Conflict simulation, peacebuilding, and development

Daily Archives: 09/01/2023

Heath: Wargames can’t tell us how to deter a Chinese attack on Taiwan—but different games might

Having posted a link earlier today to the launch of a major CSIS report on wargaming a potential Chinese invasion of Taiwan, now is a good time to also flag this thought-provoking piece by Timothy Heath (RAND) a few days ago at the Lawfare blog. In it, Heath argues that kinetic wargames focused on military operations are useful, but they aren’t optimized to tell us how we might stumble into war—or avoid it.

He states (emphasis added):

Wargames that simulate combat between the United States and China near Taiwan can provide useful insight about potential military challenges. However, analysts should be wary of repurposing the same games to explore political questions such as those related to deterrence, escalation control, alliance politics, and war prevention or termination. Asymmetries in the information requirements for political versus military topics make it exceedingly difficult to design games to explore both in a rigorous manner. Paradoxically, the deliberate falsification of facts in peacetime offers the best hope of painting a more vivid and convincing portrait of a situation that would actually confront policymakers in wartime.

Wargames featuring conflict between China, the U.S., and Taiwan have taken the Washington, D.C., area by storm in the past two years. The U.S. military has held classified wargames on the topic. The Center for Strategic and International Studies (CSIS) held 22 iterations of such a scenario, and other think tanks such as the Center for a New American Security (CNAS), CNA, and RAND have held their own wargames on the topic as well. The appeal of wargames is not hard to figure out: They provide a vivid and dynamic simulation of armed conflict. The China-Taiwan war scenario is especially appealing because the U.S. and China are locked in a rivalry and are also equipped with large, advanced, and powerful militaries. What would happen if the two fought is an inherently fascinating question. The U.S. military advantage is fading, and China’s military is growing stronger. But how the two might fare against each other in combat is unclear. Wargames offer the possibility of exploring such critically important topics, whether as part of a research design or as critical context for creative discussion.

The results of the games have generated several key findings. The most obvious and compelling lesson is that combat between U.S. and Chinese military forces would probably be immensely destructive. In the CSIS game, the U.S. lost 200 aircraft, 20 warships, and two aircraft carriers. Attacks on cyber and space infrastructure are not uncommon. U.S. missiles may strike China’s homeland. Both sides might escalate to the threat, or even use, of nuclear missiles, as happened in at least one CNAS game. Analysts have also noted the military implications for operational topics such as the importance of massing forces, adequate munition stores, and the vulnerability of surface ships on the modern battlefield.

But for many, these lessons are not enough. The frightening results of such simulations naturally raise deeper questions of a fundamentally political nature, such as: How can such a war be avoided? If it can’t be avoided, how can escalation in such a war be controlled? What can the U.S. do to deter China from attacking Taiwan? How long can Taiwan successfully hold out against such an attack? Which allies will support the U.S. in such a war? These political considerations permeate the news accounts of the wargames. In the CSIS event, for example, participants debated whether the pre-positioning of marines on Taiwan prior to war would be “too provocative” or not. Players also debated whether China should attack Japan or not. Military decisions regarding escalation also carried significant political considerations, which may or may not have been debated at the game. In one game, for example, players for the U.S. sideauthorized missile strikes on Chinese ports.

Political decisions on the initiation or escalation of war are immensely important. Yet they are also extremely difficult to answer owing in part to the dearth of reliable data. After all, a U.S.-China war remains, thankfully, completely hypothetical. That leaves virtually no firsthand information with which one can answer such questions. Wargames, and the scenarios that underpin them, have sometimes been used to explore such questions. Since they incorporate many facts about relevant combatants, wargames offer the possibility of exploring political as well as military dimensions of war through a structured, analytic method.

Yet analysts should be wary of trying to use wargames designed for military questions to analyze political questions. The analysis of political topics has fundamentally different information requirements than those for military ones. Wargames that support analysis of military decisions do not necessarily support analysis of political decisions in the same situation, and vice versa.

He further argues that efforts to game future crises by tweaking the status quo are inherently problematic:

One way to get around this problem is to incorporate as much of the current world situation as possible into the game scenario and make only those changes needed to introduce conflict. A game designer could create a scenario that depicts U.S.-China relations largely as they exist today and then inject some crisis near Taiwan to begin the war. This is, in fact, the most commonly used method to build “realistic” scenarios for wargames. But a scenario set in wartime that hews to facts as they exist in peacetime introduces a serious analytic error.

The problem is that, by definition, many factors in a peacetime situation favor peace—factors that can be numerous and diffuse. A scenario based on a contemporary, nonhostile relationship between two countries implies many incentives to avoid hostilities. A main reason why the U.S. and China have not gone to war over Taiwan, after all, is because they have many compelling reasons to favor peace. What exactly about the current situation favors peace remains in debate, but candidates include mutual economic interdependence, the presence of nuclear weapons, relatively modestthreat perceptions, and involvement in shared multilateral institutions. Injecting a “trigger event” such as a crisis related to Taiwan does not resolve the structural incentives for peace. Instead, it merely creates an artificial and unconvincing driver of war. Scenarios that aim to explore political topics in wartime but share considerable continuity with peacetime situations are thus inherently contradictory—they depict a situation with as many structural incentives for peace as one that favors war. This contradiction helps explain why so many wargame scenarios strike participants as implausible and unbelievable.

His answer is to build models of crisis escalation that build on historical examples:

A better approach to wargames would be to model the political assumptions for a hypothetical wargame on the experiences of countries that have actually gone to war. As mentioned earlier in this piece, the deliberate falsification of facts in peacetime offers a good model for what might actually happen in wartime and how policymakers would likely react. After all, the most realistic and relevant facts that confront decision-makers in a war are not those that typify situations in peacetime, but those that typify situations in wartime. The very act of envisioning a war situation that does not exist requires the imaginative visualization of a world radically different from a peacetime status quo.

For such historical data to be useful, it should be as rigorous and scientifically derived as possible. The best resource for scenario designers that aim to replicate realistic and relevant facts and incentives for political decisions lies in the historical experience of countries in analogous situations

This, of course, is what many international relations scholars do: attempt to create generalized and testable hypotheses from historical data. There are, I think, a number of challenges to this approach too. After all, generalizations are simply tendencies and not iron laws of causality, specific contexts matter, and historical analogies are often misleading because of very different circumstances.

However, good IR scholarship can offer insight into is what sort of factors (political, economic, and otherwise) might shape escalation decisions, and we can then try to model those much as we might model the factors that shape combat outcomes. Certainly the social science here is far from settled or definitive, but the mere process of constructing models forces us to make explicit our assumptions about the way the world (or an adversary) works for further discussion, research, and refinement.

In general we know that subject matter experts are not necessarily very good predictors of the future (in fact, they’re quite poor at it), in part because a tendency to be cognitively over-attached to favoured paradigms. We also know that intelligence communities often outperform other forecasters, not so much because of access to classified material (although that can be a factor) but also because recruitment tends to prioritize cognitive characteristics associated with better forecasting performance, and because a well-developed analytical process emphasizes training and methodologies to check cognitive biases while encouraging constructive challenges to assumptions and interpretations. As an academic who has worked as an intelligence analyst (and assessed the predictive accuracy of other analysts), these skills are NOT ones they teach in political science (or international relations or security studies) graduate school. My impression is that they are even less present in most PME programmes.

Heath ends his piece with an important warning about the dangers of hubris and the value of humility:

Even with such improvements, however, humility about what we can achieve is required. Re-creating hypothetical war situations based on the experiences of past wars will be imperfect at best and carry their own flawed assumptions. Carrying out different iterations with slightly different assumptions could help mitigate some of these limitations. Yet even in the most optimal case, we can at best aspire to craft a crude simulacra of the incentives and factors leaders might confront in a hypothetical situation that will carry all sorts of unimaginable complexities. Given the stakes involved, even an imperfect and partial approach offers a potentially significant improvement over current methodologies for defense planners, analysts, and decision-makers alike who seek to explore political questions in wartime.

On this I think we can all agree.

CSIS: Wargaming a Chinese invasion of Taiwan

The Center for Strategic and International Studies has issued a major report on a potential Chinese invasion of Taiwan, using a wargame specially developed for the study.

China’s leaders have become increasingly strident about unifying Taiwan with the People’s Republic
of China (PRC).1 Senior U.S. officials and civilian experts alike have expressed concern about Chinese intentions and the possibility of conflict. Although Chinese plans are unclear, a military invasion is not out of the question and would constitute China’s most dangerous solution to its “Taiwan problem”; it has therefore justly become a focus of U.S. national security discourse.

Because “a Taiwan contingency is the pacing scenario” for the U.S. military, it is critical to have a shared, rigorous, and transparent understanding of the operational dynamics of such an invasion.2 Just as such an understanding was developed concerning the Cold War’s Fulda Gap, so too must analysts consider the Taiwan invasion scenario. This understanding is important because U.S. policy would be radically different if the defense were hopeless than if successful defense were achievable. If Taiwan can defend itself from China without U.S. assistance, then there is no reason to tailor U.S. strategy

to such a contingency. At the other extreme, if no amount of U.S. assistance can save Taiwan from
a Chinese invasion, then the United States should not mount a quixotic effort to defend the island. However, if U.S. intervention can thwart an invasion under certain conditions and by relying on certain key capabilities, then U.S. policy should be shaped accordingly. In this way, China would also be more likely to be deterred from an invasion in the first place. However, such shaping of U.S. strategy requires policymakers to have a shared understanding of the problem.

Yet, there is no rigorous, open-source analysis of the operational dynamics and outcomes of an invasion despite its critical nature. Previous unclassified analyses either focus on one aspect of an invasion, are not rigorously structured, or do not focus on military operations. Classified wargames are not transparent to the public. Without a suitable analysis, public debate will remain unanchored.

Therefore, this CSIS project designed a wargame using historical data and operations research to model a Chinese amphibious invasion of Taiwan in 2026. Some rules were designed using analogies with
past military operations; for example, the Chinese amphibious lift was based on analysis of Normandy, Okinawa, and the Falklands. Other rules were based on theoretical weapons performance data, such as determining the number of ballistic missiles required to cover an airport. Most rules combined these two methods. In this way, the results of combat in the wargame were determined by analytically based rules instead of by personal judgment. The same set of rules applied to the first iteration and to the last iteration, ensuring consistency.

Based on interviews and a literature review, the project posited a “base scenario” that incorporated the most likely values for key assumptions. The project team ran that base scenario three times. A variety of excursion cases then explored the effects of varying assumptions.3 The impact of these varying assumptions on the likely outcome is depicted in a Taiwan Invasion Scorecard (see Figure 8). In all, 24 iterations of the game mapped the contours of the conflict and produced a coherent and rigorously derived picture of a major threat facing the United States.

The Results

The invasion always starts the same way: an opening bombardment destroys most of Taiwan’s navy and air force in the first hours of hostilities. Augmented by a powerful rocket force, the Chinese navy encircles Taiwan and interdicts any attempts to get ships and aircraft to the besieged island. Tens of thousands of Chinese soldiers cross the strait in a mix of military amphibious craft and civilian roll- on, roll-off ships, while air assault and airborne troops land behind the beachheads.

However, in the most likely “base scenario,” the Chinese invasion quickly founders. Despite massive Chinese bombardment, Taiwanese ground forces stream to the beachhead, where the invaders struggle to build up supplies and move inland. Meanwhile U.S. submarines, bombers, and fighter/attack aircraft, often reinforced by Japan Self-Defense Forces, rapidly cripple the Chinese amphibious fleet. China’s strikes on Japanese bases and U.S. surface ships cannot change the result: Taiwan remains autonomous.

There is one major assumption here: Taiwan must resist and not capitulate. If Taiwan surrenders before U.S. forces can be brought to bear, the rest is futile.

This defense comes at a high cost. The United States and Japan lose dozens of ships, hundreds of aircraft, and thousands of servicemembers. Such losses would damage the U.S. global position
for many years. While Taiwan’s military is unbroken, it is severely degraded and left to defend a damaged economy on an island without electricity and basic services. China also suffers heavily. Its navy is in shambles, the core of its amphibious forces is broken, and tens of thousands of soldiers are prisoners of war.

You will find the full report at the link above, including its recommendations for US,Taiwan, and allies. The launch event was livestreamed on YouTube, and can be found below. Stacie Pettyjohn (CNAS) makes a particular good point about the value of multiple organizations undertaking multiple, different games (in both the public and classified spaces) to enhance the robustness of overall findings.

For press coverage of the report, see:

For other Taiwan invasion wargames, see this summary Drew Marriott published at PAXsims in December 2021.

Smith, Ringrose, and Barker: computer vs manual wargaming

Jeremy Smith, Trevor Ringrose, and Stephen Barker have published interesting article in a forthcoming (2023) issue of The Journal of Defense Modeling and Simulation describing “an experimental intervention to investigate user perceptions of computer versus manual board wargame.”

Analysis of the literature related to wargaming identifies a requirement for the perception of immersion and engagement in wargaming. The references generally indicate that the computer is less able to facilitate collective engagement than a manual system; however, there is as yet little empirical evidence to support this. There are also suggestions that players perceive manual games differently to a computer wargame. An experiment, derived from the previous analysis, was performed to address the research question: Is there a discernible difference between the levels of players’ engagement in computer wargames versus manual wargames? The experiment provides empirical evidence that there is a difference in players’ engagement with a computer wargame compared to a manual game, in particular with the manual game providing greater engagement with other players. Hence, if engagement between players is to be encouraged and regarded as an important aspect of a wargame for defense applications, then this provides evidence that the manual approach can indeed be better.

The approach taken was to have students play two different but similar wargames—one a manual boardgame, the other a digital wargame—and then survey them about their engagement across several categories.

The two games were chosen to be as similar as possible in scale, scope, complexity, and length while team sizes were also the same in both cases and team members seated similarly closely together, with the main difference being people being individually seated at a PC in the computer case and seated round a table in the manual case. The test subjects were available and willing samples of those people taking these four courses. Each course ran each game once, with two courses running manual first and two running computer first. Questionnaire response was optional, and sometimes on different days, so that although the same people were playing both games, a paired analysis was not possible because not everybody responded to both. There is also a risk of non-response being indicative of non-engagement.

The students “varied across serving and retired military officers and other ranks, defense-related civilians, and non-defense-related civilians.” The manual game was “a very simple introductory tactical military game” developed at Cranfield University. The computer wargame used was CONTACT, “a computer-based wargame developed in the United Kingdom and used by UK MoD and several overseas military nations” used here as “a simple introduction to a computer-based military wargame, so that only a limited set of its functionality is exposed and used.”

The results showed slightly the manual game reported somewhat higher levels of personal engagement and much higher levels of engagement with others. (The “experience” row actually shows the number of respondents, 45 and 34 respectively).

The authors note some limitations to their study. Not everyone completed a questionnaire for both games and there is no way of knowing whether the “missing” responses might have systematically biased the results. They also used different games, so it is possible that the game designs were a significant factor in student evaluations. It might also be noted that while engagement is generally a desirable characteristic of all serious games, it is possible to be so engaged and so eager to win that players may potentially learns the wrong lessons—something that Anders Frank has referred to as “gamer mode.”

To date academic research on digital vs manual gameplay has largely been focused on hobby games (for example, some excellent research on the effects of automating the board game Pandemic here and here). Much of it has also been by digital enthusiasts. Smith, Ringrose, and Barker have performed a service by focusing attention on wargaming in particular.

Women’s Wargaming Network meeting, January 21

The Women’s Wargaming Network will be holding its first event of 2023 on January 21:

We would be thrilled to have you join us for our inaugural semi-annual members’ meeting on Saturday, January 21st of 2023. The meeting will be held in-person at IDA Headquarters in Potomac Yard, VA. 

Members will hear presentations and progress a progress report, financial report, and presentations from each of WWN’s committee leads. We will also share an update on fundraising and engage the group in filming a pitch video for prospective donors and WWN members. 

In vision casting for 2023 we will discuss WWN’s collaboration with RAND on Hegemony, a prospective collaboration with the U.N., and the process of building a Board of Trustees. 

We will also, of course, play a wargame together!

RSVP to let us know that you are coming, and stay tuned for additional details via email!

If you are a foreign national, please let us know that you plan to attend the event by January 17th.

If you are a US citizen, please complete the form by January 20th.

The RSVP form can be found here.

%d bloggers like this: