Oct 24, 2022·edited Oct 24, 2022Liked by Noah Smith
I think the fundamental logic here is more or less correct, but complicating the analysis and (IMO) making it even more likely that a Chinese attack on Taiwan precipitates WW3 are a few additional factors you don't get into here.
For one, this isn't a binary game between two actors - other countries which are major or regional powers in their own right will likely have a say in whether the situation escalates (most notably Japan, but possibly also South Korea, Australia, Vietnam, etc.). Okinawa hosts a major US base that is even closer to Taiwan than Guam is, and a Chinese attack on that base would also be an act of war against Japan that would likely make it extremely likely to escalate as well.
Secondly, we live in a much richer information environment than we did in the era of Pearl Harbor - with surveillance satellites, modern signals intelligence, and the like, there's no way China would be able to keep a comprehensive attack against either Taiwan or major US bases a secret until the last second. We'd see it coming and have time to prepare and/or escalate our defensive posture, which might cause the Chinese to think better of it, but also might lead to risky gambles on their part.
Finally there's the vital economic importance of Taiwan - this isn't a relative hinterland like 1938 Czechoslovakia was, this is a country that plays a crucial role in the modern global economy. China invading it would be more akin to Saddam's invasion of Kuwait than to Hitler's pre-war conquests.
I don't think there's any Taiwan invasion scenario that isn't a massive and very dangerous gamble for the CCP. That doesn't mean Xi won't be dumb enough to try it (he doesn't strike me as an especially insightful or competent leader) but we have to hope that he decides the risk isn't worth it.
China's military pattern since the revolution is to "slice the salami" with gradual escalations. (Very unlike the Russian coup-de-main approach of all-or-nothing sudden strikes!)
The "salami" Chinese approach forfeits some surprise, but maximizes political flexibility and lets them either push on weakness or pull back from unexpected strength. Also it lets both sides have a face-saving deal.
If that pattern continues, I expect China to blockade, not invade, Taiwan. Much more controllable for Beijing.
We could still get WW3 out of that, of course: suppose America sends a supply convoy to Taiwan, and a Chinese submarine sinks some American ships...
I think it's time you interview a military expert Noah, I think you'd really benefit from it. Or even any random officer who made it to O-5 in any U.S. service.
I don't use the term pejoratively just descriptively, but the ignorance in the underlying premises here is so great that it makes the entire analysis absurd. A couple of examples:
-Most of the U.S. "bases" that would intervene in a Chinese invasion of Taiwan are carrier battle groups. China is decades away from a blue water navy or any other weapon that could reliably disable the multiple carrier battle groups that can be brought to bear on any invasion. Even if they were a peer navy, which they're far from, they could only count on winning half the encounters which still leaves plenty to eliminate what is an insanely difficult amphibious assault.
-The remaining "bases" aren't Anderson in Guam. They're bases in Japan and Korea and Australia, and even Montana and North Dakota. It's simply not credible that China would have a chance of destroying all those bases, again even if they were a peer and they're not even close. Not to mention the impact of really ensuring that they draw every U.S. ally into the war.
-There isn't the capability to launch a Pearl Harbor style "surprise attack" that would have a meaningful military impact anymore, by any nation, short of an all-out nuclear first strike. The logistics of such attacks over such vast distances are simply far too visible by everyone in the world.
And we could go on and on. But basically, think of what you would think if an Army infantry guy your age started pontificating on economics with absolutely no education or knowledge of the subject, or any attempt to talk to anyone with such expertise before said pontification. That's how this post is perceived by anyone with any military background. You're usually much more aware of what you don't know, so this is somewhat disappointing.
A related reason to be worried is that here (unlike in Ukraine), “tactical” nuclear weapons do have really obvious tactical uses, especially for China. Forces in Ukraine are spread out and fast-moving, but in an Asian theater they’re concentrated: ships and island bases are high-value, small in number, slow-moving or stationary, conventionally hard targets.
Very strong incentives to go nuclear, especially in the opening surprise attack you’re considering here.
History shows us a slightly different "conqueror's game." It goes like this:
In 1914, the Germans talked themselves into believing Britain wouldn't really join the war against them, so they could go ahead and invade France.
In 1940, the Japanese talked themselves into believing America wouldn't really shut off their oil supplies, so they could go ahead and occupy China.
In 1950, Stalin talked himself into believing the USA didn't care about Korea, so he could give his strongman in North Korea a green light to invade the South.
In 1990, Saddam Hussein talked himself into believing America wouldn't really intervene over who ruled the Persian Gulf states, so he could go ahead and occupy Kuwait.
And in 2022, Putin talked himself into believing the West wouldn't seriously invest in Ukrainian resistance, so he could go ahead and strike for Kiev...
It's, uh, kind of an alarming pattern.
So this post is lovely game theory and fun to read. But there's history suggesting we should worry about the opposite problem.
Again and again, the trouble isn't that conquerors predict democracies' intervention. It's that they don't!
So the optimal strategy for the US is to spend as much as possible now on propaganda to convince the Chinese that a loss (even if very unlikely) would be five times worse than a victory (500 to 100)? If they believe that, then war never makes sense, whether they attack US bases or not.
Scenarios not considered: the war is fought in cyberspace, and success is defined by which side can cripple the other’s economy and infrastructure quicker and more thoroughly. No need for nukes.
Another scenario, based on the current Ukrainian situation: Taiwan defeats or at least stalemates a Chinese invasion using unconventional tactics and/or superior morale, aided by US intelligence help.
The Chinese military is pretty soft though, maybe not as bad as the Russian military but not similar to an all-volunteer army like the US or Europe. While they do have an excess male problem, thanks to the one child policy, I have a hard time believing that many mothers and fathers would enthusiastically send their sons as cannon fodder. I just think that it is all talk, although if Taiwan creates an excuse for China to attack, Taiwan will be conquered fairly handily. I think the leadership of Taiwan recognizes that, and has moderated away from a strict independence position, although of course if the electorate decides differently, that could be the door opener.
I'm pretty comforted after reading the article as a Taiwanese.
In my intuition, it is impossible to believe China will ever try to attack US milatary base. But according your simple Taiwan invasion "game", it looks like if China want to invade Taiwan, a flash attack to US will be an option for China. That mean Taiwan will not be alone in the war. Yeah, I know people in US will not happy to see US get involved anothoer Vietnam war. But if there's more missiles & aircrafts point at China definitely help a lot to my country.
I think there is a sleight of hand in how "Probability of Chinese victory if China attacks U.S. bases and U.S. fights: 80%" is defined. Seems what Noah means is "80% chance of a swift conquest of Taiwan". But that wouldn't be the end of the war in the scenario where China attacked US bases. That wouldn't be Vietnam but Pearl Harbor - America was attacked first. No way they just give up and go home if Taipei falls. The initial war over Taiwan ending in Chinese victory would be more akin to the 6 months of victories that Yamamoto allegedly promised the Japanese command in 1941. The war would continue on after that and in the context of "final victory" 80% in favor of China seems way too high.
I'm not sure that the value of a victory in defense of Taiwan has the same expected value for the US as a victory does for China. I'm no expert, but my impression is that China cares much more about Taiwan than the US does.
Your description of the US payoff for defending Taiwan (#2) also describes the payoff for defending Ukraine. Shouldn't the lack of direct US involvement in that conflict tell us something about how much they value that payoff?
Sorry, but this is ludicrous: your analysis of the branch where China preemptively strikes US bases seems way off.
First, if China attacks US bases in an analog to Pearl Harbor, there is a significant chance that the US would spot the run-up to the attacks and actually win the associated battles. This isn't 1941: we have spy satellites, spies, signals intelligence, and significant air defenses. And after such an attack, you'd see at least a conventional all gloves off war between US and China, one where I'd still bet on the US.
Essentially, anything short of a bolt-from-the-blue nuclear attack against those bases would *not* be a surprise. And while the retaliation for a nuclear BFB attack would be unpredictable, in any reasonable model, the expected value of your Forbidden City tickets would quickly trend to zero.
Your whole argument depends upon China deciding a this branch is the one to pursue, and I think your cost estimates and probability distributions in this branch are way off.
I think the fundamental logic here is more or less correct, but complicating the analysis and (IMO) making it even more likely that a Chinese attack on Taiwan precipitates WW3 are a few additional factors you don't get into here.
For one, this isn't a binary game between two actors - other countries which are major or regional powers in their own right will likely have a say in whether the situation escalates (most notably Japan, but possibly also South Korea, Australia, Vietnam, etc.). Okinawa hosts a major US base that is even closer to Taiwan than Guam is, and a Chinese attack on that base would also be an act of war against Japan that would likely make it extremely likely to escalate as well.
Secondly, we live in a much richer information environment than we did in the era of Pearl Harbor - with surveillance satellites, modern signals intelligence, and the like, there's no way China would be able to keep a comprehensive attack against either Taiwan or major US bases a secret until the last second. We'd see it coming and have time to prepare and/or escalate our defensive posture, which might cause the Chinese to think better of it, but also might lead to risky gambles on their part.
Finally there's the vital economic importance of Taiwan - this isn't a relative hinterland like 1938 Czechoslovakia was, this is a country that plays a crucial role in the modern global economy. China invading it would be more akin to Saddam's invasion of Kuwait than to Hitler's pre-war conquests.
I don't think there's any Taiwan invasion scenario that isn't a massive and very dangerous gamble for the CCP. That doesn't mean Xi won't be dumb enough to try it (he doesn't strike me as an especially insightful or competent leader) but we have to hope that he decides the risk isn't worth it.
China's military pattern since the revolution is to "slice the salami" with gradual escalations. (Very unlike the Russian coup-de-main approach of all-or-nothing sudden strikes!)
The "salami" Chinese approach forfeits some surprise, but maximizes political flexibility and lets them either push on weakness or pull back from unexpected strength. Also it lets both sides have a face-saving deal.
If that pattern continues, I expect China to blockade, not invade, Taiwan. Much more controllable for Beijing.
We could still get WW3 out of that, of course: suppose America sends a supply convoy to Taiwan, and a Chinese submarine sinks some American ships...
I don’t think it’s clear China has an 80% chance of victory - I mean Russia lost to Ukraine and Russia has fought more wars recently.
Plus Taiwan would require an amphibious landing to attack.
I think it's time you interview a military expert Noah, I think you'd really benefit from it. Or even any random officer who made it to O-5 in any U.S. service.
I don't use the term pejoratively just descriptively, but the ignorance in the underlying premises here is so great that it makes the entire analysis absurd. A couple of examples:
-Most of the U.S. "bases" that would intervene in a Chinese invasion of Taiwan are carrier battle groups. China is decades away from a blue water navy or any other weapon that could reliably disable the multiple carrier battle groups that can be brought to bear on any invasion. Even if they were a peer navy, which they're far from, they could only count on winning half the encounters which still leaves plenty to eliminate what is an insanely difficult amphibious assault.
-The remaining "bases" aren't Anderson in Guam. They're bases in Japan and Korea and Australia, and even Montana and North Dakota. It's simply not credible that China would have a chance of destroying all those bases, again even if they were a peer and they're not even close. Not to mention the impact of really ensuring that they draw every U.S. ally into the war.
-There isn't the capability to launch a Pearl Harbor style "surprise attack" that would have a meaningful military impact anymore, by any nation, short of an all-out nuclear first strike. The logistics of such attacks over such vast distances are simply far too visible by everyone in the world.
And we could go on and on. But basically, think of what you would think if an Army infantry guy your age started pontificating on economics with absolutely no education or knowledge of the subject, or any attempt to talk to anyone with such expertise before said pontification. That's how this post is perceived by anyone with any military background. You're usually much more aware of what you don't know, so this is somewhat disappointing.
A related reason to be worried is that here (unlike in Ukraine), “tactical” nuclear weapons do have really obvious tactical uses, especially for China. Forces in Ukraine are spread out and fast-moving, but in an Asian theater they’re concentrated: ships and island bases are high-value, small in number, slow-moving or stationary, conventionally hard targets.
Very strong incentives to go nuclear, especially in the opening surprise attack you’re considering here.
History shows us a slightly different "conqueror's game." It goes like this:
In 1914, the Germans talked themselves into believing Britain wouldn't really join the war against them, so they could go ahead and invade France.
In 1940, the Japanese talked themselves into believing America wouldn't really shut off their oil supplies, so they could go ahead and occupy China.
In 1950, Stalin talked himself into believing the USA didn't care about Korea, so he could give his strongman in North Korea a green light to invade the South.
In 1990, Saddam Hussein talked himself into believing America wouldn't really intervene over who ruled the Persian Gulf states, so he could go ahead and occupy Kuwait.
And in 2022, Putin talked himself into believing the West wouldn't seriously invest in Ukrainian resistance, so he could go ahead and strike for Kiev...
It's, uh, kind of an alarming pattern.
So this post is lovely game theory and fun to read. But there's history suggesting we should worry about the opposite problem.
Again and again, the trouble isn't that conquerors predict democracies' intervention. It's that they don't!
So the optimal strategy for the US is to spend as much as possible now on propaganda to convince the Chinese that a loss (even if very unlikely) would be five times worse than a victory (500 to 100)? If they believe that, then war never makes sense, whether they attack US bases or not.
Scenarios not considered: the war is fought in cyberspace, and success is defined by which side can cripple the other’s economy and infrastructure quicker and more thoroughly. No need for nukes.
Another scenario, based on the current Ukrainian situation: Taiwan defeats or at least stalemates a Chinese invasion using unconventional tactics and/or superior morale, aided by US intelligence help.
The Chinese military is pretty soft though, maybe not as bad as the Russian military but not similar to an all-volunteer army like the US or Europe. While they do have an excess male problem, thanks to the one child policy, I have a hard time believing that many mothers and fathers would enthusiastically send their sons as cannon fodder. I just think that it is all talk, although if Taiwan creates an excuse for China to attack, Taiwan will be conquered fairly handily. I think the leadership of Taiwan recognizes that, and has moderated away from a strict independence position, although of course if the electorate decides differently, that could be the door opener.
I'm pretty comforted after reading the article as a Taiwanese.
In my intuition, it is impossible to believe China will ever try to attack US milatary base. But according your simple Taiwan invasion "game", it looks like if China want to invade Taiwan, a flash attack to US will be an option for China. That mean Taiwan will not be alone in the war. Yeah, I know people in US will not happy to see US get involved anothoer Vietnam war. But if there's more missiles & aircrafts point at China definitely help a lot to my country.
I'm a South Korean in South Korea...I don't want to read this article..😭😢
I think there is a sleight of hand in how "Probability of Chinese victory if China attacks U.S. bases and U.S. fights: 80%" is defined. Seems what Noah means is "80% chance of a swift conquest of Taiwan". But that wouldn't be the end of the war in the scenario where China attacked US bases. That wouldn't be Vietnam but Pearl Harbor - America was attacked first. No way they just give up and go home if Taipei falls. The initial war over Taiwan ending in Chinese victory would be more akin to the 6 months of victories that Yamamoto allegedly promised the Japanese command in 1941. The war would continue on after that and in the context of "final victory" 80% in favor of China seems way too high.
For someone on vacation you're sure keeping up a high posting pace!
I'm not sure that the value of a victory in defense of Taiwan has the same expected value for the US as a victory does for China. I'm no expert, but my impression is that China cares much more about Taiwan than the US does.
Your description of the US payoff for defending Taiwan (#2) also describes the payoff for defending Ukraine. Shouldn't the lack of direct US involvement in that conflict tell us something about how much they value that payoff?
I do like the game theory method as a framework for thinking about the issue though, regardless of its predictive power. Thanks!
Sorry, but this is ludicrous: your analysis of the branch where China preemptively strikes US bases seems way off.
First, if China attacks US bases in an analog to Pearl Harbor, there is a significant chance that the US would spot the run-up to the attacks and actually win the associated battles. This isn't 1941: we have spy satellites, spies, signals intelligence, and significant air defenses. And after such an attack, you'd see at least a conventional all gloves off war between US and China, one where I'd still bet on the US.
Essentially, anything short of a bolt-from-the-blue nuclear attack against those bases would *not* be a surprise. And while the retaliation for a nuclear BFB attack would be unpredictable, in any reasonable model, the expected value of your Forbidden City tickets would quickly trend to zero.
Your whole argument depends upon China deciding a this branch is the one to pursue, and I think your cost estimates and probability distributions in this branch are way off.