Playtesting the CIA’s training games

What do you think is the biggest problem for a modern American intelligence agent? Infiltration, perhaps, by Russian or Chinese agents? Retaliation by foreign governments against your local staff? Maybe it’s the threat of leaks and the ensuing attention of do-gooders in Congress?

Nothing so exciting. The biggest problem is bureaucracy: finding a meeting time all your colleagues can make, sharing information to coordinate your efforts, and convincing other parts of the government to actually pay attention to your concerns.

That, at least, is the message conveyed by a set of board games designed by the CIA to train its employees, and recently made public by a Freedom of Information request. Having once reviewed a very silly videogame created by the FBI to stop people from becoming terrorists, I was fascinated to know what I could learn from these documents about the worldview of the planet’s most famous, and perhaps most powerful, spy agency.

The use of games as training aids is very common. In the early 20th century, the Germany Army and US Navy regularly used wargames to plan their doctrines. The Westminster terror attack in Britain last year was eerily prefigured by an inter-agency terrorism readiness exercise in which several hundred imaginary MPs were murdered. The opposition Labour party has been wargaming what might happen if it takes power and is resisted by the establishment.

The CIA’s games originated in 2008 when an Agency employee named David Clopper was seconded to the Sherman Kent School for Intelligence Analysis and asked to come up with new ways to train agents. He turned to his hobby to create two games for classroom use, intended to teach specific lessons within a limited space of time. And so, this year, I downloaded and printed their cards and counters (you can find them here), gathered a group of friends, and gave them a try.

Collection's game board, as shown in documents released under the FOIA.

Collection’s game board, as shown in documents released under the FOIA.


The first game, Collection, is described as “a collaborative training game designed to teach how analysts and collectors in the intelligence community work together against a variety of challenges”. It plays out rather like Pandemic: three global crises blossom onto the board, with titles like “Caspian Energy Crisis”, “Palestinian Intifada” and “Venezuela Causing Trouble,”  and slowly ramp up as new cards are drawn. Players must then solve these problems before they explode – but not by deploying special forces teams or wiretapping embassies.

Instead, you move your pawn between spaces representing their “engagement” with different intelligence collection bodies, such as the NSA, the National Geospatial-Intelligence Agency (which processes imagery from America’s spy satellites), and the CIA’s own Directorate of Operations. You place “engagement gems” on these agency spaces to signify your deepening relationship with the workers there. Then you roll to generate reports – the more gems placed, the more dice you get – and once you get enough reports, the crisis you’re working on is “solved.” Reports can also be spent on reducing a crisis, creating a trade-off between solving and managing.

“Solving” a crisis doesn’t actually mean doing anything about it.

In other words, Collection is a nag-em-up. Your primary action is pestering other people to collect intelligence for you. You are splitting your resources not across a map of the world but across a map of the US intelligence community. And “solving” a crisis doesn’t actually mean doing anything about it. Per the rulebook, it simply means avoiding an “intelligence failure” by providing the politicians with all the information they need. If they do nothing with that information, or act and worsen the crisis, well, hey – you did your part.

Beyond this general view, fascinating specific lessons arise. For instance: as a player, you cannot simply move anywhere; you can only move between different agency spaces on the same crisis or between the same agency’s space on different crises. That suggests a recognition of the real time cost of gaining specialist knowledge and forming relationships. It makes sense that in order to engage a new agency about a new crisis, you’d need either detailed knowledge of that agency or detailed knowledge about the crisis. Having neither will slow you down. It’s pleasing to see a game that so firmly emphasizes that attention and intellectual labor are finite resources which institutions must manage as carefully as ammunition or cash.

A slide from a presentation delivered internally about CIA's games projects.

A slide from a presentation delivered internally about CIA’s games projects.

Similarly, building up engagement with an agency is slow going. To have any hope of succeeding in your report rolls by the time the crisis hits its endgame, you will need to be rolling a lot of dice. But you can only add a maximum of two engagement gems per turn, and only roll for a maximum of two reports per turn. In our game, faced with “Haiti Refugees,” “Indonesia Tsunami,” and “Sudan Atrocities,” we wasted far too much time rolling with low success rates instead of building up our capacity to roll and then blitzing the reports once it was high. It seems that in intelligence work, there are no shortcuts.

Most amusingly, there is a whole element of crisis resolution which is out of your control. Success rates when rolling for reports are determined not only by your level of engagement with an agency but by that agency’s “response level” to a crisis. Level 1 means it’s barely on their radar; level 10 means it’s their overwhelming priority. But, with a few exceptions, players cannot change these levels proactively. They must wait for the agencies to realize the scale of the crisis by themselves, which usually happens when the crisis itself gets worse.

It’s worth taking a moment to savor this. The CIA is reviled across the world as a shadowy driving force behind major events, yet its internal training game about responding to such events is structured by a sardonic cynicism common to civil servants everywhere. Analysts, Collection argues, don’t get to wake everyone up to a problem like some air-conditioned Paul Revere. Only when the crisis escalates – when the news media mobs politicians, and the politicians scream down the phone at their officials – can collection agencies be made to pay attention.

That said, players are not completely powerless. Agency response levels will only rise alongside the crisis level if said agencies are already engaged, even if only nominally. So if you haven’t made an effort to engage anyone with a crisis, nobody will properly respond to it. This complacency ultimately destroyed our own game of Collection, impeding our report rolls for the neglected Sudan conflict.

Collection Deck

The second training game is called Collection Deck, and is much more frenetic. It’s a competitive card game, with a commercial version being Kickstarted by Techdirt founder Mike Masnick. Rather than trying to realistically simulate any of the Agency’s processes, it’s designed to teach players about specific intelligence collection techniques they can draw on in their work – but it still contains appetizing insights into how the CIA sees the world.

Here, a set of intelligence problems – “Pakistan nuclear security,” say, or “Al-Shabaab attacks in Kenya” – are laid out on the board, each one with a difficulty level and one or more categories (Political, Economic, Military, and Weapons). Players take turns to deploy cards representing intelligence collection methods, from satellite imaging to simply reading foreign newspapers, against these problems to try to match their difficulty. Other players then represent “the system,” incorporating enemy intelligence agencies as well as simple error and coincidence, and play a series of “reality check” cards to counter the techniques. The first player can counter these counters, and if after the flurry of opposing cards has finished there remains a viable plan on the board, they get to take the intelligence problem away and add it to their score.

Collection Deck is supposed to help analysts understand how different problems can be attacked and what obstacles those attacks could face. There are even a set of “challenge” cards, which allow players to challenge each other to explain how their strategy could actually be applied in real life on pain of failure. In a South by Southwest conference talk included in the FOI release, David Clopper says: “I’ve had numerous students tell me after playing that game something along the lines of ‘I had no idea that capability existed in the Intelligence Community! Do you have someone I can talk to about how to bring it to bear on my intelligence problem?’”

Mockups from CIA: Collect It All, a commercial game based on Collection Deck.

Mockups from CIA: Collect It All, a commercial game based on Collection Deck.

A close look at the cards is appropriately educational. It’s nice to see the CIA makes good use of “captured documents and electronic media,” “handheld imagery” (i.e. spies taking photographs), “biometrics,” and “computer network exploitation” (i.e. hacking). There are satellite techniques such as “COMINT mapping” (geolocating foreign transmissions), “Overhead COMINT” (“satellites monitor and collect foreign voice communications”), and “Overhead ELINT” (“satellites collect signals from radar and electronic warfare systems”).

It’s particularly interesting to read the more mundane cards. They describe FBI representatives overseas liaising with foreign law enforcement bodies; debriefings of US personnel “who interact with foreign entities or activities of interest;” “analytic outreach” (“tap into the insights of US academics”); and information from “US agencies that are not funded by the intelligence budget” (such as USAID). There are a whole set of cards to do with open source intelligence (OSINT), which show the CIA subscribes to “commercial databases” and reads Reuters like everyone else. There is even a card called “Internet,” the description of which reads: “Analysts can do their own open source research.” It’s surprisingly powerful.

Many of the cards are actually redacted.

Counters include “aggressive counter-intelligence,” diplomatic expulsions, and media blackouts, but also accidents such as “misinformed sources,” “bad weather,” and “satellite failure” (description: “How much did we pay for this thing again?”). Significantly, many of the reality check cards are about internal enemies, not external ones. You can counter an OSINT technique “Customer bias” (your bosses don’t believe OSINT is real spy work), counter any collection technique with “red tape” or “internal politics,” and boost your own capabilities not by adopting some strange new secret weapon but with “working group meeting” or “leverage personal relationship.”

Unfortunately, many of the cards are actually redacted. What are the classified satellite techniques which are “not affected by bad weather”? What is the MASINT technique (measurement and signature intelligence) which “can only target problems in Eurasia, Middle East, or Asia”? What is the CIA not telling us about its OSINT techniques which are “not affected by Linguists Reassigned or Media Blackout,” or its SIGINT technique which is, ominously, “not affected by Encryption”? At least, like Donald Rumsfeld, we know what we don’t know.

(Redacted cards also interact joyously with challenge cards, forcing non-CIA players to bluff wildly about how they would apply techniques they cannot name. In our game, one player was defeated when she was challenged to explain how she would use satellite imagery to solve the problem of “China yuan revaluation.” The problem was later solved by reading a newspaper.)

Cards from the original Collection Deck. The white boxes mark redacted information.

Cards from the original Collection Deck. The white boxes mark redacted information.

What we can read, though, is suggestive. The most difficult ranked intelligence problems in Collection Deck include “Isis leadership,” “China cyber warfare,” “Iraq instability,” and, interestingly, “Afghan opium network” (University of Wisconsin professor Alfred W McCoy has called Afghanistan “the world’s first true narco-state”). The difficulty of the Israel-Palestine conflict is only rated at 6 out of 10. And what’s absent is as interesting as what’s present.

One of my players, Ben, is a defense analyst at a global research company. He was struck by how the CIA seemed “totally uninterested” in South America (bar Venezuela), the Pacific, Australasia, Europe, and the Arctic. He was also interested by how dominated by technology the collection techniques were – how much gadgetry there was and how little human intelligence (HUMINT). “So many of the techniques,” he observed, “make use of satellites – which has implications, because satellites are easy to take out.” Indeed, the US government was warned in February that China and Russia will be capable of blinding satellites with directed energy weapons within the next few years.

The Cat and the Coup.

The Cat and the Coup.

My friend Olivia, a reporter who has covered reforms to recruitment in the British secret services, found it fascinating that there were almost no unredacted cards to do with politics – no “change of leadership,” no “radical Leftists in charge.” One card does refer to a crackdown on reform by the United Arab Emirates, while others curiously identify “global warming summit” and “G8 meeting” as intelligence problems (the NSA has previously spied on Germany’s chancellor Angela Merkel and some of her ministers). By and large, though, there is very little here about manipulating other countries’ politics – unless that stuff is among the redactions. And as Olivia says, “the 1980s obsession with communism seems well and truly left behind.”  

That’s strange, because many of the CIA’s most infamous episodes involve precisely this kind of manipulation. In The Cat and the Coup, a documentary videogame about CIA plot which in 1953 replaced Iran’s democratically-elected prime minister Mohammed Mosaddegh replacement with Shah Mohammed Reza Pahlavi, the interference of Western intelligence agencies is represented by the sinuous presence of a black cat, which flows through each level, knocking over objects and causing trouble. And the work of the black cat is visible throughout the rest of the 20th century – from the 1973 coup which installed murderous Chilean dictator Augusto Pinochet to the massive program of sponsoring American artists (including Jackson Pollock, Mark Rothko and George Orwell) in order to prove that the USA was not a cultural backwater. It’s all in laid out the Family Jewels.

In the end, this is the most interesting thing about both of the CIA’s training games. They are not about being the black cat. They are not about intervening in world affairs. They are simply about finding things out, often battling your own office politics in order to do so. On this evidence, says Ben, “the CIA are trying to understand the world, not trying to change it.”

'Characters' from Kingpin: The Hunt for El Chapo.

‘Characters’ from Kingpin: The Hunt for El Chapo.

Bonus Round: Kingpin

There is another CIA board game which does involve more direct intervention. Kingpin: The Hunt for El Chapo, designed by Agency trainer (and commercial wargame designer) Volko Ruhnke, simulates the American-led attempt to capture Mexican drug lord Joaquin Guzman. It’s “designed to familiarize intelligence analysts with manhunting methodology” – to create a “mental model” for analysts of how such operations tend to work.

It is a team game, moderated by a referee, in which Cartel players and Hunter players try to outfox each other from either side of a screen. The Cartel must move their kingpin around the board, shielding him with defenses and deceptions while fulfilling his “needs” (such as seeing his wife or, er, sex workers) while the Hunters methodically investigate leads and try to nail him down so they can arrest him. It’s a nice illustration of how American law enforcement sees organized criminals: their need to live something resembling a normal life, to participate in vices and have fun, is their biggest point of vulnerability and will lead them to expose themselves.

But even this game doesn’t exist to train people who might lead such a manhunt themselves. It exists to train analysts who might work with agencies who perform such manhunts; with the Drug Enforcement Agency or with Mexican police. That, and the other two games’ overwhelming focus on bureaucracy, illustrates a truth about modern intelligence work: that it is far more boring than most people think.

Being a CIA analyst probably isn’t that different from being a private sector analyst like Ben. Most Western agencies today are huge, sprawling employers subject to elaborate codes of conduct, justification requirements, paperwork specifications, internal guidance, legal advice, and personnel policies. They have HR departments, compliance units, diversity strategies. They don’t care whether you can pilot a helicopter and parachute out of it onto a moving train. They want you to be responsible, judicious, and have good people skills.

It would not be true to think the modern CIA only watches the world and never intervenes.

There are things that these games do not deal with. They do not simulate the CIA’s drone warfare program in Pakistan, which has killed, according to the Bureau of Investigative Journalism, between 2,515 and 4,026 people, including between 424 and 969 civilians and up to 207 children. They do not depict the CIA officers who arrived in Afghanistan’s Panjshir Valley in 2001 carrying $10 million in cash – handing it out, in the words of journalist Steve Coll, “like candy on Halloween.” They do not concern themselves with the Special Activities Division, which continues to recruit soldiers from America’s most selective infantry and sends them without identifying objects or serial numbers to watch, train and kill people in Syria, Iran, Pakistan, and Yemen.

It is true that intelligence work has become more professionalized and more bureaucratized than it once was. It would not be true to think the modern CIA only watches the world and never intervenes. The insight these games provide is useful, but partial. Perhaps their most salient lesson is that there is no puppet master who understands and control everything. Instead there are just people, working inside institutions, within structures and procedures, according to political instructions, who produce and enable all that midnight surgery – and because they are just people working inside institutions according to instructions, they don’t always know exactly what they’re doing. Or at least that’s what they want us to think.

Top image: Kentucky Route Zero (Cardboard Computer).