Forecasters often use base rates to inform their probabilistic forecasts. Conceptually, applying a base rate to a topic of present-day interest is a concession to modesty: very few events or hypothetical events are completely unprecedented. Base rates also set a marker such that a probabilistic claim that dramatically departs from the base rate is an extraordinary one and demands extraordinary evidence [1].

To help appreciate this tool, let’s look at a complex use of base rates to address an actual forecasting hypothetical, in Graham Allison’s book Destined for War. For background, one of the largest geopolitical questions of concern is whether the United States and China will fight a war in the medium-term to long-term future- possibly over a Chinese invasion to control the island of Taiwan. Analysts have noted a Chinese strategy of developing A2/AD (“anti-access/area-denial”) capabilities, such as missiles that can destroy American aircraft carriers and are tailored to evade the US Navy’s defensive systems. Over the last ten years the US has become more hostile to China economically, and- whether intentionally or not- leading politicians such as former President Biden have openly suggested the US would help defend Taiwan, a commitment that the US has avoided officially making. War games regarding a US-China conflict in littoral East Asia have become common [2].

Destined for War was a book-length effort to apply base rates to this question. Allison abstracts away what might be considered useful specifics of the case, such as Taiwan’s sovereignty and historical US antipathy towards “communism” [3], and distills it into a structural dynamic in which there is one dominant power (the US) and one rising power (China). Then he invokes The History of the Peloponnesian War as follows (page 29):

While other observers emphasized proximate causes, Thucydides goes to the heart of the matter. “As to the reasons why Sparta and Athens broke the truce,” he writes, “I propose first to give an account of the causes of complaint which they had against each other and of the specific instances where their interests clashed.” But, he warns, “the real reason for the war is most likely to be obscured by such arguments.”
Beneath these contributing factors lies a more fundamental cause, and he focuses his spotlight on it. What made war “inevitable,” Thucydides tells us, “was the rise of Athens and the fear that this instilled in Sparta.”

The book points out some commonalities within this comparison class. One is that the incumbent power has often become accustomed to status-quo policy standards, like the UK’s Two-Power Standard calling for as many battleships as the second and third largest navies in the world combined, or the US fearing the USSR (or now China) surpassing it in economic size. A policy of choking the challenger’s rise can become the only way to satisfy these standards, especially material ones; consider containment of the USSR. A second commonality is that the challenger often is motivated by a desire of importance and control of their country’s destiny. Destined for War also notes that friendliness between two countries is often insufficient to guarantee peace; the monarchs of the UK and Germany were family, and of course Sparta and Athens had relatively recently acted together to defeat Persian invasions of Greece.

Allison thus places the US-China relationship in this comparison class of “international systems featuring one dominant, incumbent power and one rising newcomer”. He then pores through history and identifies sixteen cases that he believes meet this description, and therefore comparable in this way to US-China relations. The hook of his argument is that of the sixteen historical examples of “Thucydides’ Trap”, twelve of them resulted in war- suggesting that even before one looks at particular circumstances of the US-China relationship [4], war would seem to be likely.

Base rate #1: 12/16 historical incidences of Thucydides’ Trap led to war.

So are the chances of war between the US and China 75%? Allison himself wisely cannot help but notice the cases he has identified usually result in a great power war but the Cold War did not, which for various reasons should be considered at least somewhat more relevant than cases dating from medieval Europe. Among these is the unique presence of nuclear weapons in this case: “nuclear weapons have no precedent” (page 206, emphasis in the original).

Base rate #2: 0/2 historical incidences of Thucydides’ Trap in the nuclear era led to war [5].

So we see how (1) base rates are an art; (2) they are subjective, and people who prefer to arrive at particular conclusions can advocate for various comparison classes where the resulting base rates support their position; (3) we can at least shift the topic of debate to evaluating competing comparison classes rather than the more immediate, and probably more ideologically charged, circumstance.

But within the framework of base rates, we can keep thinking about what are the crucial elements in the US-China case, and see how that changes our comparison class. Allison spends a good deal of time on another particular case: the rise of the United States itself, and in particular its tensions with Britain regarding influence in the Western Hemisphere. Here, he actually finds reasons to discount the lack of war: Britain had its hands full worrying about a rising Germany and other European threats, and was not able to prioritize containing the United States. It’s also notable that while Britain was certainly proud of its empire, it did not have to worry about a US invasion of its home islands [6].

Base rate #3: 5/6 historical incidences of Thucydides’ Trap under what I’ll call “no higher stakes” for the world’s superpowers led to war [7].

How else could we filter Allison’s reference class? One obvious way is that all these cases include Western powers, which may have particular ideas about power and the state; for that matter, the earliest case was Portugal vs. Spain in the 15^th century.

Base rate #4: 3/3 historical incidences of Thucydides’ Trap between a Western and a non-Western power led to war [8].

Another is that the US and China are separated by an ocean; this is another way in which the UK-US and US-USSR cases, which did not result in war, are more applicable. Note that the particulars of the US-China relationship often focus on the US’s ability to project power across the Pacific Ocean.

Base rate #5: 1/3 historical incidences of Thucydides’ Trap requiring projection of power across an ocean led to war [9].

We could also imagine working on ways to augment the number of “Thucydides’ traps”. The antebellum United States, for example, could be modeled as a dominant South and a rising North; certainly, if one imagines extending tensions over slavery deeper into the industrial era, the balance of power between the two regions would have eventually become colossally skewed. Carthage and Rome, or Byzantium and the Ottoman Empire, seem to at least rhyme with the phenomenon- but be careful not to only look for Thucydides’ traps where we know a war eventually occurred!

There might be still more cases which slipped his group’s analysis because they didn’t occur in the West, and so there is less scholarship about them (is there really no example in Chinese history, for example, of a dominant incumbent considering how to respond to a challenger?) What about in business? Surely there was a time when Microsoft could have been considered a dominant power and Apple a rising one, or Detroit’s Big Three versus the Japanese automakers, or Blockbuster versus Netflix (followed by broadcast/cable TV versus Netflix and its imitators). Can we accumulate more cases by defining “war” in business as price cuts or negative advertising? Are there other fields- cultural, for example- that we can turn to for base rates?

[1] See here for more discussion

[2] See here for an extensive review of one scenario, as well as helpful background on the topic: https://www.csis.org/analysis/first-battle-next-war-wargaming-chinese-invasion-taiwan

[3] Destined for War was published in 2017; since that time, support for restricting trade with China has become more bipartisan in the US, an “AI race” between the two countries has emerged, and a spectrum of accusations have been leveled by Americans against China for its government’s neglect or complicity in the origins of the COVID-19 virus.

[4] Note that many observers would assume US-Athens and China-Sparta links; this structural approach completely ignores the factors behind this intuition.

[5] There were some number of “close calls” during the Cold War, such as the Cuban Missile Crisis. Allison chooses to treat post-unification Germany as a rising power in Europe, against incumbents UK and France.

[6] The book also speculates that Britain failed to intervene in the American Civil War, and by the time they recognized the US as a rising power they lacked effective options for containment.

[7] France vs. Habsburgs (war); France vs. Britain (war); Britain vs. France (war); Britain vs. Germany (war); Britain vs. Germany (war); US vs. USSR (no war).

[8] Hapsburgs vs. Ottoman Empire (war); China/Russia vs. Japan (war); US vs. Japan (war).

[9] Britain vs. US (no war); US vs. Japan (war); US vs. USSR (no war). This would rise to 2/4 if you also include the events leading up to the Russo-Japanese war, on the grounds that Russia did actually have to deploy a large military force from eastern Europe to East Asia.