes. In many educational exercises it is essential to check the simulation outputs and, if necessary, adjust them to ensure the educational objectives are met. Having your entire air force destroyed on the ground on day 1 will teach valuable training lessons; but what about an educational exercise where one of the objectives is to learn how to produce an Air Tasking Order and manage the air-land battlespace?

    Ability to rewind. It is unlikely that a training exercise will ever be stopped then re-started at a previous point in time because the players have done badly (loss of face for starters!). On the other hand, it is often desirable to have this option available in an educational context. The design considerations (and potential simulation costs) to allow this ‘re-wind’ feature are considerable. I’ve seen it used effectively in an educational exercise where it was essential to achieving the objectives; but both the simulation and – critically – the exercise processes and timelines had to be specifically designed to allow it.

    I’m sure many people will read this and think the distinction – and the examples – insignificant. In many exercises they are. But I’ve seen too many exercises where failing to consider the differences, slight though they appear, has led to considerable problems including outright failure.

    Graham LB

    P.S. I’ll cross-post this onto simulatingwar as I think it’s maybe a little tangential to this thread but might have some mileage as a separate topic over there.

  10. Bill Haggart 14/07/2012 at 12:45 pm

    Graham wrote:,
    With regard to the difference between training and education, it does matter. Education, training and on the job experience lie on the same continuum, but the approach to each often needs to be quite different. When it comes to designing a simulation or a wargame there can be major implications. The best explanation of the distinction between education and training is a quip used by General Andrew Graham, recent Director of the UK Defence Academy, who thought the difference worth noting. Apparently light-hearted, it sums up the distinction perfectly: your 14-year old daughter comes home from school and tells you that she has been taught sex. You ask her if it was sex education or sex training…

    A lot of great discussion. Impossible to comment on it all. I did want to respond to Graham’s explanation [which I appreciate]. I have found that one of the real problems in education, specifically school and unversity education is that there IS that distinction. [ten years from now you’ll be glad you had Algebra] And lets use General Graham’s distinction. I love the military for making such issues soooo concrete. ;-7

    The imagined distinction is that sex education is cognitive, simply information, while sex training would be the physcial practice of doing ‘it’. A real distinction, no doubt. However, in practical terms in regards to teaching, what use is the ‘sex education’ information if it is of no practical *use*? In other words, the only value of the sex education is if the students DO something with the information. That ability requires context, situational awareness, and *practice* in the application of the knowledge. So while there certainly wouldn’t be any ‘sex training’ regarding the physical act, there would have to be training in a very real sense if the sex education instructors could have any expectation that the course knowledge would be used in real life.

    In that sense, there isn’t a real distinction between education and training in practical, rubber-meets-the-road terms, but rather in the specific training topics. ;-7 Certainly one of those goals would be NOT include practicing the sex act itself. Even so, the most effective approach to sex education and/or sex training should look remarkably alike in methodology, only different in content…

    I do find it interesting that military folks would make that distinction, because one of the things I have always admired about military education in general is that it most often looks and feels like training.


  11. Brant 11/07/2012 at 2:13 pm

    “On the Joint Staff we are using PSOM for planning and analysis.”
    To me, that almost shoots it down as a solution to CGSC’s problem right there.
    Training tools should be focused on training, not predictive analysis or decision support. Conflating the two because they have similar interfaces (to people who don’t know any better) will only result in HUGE problems with outcomes, and poor results on all fronts (often critiqued by people who *should* know better).

  12. Julian Ouellet 11/07/2012 at 11:58 am

    PSOM is alive and well. It was used by IJC twice in the last year for planning purposes and it is being used for training in a number of places (NPS, USPTC, etc.). It continues to evolve in terms of its capabilities and user community. On the Joint Staff we are using PSOM for planning and analysis.

    For some information on present use and development check out:
    Journal of Defense Modeling and Simulation Vol 8(2) April 2011 issue. The folks at Dstl that designed and run PSOM have a number of articles on its function and use in the issue.

    Additionally some info on one of the Afghan events can be found here:

    If you would like additional information on DOD use of PSOM for training, planning, and analysis feel free to contact me.

  13. James Sterrett 10/07/2012 at 12:38 pm

    Everyone, this has generated far more discussion than we anticipated – which is *great*! We appreciate the time you’ve spent in providing feedback.

    Please keep posting anything you think we ought to know.

    However, we didn’t plan on this, and I’m about to be out of electronic contact for two weeks. In the interim, Michael Dunn, from my team, will be monitoring this and continuing the discussion. He will also be at Connections (he won the coin toss on who got to go :) ), so the discussion could possibly continue there as well.

    Again – *thank you* for the time you’ve taken to help us, and please don’t stop.

  14. James Sterrett 10/07/2012 at 12:35 pm

    Phil, why stop there? :)

    We’d be happy to have a look at IW TWG. Does the team have people at Fort Leavenworth?

    I see your points on timescale and human interactions…. but I don’t write the POI.

  15. James Sterrett 10/07/2012 at 12:33 pm

    Paul, thanks for the explanation. I think we differ in terminology, but we’re aimed in the same direction — you are correct that I’m not looking for something terribly involved. (That is one of the lessons of all this feedback – I fear we need to provide a more precise example of the level of detail we are looking for.) We want something with more links in it than a straight choose your own adventure, with a clear causality chain, and certainly not aiming to simulate everything – processing moderately large number of chains such as “You chose to allow artillery fire in this region, which damaged the schools, so now the locals are angry, so now the insurgents have more recruits”, and doing it automatically instead of in the instructor’s head, by defining the links and then having player actions trigger the responses through the system.

    Personally, I’d call that a simulation, albeit not a hugely complex one – but I see your point that many people are going to misunderstand what we’re aiming at. Thank you!

  16. Rex Brynen 10/07/2012 at 11:56 am

    Phil: We did briefly cover the IW TWG at PAXsims based on media reports (, but we would love to have a full blogpost from you on the subject (hint, hint!). An email is on the way…

  17. Phil Haussmann 10/07/2012 at 11:47 am

    Wow! This is a an awesome thread. I’m surprised it hasn’t come up yet, but someone needs to mention the TRADOC Analysis Center’s (TRAC) Irregular Warfare Tactical Wargame (IW TWG, of which I am a developer). Here is a brief overview:

    The IW TWG is designed to provide a relevant and credible capability to effectively inform decisions concerning operations within the IW operational environment (OE). It enables decisions about the impacts of doctrine, organization, training, materiel, and leadership and education (DOTML) change – specifically organizational, doctrinal/TTP, and relevant materiel changes; enables decision on how to invest in change; and enables decisions on how to conduct operations within the IW OE. The IW TWG was designed to account for and represent the relevant relationships and prevailing interactions between ground force tactical units and local (host nation and regional) populations; the influences of tactical unit interaction on the populations; local and higher diplomatic, information, military, and economic (DIME) influences on the populations; and how population and IW OE states in turn impact tactical and operational unit operations. In its current form, the IW TWG is set up to use an Afghan scenario in Helmand province. 14 players (1 BDE commander, 1 BN commander, 4 Co commanders, 1 Afghan National Army commander, 1 Afghan National Police Commander, 1 Government of the Islamic Republic of Afghanistan leader, 2 Taliban commanders, 1 Criminal, 1 NGO, and 1 Other US Government and Provincial Reconstruction Team leader) plan for a one week turn which is adjudicated through a simulation called the Planning, Adjudication, and Visualization Environment. At the route of this simulation are things called Task, Event, Outcomes that define a set of things that can happen when players planned Tasks are adjudicated (including interactions with the populace, key leaders, and other players’ Tasks). Population responses come from a seperate, data driven model. The players interact with each other through a chat function in order to capture everything.

    As you can tell, it is supposed to be used for analysis, not training. Of course, several players over the last three years have asked for training versions. No one has created such a thing, since TRAC doesn’t do training unless asked to do so. I contend that it would not be difficult to modify it to a point where adjudication would take minutes instead of one hour. This would allow it to be run for the required 12 to 18 months of game time. Luckily, our software developer (an Army civilian, so you get the source code) has recently created a scenario generator. I believe that solves most of the problems, but I haven’t seen it in action.

    The most difficult part of a game including a large number of players is monitoring their interactions and intent. Several attempts have been made to capture player intent, including the chat function, interviews, and “player intent cards”, but as far as I can tell, in the six months of analysis that have been performed on the game since the last run, no one has actually gotten any results from these pieces of data. Possibly, they simply haven’t looked at the data. The amount of data generated automatically by the simulation is truly astounding, and the analysts at White Sands Missile Range have been spending their time on that rather than the subjective analysis required from analyzing conversations. All this leads back to Paul’s point about MOPs and MOEs. If the MOE is “a player can function on a staff during a stability operation” then this game will probably prove suitable. That said, there exist many problems with the game which I’m sure my boss (Paul Works) would probably not want me to mention here.

    Now that my selfless plug for the IW TWG is over, I want to briefly address some of the issues that have appeared on this thread regarding Stability Operations. First of all, on the issue of computer vs. tangible gaming, I lean towards the tangible gaming option in the case of SO. SO, as was mentioned earlier, is all about personal interaction. Players siting in front of a computer screen simply do not get the same experience. My part of the TWG was a board game in which four of the 14 players had the opportunity to argue. Topics of discussion included which piece of infrastructure was more important and how to get the Green player to stop corruption (I paid him real money to hoard his resources). Although these discussions also came up in the PAVE environment, none of the players were immersed enough to actually interact in a realistic manner (getting angry or disappointed).

    Second, re: timescale, although the results of Blue actions can rarely be felt in the timeframe required unless they screw something up (don’t piss on the Koran), the players are not actually going to be in country for more than 12 months on most occasions. When playing a 12 month game, we run the risk of training them to ignore the long term consequences of their actions. The Army has consistently made this mistake in Afghanistan. We give the locals generators instead of fixing the electrical grid, then the generators break after our 12 month rotation, but we don’t care since we’re out of there. We take the lead in most offensive operations (how else will we achieve our MOEs?) instead of actually letting the ANA make their own mistakes and learn to lead. I doubt I can convince the CGSC to expand their timeframe, so I propose this simulation include injects that show the consequences of previous short-sighted decisions.

    I have many, many more things to say on this topic, but I’ve said enough already.

  18. Brant 10/07/2012 at 10:14 am

    Paul – my winking comment was a bit of an inside joke aimed at Dr Sterrett, as we’ve had the same discussion going on for about 6 years now :)

  19. Paul Vebber 10/07/2012 at 10:13 am

    “Paul, I’m not sure I follow the difference between that and a simulation?”

    “Maybe it’s a game and not a sim? ;)”

    Maybe its not really either, but more like what what used to be called a “game assist program”?

    I think in the context most people are commenting on this – a simulation of stability operations “reproduces to a certain degree of abstraction all the process associated with the environment stability operations are conducted in”. The problem I and others have is that since we really don’t know much about those process (and even when we try, people joke about the result – the infamous “incomprehensible powerpoint slide”) that ‘simulating’ it to any degree is really hard, and perhaps impossible.

    In the simulations I have dealt with like EADSIM, JSAF and NSS, the “simulation engine” takes inputs, has internal models of the interactions between entities in the sim, and can chug back at least abstract results that one can follow a causal chain through the sim from input, through a model, to output, that it can be argued is at least “representative (I get into heated arguments its NOT predictive, but others maintain it is.

    You this process “causality chain” – each link of which can be subjected to some level of VV&A for a simulation to be credible. These sims can do that because they work on physical entities interacting based on laws of physics, or at least abstract representations that don’t violate physics (don’t get me started on “physics-based modelling…). The operating environment of stability operations, despite many protestations, is no the physical domain of war, but the moral and cognitive. Its not about missile locking onto targets, but convincing people to change their minds about things, or at least not to act out in what WE consider inappropriate ways because they disagree.

    How those processes work are being looked at, and a number of theories are out their, and “human terrain modelling” is getting a lot of buzz, but I look at what they appear to promise, and look at day to day family life and scratch my head and go “really?!?” If the kind of things we want are possible, then we would have invented them long ago to solve the vexing problems of day -to -day relationships right here at home…

    In any case, what this says to me is “simulating” stability ops is at best extremely expensive and most likely highly dubious not just predictively, but fundamentally “in theory” as well.

    So I read statements like:

    “when run well, keeping up with continuity in the situation is a lot of work for faculty who are supposed to be mentoring and evaluating students.”

    “A staff-centric game produces more data for more people to chew over. Both of these contain a spectrum of options, especially staff-centric.”

    “The purpose is to learn staff processes in a stability operations invironment, while having a richer dataset in response to their decisions than a small team can come up with on its own in a reasonable timeframe.”

    “but it’s supposed to let us build the broad brush of a given situation and let individual classes run with it.”

    and to me the money quote:

    “The sim is the background, providing data for the decisions. The instructor may well tweak the simulation’s results in order to set things up for some learning point. (So why not simply have the instructor provide all the answers personally? Workload! It’s been tried.)”

    Simulations of the type I deal with are as complex as they are because for a given set of inputs, the output is not know. Given all the factors in play, will the seeker lock onto the target, will the sonar detect the target? Does the new algorithm in the system do this more often than the old one, does it introduce other problems we didn’t think of?

    NOw I spent time in the training community and understand “training sims” answer different questions – but they still are “simulations” because they provide feedback to student inputs based on “real capabilities interacting” – usually with different emphasis than an analysis simu, but they are related.

    In your case it seems that you know the universe of outcomes that different student decisions should produce to achieve your education objectives. The workload of doing that manually is too high for the time you have to do the game.

    You don’t need a simulation to execute a “causality chain” – you just a faster and less manpower intensive tool to map elements from a set of plausible outcomes it seems you already have identified, to inputs the student chooses. To me this is different from a simulation – the “fix is in”! A sim has internal rules that produce results based on trying to represent HOW the output is linked to the input. You just need a “tool” that lets the instructors “drive” – rather than really on hoping the simulation gets him where he wants to go. It more like setting up one of those books where you “choose what you do at the end of the chapter” and then you go to page 34 or page 87 to carry on the narrative from there.

    You want a plausible narrative to be constructed that teaches the students a lesson, and which rewards behavior you are trying to reinforce, and punishes alternative. My mentor in the training community CDR Don Cope, used to say “training is about making sure the student knows HOW to follow procedure, education is making sure he knows WHEN to deviate from it and WHY…

    TO me to do this you want a game (a decision-centric tool) that has a computerized “instructor workload reduction tool” so you can turn the crank on adjudicating turns fast enough to get your training objectives accomplished in the time you have.

    I don’t think that is a sim, and based on my experience in the community, if you ask for a sim to do the job of that tool, you are going to WAAAAAY overspend for a host of capability you don’t want or need because a lot of effort will be sent trying to “simulate” things that you already know how to do, just want to do faster.

    I sent you the POC at OSD CAPE – probably best for you to talk directly to him, rather than me be the point of failure between you!

  20. Brant 10/07/2012 at 6:27 am

    on 09/07/2012 at 11:17 pm, James Sterrett said: “Paul, I’m not sure I follow the difference between that and a simulation?”

    Maybe it’s a game and not a sim? ;)

  21. Brant 10/07/2012 at 6:25 am

    Peck’s new article for TSJ dances close to this issue as a part of a broader issue about training:

  22. James Sterrett 09/07/2012 at 11:17 pm

    Paul, I’m not sure I follow the difference between that and a simulation?

    Regardless, I’d be happy to know more about the Oz Wargame Integration Toolkit.

  23. Paul Vebber 09/07/2012 at 4:52 pm


    It sounds what you want is not really a simulation, but a data automation tool. That is a very different thing than a simulation. A data automation tool addresses the workload issue you describe, but is not actually simulating all the myriad processes that are involved in an actual stability operation. It simply takes a set of player actions that are input, and chooses, based on a set of rules along the lines of Outlook’s “rules manager” from a set of “pre-scripted responses”. It can get as complicatd as you want, but typically in a taining environment you want the universe of results to be limited anyway.

    That is a WHOLE lot cheaper than full blown simulation, As you say, outputs need only appear plausible, and can go through an SME vetting process to weed out “game breaker’s”. It can also follow a “Campaign Tree” as suggested by Graham, with branches that can trigger based on certain player actions, or by white cell.

    An engine that can be used in this way, as well as to facilitate data capture is Oz Wargame integration Toolkit that OSD CAPE is developing. I’ve been a Beta tester and it seems to have a lot of the features you want in something that already exists. If you are interested I can send you some backround material

  24. James Sterrett 09/07/2012 at 3:49 pm

    And while I got the prior comments sorted, more rolled in! Again – thank you!

    Brant, we are not so much looking for what the students ought to be taught. So, getting a bit more specific:

    They get put into one of two scenarios. One involves a relatively developed country which they have recently rolled into and liberated (in the prior exercise, a division-level major combat operation staff exercise). During this exercise, the Future Ops cell develops the transition plan for the stability operaion, and in principle that is the order handed to the brigade in the stability operations exercise. Instructors have discretion on exactly when the stability operation starts – right after the end of hostilities or several months down the road. And the instructor can modify the initial conditions and the student order (if, for example, the FuOps cell did a terrible job, you don’t want the next exercise crippled by it; or conditions may need to change to focus on something the students are not good at handling).

    The other involves Cerasia, as part of the Eagle Owl joint exercise between CGSC and the British JSCSC. Cerasia is a NATO scenario, is completely unrelated to the prior exercise chain, and involves a largely undeveloped area. Cerasia has been largely unsupportable by the simulations we’ve looked at so far, most of which revolve around restoring essential services (security, water, etc) to places where those services have been destroyed. If there is a PSOM scenario for Cerasia, that would be very interesting news.

    The primary objective is to educate student on executing staff processes in a stability operations environment.

    They have to walk out of the door at CGSC, diploma in hand, ready to take on the role of such a staff officer.

    They may not be brilliant at stability operations… but they have to be able to work with the rest of the brigade (or other echelon) staff when they walk out.

    We do want them to understand stability operations – otherwise we wouldn’t use it as the backdrop for a staff exercise. But while the average student will have a better awareness of stability operations history, doctrine, and theory, the key to make sure that all our students can go, fit into a staff slot, and have enough background on whatever kind of operations they face that they can get OJT fast enough to succeed.

    Graham, I will file General Graham’s explanation for future use. :)

  25. James Sterrett 09/07/2012 at 3:32 pm

    Hello! Thank you, all of you, for your feedback!

    Some replies to points made, and clarifications to our intent:

    We’ve tried CELTS, Elusive Victory, UrbanSim, an older version of PSOM (c. 2008), and some homebrewed spreadsheets to run this. As it stands, it is run as a series of 1-2 page vignettes. When run well, the vignettes serve quite well. When they are not run well…. And, even when run well, keeping up with continuity in the situation is a lot of work for faculty who are supposed to be mentoring and evaluating students.

    None of the sims do everything that is needed. We are quite interested in seeing any tools that are out there, which certainly includes your sim, Rex, and a version of PSOM newer than 2008. (We’d heard PSOM had died!)

    “Staff-centric” means capable of driving a staff through an exercise, as opposed to Leader-centric. Any single-player game is, in principle, leader-centric: it produces data for one person to interpret and make decisions on. A staff-centric game produces more data for more people to chew over. Both of these contain a spectrum of options, especially staff-centric. We make a point of it in part because there are those who are telling us to the UrbanSim to drive this exercise. UrbanSim is a great leader-centric tool, when you have the scenario you need. Unfortunately, there isn’t enough detail in it to really drive a staff, and building new scenarios requires access to source code.

    How we envision the exercise – based in large measure on prior pilots:
    2 faculty, 1 or 2 white cell students, 1 student for Red, and the other 13 or 14 so students are all holding positions on the Blue staff – including battalion commanders, some might be other agencies or NGOs.
    The students are handed the initial situaiton and come up with a plan, the Red student comes up with a Red plan, briefings to instructors occur, and the plans are put into the sim – including by the White cell for any factions owned by White.
    Run a turn or several turns. Results come out of the sim. Students analyze these, and run through their staff process again.
    The purpose is to learn staff processes in a stability operations invironment, while having a richer dataset in response to their decisions than a small team can come up with on its own in a reasonable timeframe. Oddly, one of the staff processes is RDSP (Rapid Decision-making and Synchronization Process), which doesn’t suit the rest of the exercise structure; that’s an artifact of the way the course has evolved over the past 10 years.
    Rex’s points about ARs are on target; why students made a decision is key. The instructors are supposed to be capturing this, as they are present in the room. I’m not sure the instructors would have the time to review a huge database in the time between the end of the exercise at the end of on Thursday and the AAR on Friday morning – though how the instructors monitor communications in a distance learning exercise I am not certain.

    When the exercise runs, we will be looking at approximately 48 simultaneous independent exercises going on, each run by one of two faculty. We have 5 Army civilians in the Sims Division (subject to potential downard revision in budget cuts…) No two faculty run the exercise in quite the same way. “Low overhead” isn’t being snuck in there. If it isn’t easy to run then it will fail.

    Our biggest group of SMEs are our faculty – it is their learning points we need to drive, and they decide what the students have to learn. This isn’t strictly an “Ours is not to reason why” situation, but there are limits to the value of external SMEs.
    The second big group of SMEs is hired as consultants during the exercises; these are already brought in, with backgrounds in media, other agency, and/or NGOs. These people come in and discuss their perspective with the students and comment on their planning. In addition, a handful of our students are from other agencies – perhaps 2 per year in a student body of ~1500. 1 in 16 students are non-US officers, from nearly anywhere in the world.

    Several of you brought up timescales; the declared timescale by the faculty is around 12 months. Realistically, the exercise might be much shorter (6 months) and could run longer, to 18 months. We left the shorter end out, and kept the longer end in, to drive home ease of use considerations.

    Paul Vebber: I’d like to be able to create a family of games. Not necessarily going to happen unless and until CGSC regains the freedom to directly contract for development. This sim isn’t intended to do everything – but it’s supposed to let us build the broad brush of a given situation and let individual classes run with it.
    Entirely agreed that the sim does not create the learning experience. The sim is the background, providing data for the decisions. The instructor may well tweak the simulation’s results in order to set things up for some learning point. (So why not simply have the instructor provide all the answers personally? Workload! It’s been tried.)
    I also agree that the simulation cannot and will not do it all. And your point on the simulations’ validity is correct – which is why we don’t seek to make it predictive. Correct in terms of the real world doesn’t matter… correct in terms of being plausible enough that the students don’t throw the BS flag at it is good enough. From experience, that is a surprisingly low bar.

    Rex’s #2: I’d envision a melding of #1 and #3: the sim resolves the outcome based on student orders, unless the instructor, seeing what’s going to (or has) happen, takes a hand and (quite possibly after running the turn), presents the situation to students, uses their decision, and edits the outcomes to suit. [This is why there’s a steady drumbeat of “white cell editable” shot through this.]

    Dropping into detailed focus is great, but it takes time, and time is short; it needs to happen only when the instructor decides the time spent will be worthwhile.

    AI: Honestly, in the most common projected use, the AI won’t get used at all. We probably ought to move AI to being an Objective level objective instead of Threshhold. (NSC likes to term capabilities as Threshhold (gotta have it) and Objective (nice to have) – I think those get clumsy, and ideally we’d have more gradations, but what the heck.)e is, in principle, leader-centric: it produces data for one person to interpret and make decisions on. A staff-centric game produces more data for more people to chew over. Both of these contain a spectrum of options, especially staff-centric. We make a point of it in part because there are those who are telling us to the UrbanSim to drive this exercise. UrbanSim is a great leader-centric tool, when you have the scenario you need. Unfortunately, there isn’t enough detail in it to really drive a staff, and building new scenarios requires access to source code.

  26. Graham Longley-Brown 09/07/2012 at 2:12 pm

    In defence of the GCSC not specifying the aim of the training, I’m pretty sure that they will have the course Training Objectives, Enabling Objectives etc well mapped out. These will (I hope!) have shaped the simulation requirements but simply haven’t been provided to us with the other papers. That said, I agree completely with Bill and Rex with respect to the right way to go about designing a wargame. The wargame design process I use for training purposes is derived from Peter Perla’s and hasn’t changed much since ‘The Art of Wargaming’ was published in 1990:

    1. Specify the exercise Aim and Training Objectives (TOs)
    2. Identify the people to be trained, their roles and the decisions they will be expected to make
    3. Determine the desired effects on the players, and the exercise activities required to achieve these
    4. Determine the scenario and types, level and sources of information the players will need to make their decisions and to enable the TOs to be achieved
    5. Identify the structures and processes required to achieve Steps 3 & 4
    6. Identify or design the tools, technology and SMEs needed to populate and enable these structures and processes
    7. Create an audit trail by documenting all decisions taken and the reasons for them

    Consideration of the tools & technology (including the simulation(s)) is just about the last thing you do. If something akin to this process hasn’t been used by the GCSC folk in determining the simulation requirement it will prove problematical.


    With regard to the difference between training and education, it does matter. Education, training and on the job experience lie on the same continuum, but the approach to each often needs to be quite different. When it comes to designing a simulation or a wargame there can be major implications. The best explanation of the distinction between education and training is a quip used by General Andrew Graham, recent Director of the UK Defence Academy, who thought the difference worth noting. Apparently light-hearted, it sums up the distinction perfectly: your 14-year old daughter comes home from school and tells you that she has been taught sex. You ask her if it was sex education or sex training…


  27. Brant 09/07/2012 at 12:47 pm

    “Specificially, what the participants were to learn or train for.”

    I think it was intentionally omitted. I think that’s part of what the CGSC guys are looking for: What do *you* think they should be training for?

  28. Bill Haggart 09/07/2012 at 11:59 am

    Rex has hit on one thing I was surprised not to find in the GCSC specs: Specificially, what the participants were to learn or train for. The Specifications are pretty much the design of the simulation, what it is supposed to do. What is completely absent is the ‘So What?” as Graham points out. I realize that this is meant for brigade and battalion level staff, who all have very specific roles in any “stability” operation, so a great deal of the ‘So What?’ is assumed, or from the design specs, the purposes of the exercises are left to the instructor using the sim.

    From my experience this is completely backwards in approach. [Not to say it is wrong necessarily, just opposite.] When a bid is out, the specifications detailed are what participants should learn, be able to do, when, where and how: The environment in which they will be expected to use the knowledge and skills learned in the simulation. Also the constraints on the learning process in time, resources and delivery are outlined.

    The actual architecture of the simulation is left to the designer, where he chooses the best methods and mediums to achieve the simulation goals. With the GCSC specs, the simulation structure has been detailed and the medium [computer programs] chosen, but nothing has been said about what is being trained or what the participants should learn.

    As a side comment, I am not sure what the difference is between Education and Training, a distinction Graham makes. In either case, education or training, the participants will be expected to use what they have learned in real situations. If not, what’s the point?

    In anycase, asking what simulation structure is ‘adequate’ as Rex asked, can’t be answered unless the learning goals of the simulation design has been detailed. For instance, depending on the goals, vignettes and injects may be completely irrelevant, or significant elements. It all depends.

  29. Brant 09/07/2012 at 11:45 am

    no – it requires a multi-polar set of participants in which you have other players at the table. There’s no reason not to give USAID a seat at the table, with a set of tools available to them that might realistically replicate what’s available to them on the ground.
    You could invite the Red Cross* to come in and play, too, and represent themselves, but I don’t know if they would. While it might give them some excellent exposure to what the US can do and how they can work together, a lot of those NGOs might feel their neutrality is being compromised by training with the US like that.

    * or ______ NGO

  30. Rex Brynen 09/07/2012 at 10:02 am

    Riots in Bugtussle again, Brant? I say we nuke the entire site from orbit. It’s the only way to be sure….

    THOUGHT #3: A lot hinges on the AI.

    The draft requirements outline as platform in which the behaviour of RED and others is largely driven by AI. The problem here is that AIs commonly behave in very reactive and non-adaptive ways. The actual parties encountered in Stability Operations, however, are typical very dynamic and adaptive. Moreover, their behaviour is often one or trying to manipulate external forces in a way that advances their own domestic interests.

    This is an important aspect to depict in the learning process, I think. It is very easy for military officers to view themselves as being in the drivers’ seat, as a user of tools—and, in so doing, to embody a dangerous amount of hubris. Instead they need to recognize that they amy be only one of many rival drivers, and that others may view them as tools to be manipulated. Rival villages or warlords may provide false intel about each other, for example, in the hopes of getting external military forces to provide support or to do their dirty work.

    Are vignettes and injects adequate to do this? Or does it require something more within the AI itself?

  31. Brant 09/07/2012 at 9:28 am

    two thoughts at the ‘execution level’ of actually building something

    1. There needs to be a *lot* of scoping of what’s really expected to be accomplished by the unit being trained by this sim. The timeline for the training (line 156) is to compress 12-18 months into 2-4 days of real time. However, as noted above, there’s significant research that to truly stabilize a country can take 20-40 *years*. How much do we want to have the brigade staff “reinventing the wheel” and how much should them be falling in on an existing mission? Think of the Sinai peacekeeping mission for a minute, where we’d had US observers for 25 years as a part of the UN mission there. The first 3-5 rotations did a lot of planning, adjusting, modifying, etc, but since about 1988, they’ve been falling in on the same basic plan with minor adjustments. How many years do you want to have already been ‘under the belt’ when the scenario starts.

    2. Playing out the time. (similar to Rex’s comment of 08/07/2012 at 1:22 pm) Once the execution is underway, do we run a continuous time scale, where the game runs x-number of equal-time turns? Or jump around to key events? There’s probably a limit to the training value of week-to-week tedium of “your guys go on patrol; nothing happens” until it’s interrupted by “holy shit, the town of Bugtussle just erupted in protests and violence!” At that point, you may want to shift the time scale closer to real-time to allow the participants to actually exercise their proper processes for developing their plans and making their decisions without having to move at warp-2 just to keep up with the game.

    Bigger picture is this:
    What are the types of decisions we want the player to make, and to train their brains to make, and within what time frame do we expect them to be made?
    That’ll drive the echelons represented in the system(s) and the feedback loops (positive and negative) presented to the players based on their actions.
    Finally, how wide do you expect the players’ decision-making to reach. I would suspect that no one is going to ‘win the war’ in 12-18 months, but you sure could lose it. That’s a pretty crappy definition of success for a unit training: “Don’t f&#k up” I think most commanders are looking for something a little more meaningful, with the understanding that the unit might do everything perfectly right within their span of control, but some do-gooder missionary NGO starts handing out Bibles with emergency food in an Islamic-dominated disaster area and suddenly everything goes into the crapper.

  32. Rex Brynen 08/07/2012 at 5:29 pm

    Graham: Thanks! I’ve reposted your comment as a blogpost here: …so that I could include the diagram.

  33. Graham Longley-Brown 08/07/2012 at 4:32 pm

    Rex, Paul,

    What you are saying makes a lot of sense; I hope the GCSC folk take note! Frustratingly, I can’t post an image here so you might want to skip over to Phil’s simulatingwar yahoo group to see what this post refers to (if someone can tell me how to copy an image here please tell me!).

    The image is of a ‘Campaign Tree’ process used in a ‘Hybrid Warfare Tactical Wargame’ at our Land Warfare Centre. This supports training for company HQ and/or battalion HQ commanders and staff. It addresses some of the points discussed above – and it works! The concept is simple: the Training Audience (TA) move through the Campaign Tree from vignette to vignette along a path determined by their own decisions and actions. Vignettes are played out in real-time and use a real-time (largely kinetic) simulation. The TA decisions and actions taken – under pressure – during each vignette are adjudicated by Excon SMEs and dictate the path taken to the next vignette on the Campaign Tree. The periods between vignettes are modelled using a soft factors simulation and last from 3 – 8 weeks. Hence the consequences of the actions taken by the TA during the vignettes, combined with their ongoing Concept of Operations and decisions taken in response to injects and events fed in by Excon are played out during the longer time periods.

    The process integrates two simulations, adjudicated and moderated by Excon ‘Rainbow Cell’ SMEs. MEL/MIL injects are used as required to bring out Teaching Points; these are, in the main, pre-considered but can be dynamically scripted. Likewise the vignettes are pre-considered and pre-loaded in the real-time simulation but are modified just before going live depending on the TA plan during the preceding time period and can be executed as deemed appropriate by Excon. The diagram doesn’t show AARs, information flows etc – it’s just the bare bones concept.

    Although I think this is quite simple in concept it’s hard work to pull all the elements together in the space of a 1-day training event that spans all or most of an operational tour deployment in game time. But it works…

    The thing I love most about it is that it allows the TA to create their own narrative; it’s their actions that determine the path through the Campaign Tree – their story. Hence they are more likely to internalise lessons learned. Check out Peter Perla and ED McGrady’s article ‘Why Wargaming Works’ (Naval War College Review, Summer 2011, Vol. 64, No. 3) for more on why a created narrative, as opposed to a presented narrative is so strong a learning mechanism.

    It’s also very flexible. Delete the soft factors sim and insert a board game if you like. Run it all using just Military Judgement.

    So what? Paul makes the point very well that the GCSC requirement is assuming a computer simulation solution that can do everything. I don’t think such a sim exists, or will do in the near future, so a more flexible approach is needed that integrates a number of simulation methods and exercise processes. Paul’s ‘Right answer’ of a ‘family of games (decision-centric tools) where the students use a variety of small, purpose focused games to get at specific aspects of the problem’ is spot on. Rex’s ‘preselected teachable moments’ are encapsulated in various places in the Campaign Tree, and lessons learned are reinforced by the narrative created by the TA themselves. In summary, I suspect that this requirement will need some innovative thinking rather than assuming (hoping?) that someone will come up with a sim that walks on water.

  34. Rex Brynen 08/07/2012 at 1:22 pm

    Paul, I couldn’t agree more about the richness and complexity of the “usual” stability operations environment (there being no “usual,” of course–which just complicates the educational challenge all the more). That, in turn, leads me to…

    THOUGHT #2: Conventional vs unconventional scaling/level of analysis in simulation design

    The CGSC requirements are quite clear that the simulation is intended for the brigade level, and that they do not want detailed rendering of combat operations so as to avoid what Graham has rightly referred to as the “Nintendo effect.” There are good reasons for this—not only is it distracting and diverts from the primary educational purpose (refining SO planning skills), but it also encourages the sorts of micromanagement that can be a problem in the field.

    That being said, I do think it is useful for operational, even strategic games to provide an opportunity to zoom down into the weeds, so as to provide some teachable moments about how actions taken at a very low level (the “strategic corporal”) can have multiple higher-level second- and third-level effects, and to similarly illustrate how plans and orders that make sense at the HQ level may be rather more complicated on the ground.

    A good illustration of this is provided in Rory Stewart’s excellent book The Prince of the Marshes, which recounts his time as a CPA deputy provincial governor in Iraq. At one point, a local political dispute leads to protests against the newly installed Iraqi governor. British troops are sent to defend the facilities of the new local administration (a sensible order, no doubt, viewed from above) but in the end the crowd are not deterred by a show of force. Since the small number of British troops aren’t about to open fire on a large number of unarmed demonstrators pushing past them, they essentially have to abandon the offices to be sacked.

    On can imagine the proposed CGSC sim handling this in three ways:

    1) Higher-level command determines ROEs. and these then shape the outcome and secondary effects of operations (such as an order to secure a location). The result is presented to the player implicitly in SIGACT or other data. This seems to be what is envisaged in the draft requirements.

    2) The sim generates a vignette of an outcome based on the ROEs (say as a detailed unit AAR), in a much more description and hence immersive form.

    3) For preselected teachable moments, the sim suddenly shifts the player from brigade staff officer into the shoes of senior NCO or young LT on the ground, faced with a difficult decisions. The orders say secure the location. We can’t hold the crowd back. They’re throwing rocks, we’ve seen some guns, but we’ve taken no fire. Do I do A, B, or C? The choice has significant impact on how that vignette unfolds. This isn’t done often enough to represent micromanagement, but on selected occasions to offset the dry abstraction of orders and planning.

    As is probably evident from my comments, I rather like the third approach, and use it myself. In this years Brynania sim, for example, the UN force commander is generally confined to making decisions about the deployment of company and battalion sized contingents, and it is assumed they just “do their thing” once in position. However, at one point a UN patrol stumbled across evidence of major rebel movements in violation of the ceasefire, and had only a short time to avert a possible surprise attack on the government. Consequently, I put the UN force CO in the shoes of an Indian army Major with on the ground who was given less than an hour in which to do something. Similarly, the insurgent military CO was temporarily made an insurgent battalion commander who found a few dozen Indian peacekeepers blocking a key road. It provided a useful opportunity to highlight the complex and contingent SO interplay between the balance of forces, signalling, deterrence, bluff, consent, and legitimacy—and how small incidents can go badly wrong (or in this case, very right–the rebels backed down).

  35. Paul Vebber 08/07/2012 at 11:40 am

    Rex brings up an increasingly problematic feature of what has become “plannning-centric warfare”. The spec details a huge amount of detail that the sim needs to represent, from graffiti to drive by shootings, to key leaders (and by inference) the networks they are the leader of. So what is the relationship between the actions the players take in the game, and the “sigacts” that occur?

    In the land of planning-centric warfare, if everyone agrees during planning that 21 acts of grafitti, 7 protests, 4 drive-by shootings and a leader trying to create an IED building capability can be addressed by (insert course of action determined by consensus of what the players *believe* should work. Then if the (Measures of Performance) indicate that the course of action was executed successfully, then it is ASSUMED they will have the desired effect on Measures Effectiveness (MOEs). The MOPs can be measured accurately, quickly, and relatively easily. The scheduled activities happened as planned, except for (insert issues here), on a daily if not hourly basis. The MOEs can be measured, but thee timescales vary. They tend to be weeks to months, not hours and days, and sometimes can be years.

    The issue that makes ‘planning-centric warfare’ problematic is that there is no “theory of action” to connect the MOEs to the MOPs, We have “successful simulations” for conventional warfare because we have a moderately successful “theory of action” for the mechanized battlefield. At least at the Brigade/battalion level where the effects of training, morale and leadership differences “even out”. This works for “planning comparison” where, holding the intangibles equal, you can get a decent idea of “in theory” which course of action is better, and thus plans can be assessed, at a high level, and a decision made.

    Even this “battlefield kinematics” theory gets dicey when we get into the small unit actions, where those issues of training, morale and leadership get increasingly important the further below company level you get. Given that SO combat tends to live down there almost exclusively, even what we think we know, we really don’t.

    So this brings us back to the “simulation spec” – which is really a “display spec”. Here are all the categories of things that the simulation has to be able to represent. The problem is: “Where is the simulation”? Where is the basis for how whomever makes this simulation does all the “complicated part” about how all these “display items” interact?

    The Stability Ops manual has some “motherhood and applepie” about it, and to be fair, gets into some of the knickers in a few places, but as a “theory of action” for a simulation representing as varied and detailed a set of “stuff” as this spec includes, well, I guess that is left as an exercise for the white cell.

    The question here is how to make a simulation of something you don’t understand? How do you do the comparison “in theory” between course of action, when you have only a few touchpoints on the elephant you can’t see? Given the complexity of the situation, is it even “theoretically possible” to know the effects of two different courses of action 1 year or 5 years down the road?

    The goal of the project is laudable, but if it comes down at the end of the day as another case of a “simulation” being in the same room as SMEs to give the opinions of the SMEs (all you have at the end of the day here) more credibility (they used the sim, its not ONLY their opinion) then not only will it be a failure, but its intellectually dishonest.

  36. Rex Brynen 08/07/2012 at 12:55 am

    I’ve had quite a few thoughts on this, but rather than offer them in one omnibus comment, I’ll try to fire them off as time allows.

    THOUGHT #1: It doesn’t just matter what they did, but how and why they thought it was the best idea at the time

    One of the things that strikes me in the simulation, as outlined in the requirements, is that while the key educational purpose is to explore planning processes, much of the planning process takes place outside of digital component, and presumably will not be documented within it. Am I wrong about this?

    Let me explain what I mean. Students—representing staff officers—develop a course of action and undertake the necessary planning to support it. These discussions are presumably occurring face-to-face, since the requirements say nothing about integrated communications components within the simulation package. Having decided what is to be done, the necessary instructions are entered into the simulation by the student operators—the software then does its black-boxed magic, and the effects are then outputted and the turn advanced.

    The problem here is that while the software (as per the requirements) can document what was done and what effects it had, it does not document how that decision was made. Of course, this is one of the things that course instructors ought to be doing, and some introspection into the decision process itself can be provided by student debriefs and hot washes. Still, if you are going to design a purpose-built simulation platform to provide PME on stability operations, why pass up the opportunity to systematically record data on process as well as action/outcome?

    While my own educational simulation on peace operations is far more low-tech and seat-of-your pants, it does have the virtue of capturing massive amounts of information (typically, 10,000+ messages) not just on what students did, but why they did it, and what decision-making dynamics (limited information, strong personalities, poor consultation skills, a toxic discussion environment, etc) helped to determine the COA adopted. This is enormously useful after the simulation, in reviewing information management, leadership style, process-tracing decisions, etc.

  37. Gary Milante 07/07/2012 at 1:15 pm

    Just reviewed both docs and here are some quick thoughts, with more to follow, I hope, as it is an interesting simulation concept and looks to be the foundation of a great discussion. Thanks to James for posting this and opening it to comments. Graham’s observations were very insightful, so I’ll build off of them. For now, I’ll focus on timescale, players and interface.

    To follow on Graham’s Timescale comments: Let’s simplify and say there are only four international actors involved in a stability operation – a humanitarian actor, a defense actor, a diplomatic actor and a development actor. As I’ve discussed elsewhere – humanitarian and military actors engage with the local populations over weeks and months, delivering timely responses to fluid and changing environments during (relatively) short deployments usually with specific targets and exit strategies. Development actors engage at much longer time frames – years and decades even. In the WDR 2011, we found that reform of corruption and building good governance, not to Denmark, mind you, but from current South Sudan to current Vietnam or Ghana, takes 17 to 25 years for the fastest reformers. Likewise, reducing poverty and building education systems take decades of slow and laborious progress, often interrupted by setbacks in the case of fragile and conflict-affected states. Diplomatic actors in some sense transcend the shorter time scales, but are often engaged during particular crisis. As a result, military and humanitarian actors are typically long gone by the time a development actor sees any of their glacial progress.

    These realities make agent-based modeling of stability, peacekeeping or peacebuilding (depending on your perspective) really complicated and creates a tension in time scale. To keep the simulation interesting and engaging, you want to have the immediacy of “real-time” – kinetic events, engagement with enemies, responding to crisis, but to reflect the kind of change that is necessary to build a stable/”resilient” state/society that has truly “escaped” fragility and survived out of a crisis, the simulation would need to cover at least the 1.5 years described in these concept notes.

    These time frames complicate play as well: Imagine the conversation in Blue Team (excuse the Banker trying to talk military for a moment):


    Military Commander 1: “I’ll deploy two squads of elite commandos here and here – they’ll engage in night missions to identify and eliminate the hostiles rumored to be in this area.”
    Military Commander 2: “I’ll send in air support from this base here to and regular drone missions to increase our odds of catching them in the open.”


    Humanitarian Actor: “Just for the record, I’m not coordinating at all with this military action, but since I know it is happening and that traffic on this road will be interrupted, I’m going to put food and water in place here, here and here so that we can make sure that people are taken care of if things get out of hand.”

    Somewhat less energetically:

    Diplomat: “We’re going to continue to meet in the capital with representatives of the fringe rebel group in hopes that we can get the talks


    Development Actor: “We’ve finished lining up the financing, so we’re now going to contract firms through a competitive bidding process so that they can begin work on rebuilding that irrigation system that was destroyed three months ago – we should have the firm ready to start work in four months and a brand new irrigation system eighteen months out – wait, the game ends next month?”

    Now, obviously, you could just ignore the less interesting roles and script them or run them by white team as exogenous events to the simulation. But the problem is that the population is interested much more in the things that the diplomats and the development actors ultimately are delivering – yes, the people don’t want to be shot and killed, but victory in any meaningful sense for a stability simulation will be defined by success in getting lots of enemies to lay down weapons, by rebuilt infrastructure, good jobs, investment and growth in the economy and, in the end, consumption and prosperity of the population. So, in my opinion, a simulation would need to include the modelling of realistic interactions of humanitarian, defense, diplomacy and development actors who ultimately need to deliver all of this through coordinated and sustained action – not easy to do. Perhaps the timescale should be more fluid? Allowing white team to accelerate time while “peacebuilding” is going on and slowing down time when there is violence. Indeed, in such a setup the real fatigue of the players will mimic the fatigue of the international community as players would want “quick wins” and would feel encumbered or slowed down by in a morass of conflict.

    I also think that the rainbow cell suggestion is going to be a better fit for what you are trying to get out of this experience. As I understand it (or, more accurately, as I conceive of the world), individual actors in this environment are be beholden to certain interests (whether populations on the ground, donors or congresses at home) and have their own victory conditions. The purpose, then, of such a simulation is teaching agents who might find themselves in such a situation how to identify stakeholders and shotcallers, understand their obligations and objectives, and determine whether they are compatible with one’s own interests and objectiives. We try to do this in Carana, but honestly, we still have challenges in getting players to play their interests.. A rainbow cell approach which would allow for allegiances and alliances to form and break over the course of the simulation.

    This brings me to interface. Even though I am a huge fan of roleplay and boardgames – I actually think if you’ve got the budget for something virtual allowing for users to network in to a platform that does all the math for you and runs all the fiddly parts of a very complicated model, you should use it. What I would recommend is more of a hybrid model for this size and scale of simulation – players interact with each other for fifteen minutes in face to face and then return to their computers to put in their “orders” (a la Diplomacy). Then all orders are executed simultaneously and readouts are given to each actor. This gains you two things:

    1) The complexity you need for this scale of simulation, managed by a computer rather than people, so that the white team can spend their energy on ensuring that the broader narrative is developing and that the right lessons are being learned (since this is a teaching sim).
    2) Much more prospect for hidden/incomplete/incorrect information – computers are great for rendering fog of war, giving incomplete information to players (I hope things like the “governance” measure and relationships between factions would be fuzzy for the players – when do we ever really know what the quality of governance or strength of an alliance is?). Much of the challenge of accomplishing anything in fragile environments is understanding the context, the relevant stakeholders and what is and is not possible and simulations that provide complete pictures do a disservice to people going into these environments when they suggest that we can ever know everything we need to know.

    Those are some quick thoughts, biased by my own experiences with Carana. Looking forward to seeing how this conversation continues and will reflect a bit more on the notes for some additional comments.

  38. Paul Vebber 07/07/2012 at 11:59 am

    To riff off of Grahams excellent starter, the biggest danger in this sort of effort is trying to create a single, monolithic response that will die under its own weight. What I see the “Right answer” being is a family of games (decision-centric tools) where the students use a variety of small, purpose focused games to get at specific aspects of the problem.

    No one simulation is going to be able to wrap EVERYTHING associated with stability ops up and tie a bow around it in such a way that students will learn anything other than “its really complicated”and “individual actions have good results only in the aggregate over long timescales, but can have really bad results overnight – and I (and my chain of command) have to learn to live with that”.

    I see some of the “games” being Euro-style games, some of them being multi-play “Civilization” style games, some of them being classic “hex and counter” board games some being more like “CPX’s that is seems the requirement is focused on.

    The problem’s solution in this requirement is set up for failure becasue it appears to expect the tool to create the learning experience. That is like saying I want a simulation of the white collar workplace that I can have college students play so they will be successful when they have to operate in it. That is not how College works – too educate people through a variety of means, to have the toolset they need to be successful. The “end state” they want is not going to be achieved by “creating the game”. It requires the establishment of a cadre of gamers and game designers who can work with subject matter experts and craft a game-centric educational experience that provides leaders the toolset they need to survive in the “stability operations environment”.

    That is not going to be accomplished by “immersion” in a “simulation environment” and seeing who doesn’t drown in the deep end of the pool.

    Paul Vebber

  39. Graham Longley-Brown 07/07/2012 at 9:54 am

    Hi Rex,

    That’s not asking for much, then, especially with the ‘low-overheads’ caveat snuck in!

    Here are my thoughts and comments.

    1. Education not training (or analysis). It’s important to note the educational context as this has significant design and execution implications. Deciding at the outset to use a turn-based approach is sensible: adjudication (and potential modification) of simulation outcomes by instructors/SMEs helps steer the exercise and ensure objectives are met; and real-time computer sims risk causing a ‘Nintendo effect’ whereby students focus on the current battle unfolding before their eyes rather than planning and thinking. Educating students necessitates more built-in review points, hot-wash AARs and – potentially – the need to re-wind the sim to a previous point in time. These features are far more important than in a training context.

    2. It’s the Process, Stupid! The process will be more important than the sim. Some of the features mentioned above rely both on the correct sim and a robust exercise process. This is often overlooked. It is even more important in exercises set in the Contemporary Operating Environment (COE) than in conventional warfighting because there are more factors to consider and many more information flows and sources; how these are developed and presented to the students is critical.

    3. Timescales. As the initial comments on Phil’s simulatingwar group have shown, modelling the long timescales involved in Stab Ops is an important factor and a challenge. The usual Military Decision Making Process tends to be elongated, with targeting processes, coordination boards et al occurring far less frequently than in conventional warfighting. The emphasis for this requirement might well turn out to be more on command decisions and medium- and long-term Consequence Management (2nd and 3rd order effects) than on time-pressured staff work. I wonder why the requirement states ‘staff-centric’?

    4. SME input is critical. No sim will capture the complexity of the COE in the foreseeable future as well as a good team of SMEs with real-life experience. I know the requirement – quite rightly – states that the sim is not to be predictive, but SMEs will deliver more insights and real-world lessons to the students than the best sim. Hence real people should be involved if possible – see the next 3 bullets.

    5. Rainbow Cell concept. The exercise can’t just be Red vs. Blue. I know the requirement mentions multiple factions but this needs to feature more strongly. I use the concept of a ‘Rainbow Cell’ (staffed by SMEs) to represent all the Excon colour functions: White (political, IC etc), Brown (local populations), Green (indigenous security forces), Orange (armed non-state actors) and Black (criminality). If not specifically mentioned (and built in to the process) these risk not being considered.

    6. Outside perspective. The IO/NGO perspective and input is critical. However, IO/NGO personnel tend to be anti-military and unwilling to take part in anything as militaristic as a wargame. How you capture these guys’ crucial inputs is key. Quick example: a real Course of Action Wargame had to be termed a ‘synchronisation conference’ just to get IO/NGO reps to come.

    7. Other Government Departments (OGD). Also critical in Stab Ops for obvious reasons. While more willing to participate, I find that – in the UK – these guys do not have the budget to allow them to support many military exercises. The requirement must stress the benefits to OGD personnel if the military want them to participate.

    8. Design principles. The principles I think are important in designing an educational Stab Ops/COE wargames are listed below and addressed in para 9:

    • Speed of set up
    • Flexibility
    • Transparency of inputs and outcomes
    • Adversarial approach
    • SME inputs

    9. Computer Sims (tend to) fail in most of these respects! The time taken to initially configure data (especially geo-data) and create a digitised setting and scenario is measured in months. The argument that once done this doesn’t need repeating rarely applies because, for various reasons, people want better or different scenarios. The Cerasia setting is mentioned in the requirement document, and might be used. It belongs to NATO (Joint Warfare Centre) and is available free to NATO Nations. I helped write it and know that developing something similar is a huge task. The time needed to re-configure a computer sim reduces flexibility. The sim needs to reflect the ever-changing COE. This is especially the case when it comes to Human Terrain (HT). Producing the HT for a setting takes a long time, and there will be inevitable pressure to develop or change this as time passes; it’s not a once-only event. The tendency for computer sims to be ‘black boxes’ has to be avoided; the students need to understand the cause and effect of their actions.

    10. So what?

    a. I’d consider a manual solution – i.e. a board game. I’m no Luddite (and have spent most of my professional wargaming career integrating computer sims into exercises) but in an educational and low-overhead context I’d consider further the ‘military chess’ comment in the requirement paper. There are a number of existing manual sims that might fit the bill: Brian Train’s Kandahar and a number (or combination) of Phil Sabin’s MA students’ games spring to mind. With the right SMEs delivering injects these naturally tick all the design criteria above.

    b. If it has to be a computer sim (and the requirement is from the Digital Leader Development Center, after all!) then I’d think hard about:

    (1) The overall exercise process. This is more important than the sim and should be determined before the sim is selected or designed.

    (2) How to ensure SME inputs and the ‘outside perspective’. Money spent on SME support will deliver more insights and educational value than endless software and data configuration and re-configuration.

    (3) Methods to ensure ease of set-up and built-in flexibility to ensure the ever-changing COE can be reflected.

    (4) Sharing existing settings and scenarios.

    c. The best computer sim I’ve come across for modelling Stab Ops is the UK Dstl’s Peace Support Operations Model (PSOM). This is mature and has been used ‘in anger’ on ops. However, it’s a theatre-level model, but with the possibility of reaching down to the tactical level.

    I’ve cross-posted this on Phil Sabin’s simulatingwar yahoo group too.

    Graham LB

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: