Conflict simulation, peacebuilding, and development

Daily Archives: 28/10/2013

Simulations miscellany, 28 October 2013


Some recent items on simulations and serious games that may be of interest to PAXsims readers.

* * *

Wired UK has an article on “Battling insurgency in war-torn nations with immersive 3D sims”  that suggests an optimistic (and I would say ridiculously optimistic) vision of the predictive utility of computational simulations addressing complex social and political processes:

“Simulation science will transform how humans make decisions for the next 100 years.” If Justin Lyon sounds evangelical, that’s because he is. The Texan-born founder of Simudyne is on a mission to persuade public and private sectors that their reliance on old maths is deeply flawed. That the buzz of big data is no longer a buzz — it’s sitting in his cloud-based simulation software Simudyne, and to not use it to help make decisions, particularly in the context of warfare, should be criminal.

“Humanity’s continued reliance on mathematics the ancient Egyptians understood is the wrong thing for our society,” Lyon tells, exasperated by what he sees as the world’s snail pace attitude toward adopting complex models to analyse a complex world. “The vast majority of consultants still use spreadsheets to make decisions based on that same math.”

The world is, however, ready for a change Lyon believes, as demonstrated by the epic failures of the last decade: ourfailure to predict the danger of an unregulated financial market  has proven catastrophic, as have interventions in Iraq and Afghanistan. Both prove what happens when policy is based on a flawed model. Lyon believes Simudyne — a neat package of system dynamics and agent-based modelling, visualised in 3D virtual realities via algorithms — provides the disillusioned with a new way to fail safely. Policymakers can run simulations with meticulous detail thousands of times to allow for insights and generate statistical distribution.

It took a near total economic collapse for Alan Greenspan, former chairman of the US Federal Reserve, to realise the flaw in his policy. And Lyon argues his mistake was not just to wait for real-world failure to realise that flaw, but to place trust — an inherently human characteristic — in other humans. Greenspan believed in a free-market run by rational people, who would always act to protect shareholders to maintain the status quo. Catastrophe has too often followed the errors of individuals. So why not let the machine do some of the heavy lifting, and point out our mistakes before we make them?

“For every single exceptional human, there are a million David Brents out there,” says Lyon, “we’re trying to equip humans with the ability to interact with computers that allow humans to better understand the consequences of decisions.”

So far it’s been used it to create realtime emergency response solutions (in a trial 98 percent of actors received the correct escape route on their phones when trapped in a tunnel), map the US healthcare system and replicate an entire country of three million people.

But it’s in counterinsurgency operations where the applications get most messy, risky and — if they work — perhaps most revolutionary. Simudyne has worked on defence projects in Afghanistan and Iraq, working with local businesses and elders and using open source intelligence from the World Bank, IMF, UN Agencies, charities and NGOs.

There was a precursor to Simudyne’s work, in the form of the COIN (Counterinsurgency) Dynamics Afghanistan Stability Chart. It’s a mathematical model to describe counterinsurgency, but to the untrained eye the chart’s a mass of neverending circles pointing back at our own failure to comprehend it. General Stanley McChrystal famously said “When we understand that slide, we’ll have won the war”. Lyon tells us, “I would have said ‘you’re absolutely right sir, if you had understood this, you would have won the war'”. It measured the turning point from civilian to insurgent, taking into account things like accessibility to guns and opportunity. Its says increasing the killing and detaining of insurgents increases the flow of new recruits, but Lyon believes simulation can also help us understand why that happens, and why getting the ratio of things like counterinsurgent troops dedicated to combat or intelligence patrols right can positively affect the outcome.

The point of an all-encompassing system like Simudyne’s is to seek out the “unintended consequences” of our best intentions. And it’s during counterinsurgency these are perhaps greatest. The system provides another point of view, highlighting things policymakers may never have thought of. When mapping the US healthcare system, for instance, it showed people were not using bike paths planners had spent millions on, because not enough was spent on policing — people were wary of crossing dangerous areas to reach those paths. “These obstacles frustrate our best intentions — we have to connect them together and allow decision-makers to play the game.” Like a pilot training for hours in a simulator, they know it will be different in the real world, “but the more training they have, the more likely they are to survive.”

Fortunately, the article goes on to cite some skeptics, who bring as back to earth. “You can have a sophisticated set of simulations and very smart people working on them, but if your assumptions are wrong it’s worthless,”  warns Celeste Ward Gventer. Phil Sabin also nails it:

Philip Sabin, a professor of strategic studies at Kings College London, argues the variables that matter most only occur to us once they’ve happened. This, is why presumptions will drive failure. In Afghanistan and Iraq allied forces instigated the insurgency, in part because they were operating off a bad model and poor assumptions that failed to take one key thing into account.

“In Iraq the invasion went reasonably well, and Bush said mission accomplished — in conventional terms they thought the war was won,” said Sabin. “Then the insurgency came. They hadn’t even considered these people wouldn’t be delighted to be liberated from Saddam Hussein. Their model was totally incorrect because it said it will be like 1945 and once you get rid of the terrible regime — the modern equivalent of Nazi Germany — people will proposer and be happy. The idea they’d fight for the remnants of Saddam’s regime was seen as absurd.” Like 1945, the entire regime was removed, but no one asked what those people would do. “Become insurgents of course,” says Sabin, “because you’ve given them no new way forward.” These are the kinds of factors a model won’t tell you, if you’ve built it from a position of bias. It demonstrates how ineffectual any such simulation can be, if the perceptions and politics at its core are inadvertently in direct conflict with the desired outcome. It’s the human factor that’s the simulation’s undoing.

I like Isaac Asimov’s Foundation series of books as much as the next science fiction nerd, but let’s face it—we are very far away from having psychohistorical model of human history that allows us the sort of predictive accuracy suggested in the article. Certainly, computation modelling (whether agent-based or otherwise) is really interesting stuff, and it can illuminate some really interesting issues. However, overselling the methodology it is to do it a disservice (unless, of course, you’re trying to sell contractor services to the military).

h/t Crispin J. Burke

* * *

A forthcoming issue of International Studies Perspectives will feature an article by Emre Hatipoglu, Meltem Müftüler-Baç, and Teri Murphy on “Simulation Games in Teaching International Relations: Insights from a Multi-Day, Multi-Stage, Multi-Issue Simulation on Cyprus.” You can access the early-view prepublication version at the link, if you are a subscriber.

This article reviews experiences from a large-scale student simulation, which concluded the Istanbul Conference on Mediation: Enhancing Peace through Mediation that took place in February 2012. We share insights on two unique aspects of the simulation. First, the paper examines a rare case where the simulation crossed paths with real life: a number of the impersonated officials (and offices) including the president of the General Assembly of the UN, the Minister of Foreign Affairs of Turkey, and the Director of the Policy and Mediation Division of the UN Department of Political Affairs were in the audience and shared their impressions. Second, the setup of the simulation was more complex than its typical in-class counterparts. Our insights from this multi-day, multi-stage, and multi-issue simulation can inform colleagues who plan to run larger scale simulations. Besides sharing experiences on a number of logistical points, we especially draw attention to the constructive role facilitators can play in augmenting the learning benefits accruing to the students from simulations.

* * *

The September 2013 edition of US Department of Defense Modeling and Simulation Coordination Office (M&SCO) M&S Newsletter is now available.

* * *

The Russian government is worried that there aren’t enough patriotic video games, as Michael Peck reports at Forbes.

* * *

Play the Past always has interesting material. However, we particularly enjoyed Peter Christiansen’s piece earlier this month on videogames and scientific revolutions.

As I discussed in a previous post, videogames tend to take a very deterministic view of technological development. Nowhere is this more apparent than in the concept of the “Tech Tree.”  While dedicated players and modders are usually quick to point out other flaws or deficiencies in games (often holding them to an almost absurd standard of historical accuracy), this strong thread of technological determinism is generally left unquestioned.  I attributed this lack of critique to the fact that technological determinism is so deeply ingrained into Western culture, especially the culture of tech industries like videogames.  So while the fact that a spearman has a slim chance of defeating a tank in combat may incite a minor revolt in some online forums, the fact that every culture on the planet, even landlocked ones, will develop sailing, optics, and the compass (and always in that order) never gets a second glance….

%d bloggers like this: