Taiwanese President Tsai Ing-wen met US House Speaker Kevin McCarthy in California on Wednesday. Three days later, China’s People’s Liberation Army began major military exercises around Taiwan involving aircraft, warships and simulated operations against the island. The sequence resembles Nancy Pelosi’s visit to Taiwan last August closely enough that the comparison is unavoidable, but the differences may be more politically useful than the similarities.
Last August, Pelosi travelled directly to Taipei despite repeated warnings from Beijing. China responded with unprecedented military exercises around Taiwan, including missile launches and operations that crossed boundaries previously treated more cautiously. At the time, I argued that the visit illustrated a credibility trap: Washington did not want Chinese threats to determine where American politicians could travel, while Beijing did not want its threats to appear meaningless. Both sides therefore had incentives to demonstrate resolve even though neither appeared to want a military confrontation.
This week’s meeting appears to have been designed partly around the lessons of that crisis. McCarthy had previously expressed interest in visiting Taiwan himself. Instead, Tsai met him at the Ronald Reagan Presidential Library in California while transiting through the United States. The substantive political signal remained clear: the leader of Taiwan met the Speaker of the US House of Representatives. The geography changed.
That distinction may look cosmetic. In international politics, cosmetics can carry strategic information. Diplomatic protocol, meeting locations, official titles and even which entrance a politician uses can communicate how governments want an interaction interpreted. States frequently design symbolic actions so several audiences can read different levels of commitment into the same event.
The Tsai-McCarthy meeting can therefore be understood as an attempt at calibrated signalling. Washington and Taipei maintained a high-level political interaction, demonstrating that Beijing does not possess a complete veto over US-Taiwan contacts. At the same time, McCarthy avoided travelling to Taipei, reducing one of the features that made Pelosi’s visit especially provocative from Beijing’s perspective. The message was not withdrawal and not maximum confrontation. It occupied the space between them.
China still responded with major military exercises. The People’s Liberation Army announced three days of drills around Taiwan, describing them as a warning against separatist forces and external interference. Chinese aircraft crossed the median line of the Taiwan Strait, warships operated around the island and military exercises included simulated precision strikes and operations that could contribute to rehearsing encirclement or blockade scenarios.
It would be easy to conclude from this that calibration failed. If China conducts military exercises anyway, perhaps there was no strategic benefit in moving the meeting from Taipei to California. That conclusion is too quick because escalation is not binary. The relevant comparison is not between Chinese response and no Chinese response, but between different possible levels of response.
International crises operate on gradients. A diplomatic protest is escalation relative to silence. Military exercises are escalation relative to diplomatic protest. Missile launches can represent another level. A blockade another. Direct combat another. Policy should therefore be evaluated according to whether it changes the magnitude and probability of escalation rather than whether it eliminates every hostile response.
This is where deterrence becomes more complicated than simply appearing strong. Political rhetoric frequently treats deterrence as though resolve has one direction: do something the opponent dislikes, refuse to retreat and thereby demonstrate strength. But effective deterrence communicates two messages simultaneously. The first is that certain interests will be defended. The second is that the defender does not intend to threaten interests beyond those boundaries.
Thomas Schelling’s work on deterrence emphasised that coercion depends partly on communicating limits. An opponent needs to understand not only what will produce costs but what behaviour will avoid them. A threat without an exit becomes less useful because the target has fewer incentives to comply once confrontation begins.
Taiwan therefore requires both deterrence and reassurance. Beijing needs to believe that attempting to seize Taiwan through force would create enormous military, economic and diplomatic costs. It also needs reasons to believe that maintaining the status quo will not automatically result in Taiwan becoming formally independent under an expanding American security guarantee.
Taipei faces the mirror-image problem. Taiwan needs confidence that China cannot gradually coerce it into surrendering its political autonomy. Yet Taiwanese leaders also need to avoid actions that could convince Beijing that peaceful unification has become permanently impossible and that waiting therefore only worsens China’s strategic position.
Washington operates between both. It wants China deterred from invasion while Taiwan remains cautious about formal independence. That is the logic behind America’s longstanding strategic ambiguity, although statements by President Biden have periodically created questions about how ambiguous American policy remains.
Three actors are therefore signalling to one another simultaneously, and each signal has several audiences. Tsai speaks to Taiwanese voters, Washington and Beijing. McCarthy speaks to Republicans, Congress, Taiwan and China. Xi Jinping speaks to Chinese citizens, the People’s Liberation Army, Taiwan and the United States. Domestic politics becomes embedded inside international deterrence.
This creates a multiple-audience problem. A statement sufficiently strong to reassure one audience may alarm another. If an American leader emphasises that the United States will never abandon Taiwan, Taiwanese citizens may feel safer while Chinese officials perceive movement towards a defence guarantee. If the same leader emphasises restraint to reassure Beijing, Taiwanese observers may worry that Washington is weakening.
Politicians therefore often choose symbolic ambiguity deliberately. They need words capable of carrying different meanings to different audiences without becoming outright contradictions.
This can look dishonest from the outside. Sometimes it is strategic necessity. International agreements and diplomatic formulas frequently survive precisely because different governments can interpret them slightly differently. Forcing complete semantic clarity can destroy arrangements that depend on each side retaining enough political space to avoid humiliation.
The Taiwan status quo is perhaps one of the clearest examples. Beijing maintains that Taiwan belongs to China and that unification must eventually occur. Taiwan operates as a self-governing democracy with its own institutions, military and elections. Washington recognises the People’s Republic of China diplomatically while maintaining extensive unofficial relations with Taiwan and providing defensive arms.
These positions are not intellectually harmonious. Their incompatibility has been managed through ambiguity rather than resolved through agreement. The arrangement survives because each actor has historically preferred an imperfect status quo to the risks of forcing final resolution.
That preference cannot be assumed forever.
China’s military power has grown substantially. Taiwan’s political identity has evolved. US-China strategic competition has intensified. Economic and technological rivalry increasingly affects semiconductors, investment and supply chains. Every trend changes the expected cost of waiting.
This is where dynamic deterrence matters. A status quo can remain stable today while becoming less stable tomorrow if one actor believes the balance of power is moving against it. If Beijing expects its military position to improve, waiting may appear attractive. If it expects American alliances and Taiwanese identity to make eventual unification increasingly difficult, waiting may appear costly.
The same calculation exists in Washington. If American strategists believe China’s military capabilities around Taiwan are improving faster than America’s ability to respond, they may favour strengthening deterrence sooner. Those actions can then convince Beijing that Washington is changing the status quo, accelerating the security dilemma.
Each side’s defensive preparation becomes the other’s evidence of offensive intent.
China increases military capability because it fears Taiwanese separatism and American intervention. The United States strengthens regional alliances because it fears Chinese coercion. Taiwan purchases weapons because it fears invasion. Beijing observes the weapons and alliances and concludes encirclement is increasing.
Nobody needs to want war for the probability of war to rise.
That is the classic security dilemma.
The difficulty is determining which actions are genuinely defensive because almost every military capability can be described that way by the actor possessing it. China describes exercises as responses to separatist provocation. Taiwan describes defence spending as necessary because of Chinese exercises. Washington describes regional deployments as deterrence. Beijing interprets those deployments as containment.
Intentions are private.
Capabilities are visible.
States therefore respond heavily to capabilities.
This is why diplomatic signalling remains important despite appearing weaker than military power. Diplomacy provides information about intentions that weapons cannot.
The Tsai-McCarthy meeting’s location may be one such signal. Washington can say, in effect, that high-level engagement will continue but that it understands some forms of engagement carry greater escalation risk than others. Whether Beijing interprets the distinction that way is another question.
The Chinese response suggests Beijing still wants the political cost of these meetings to remain high. That creates an incentive problem for Washington. If every US-Taiwan interaction produces military exercises, American policymakers may conclude that restraint does not buy anything and therefore stop calibrating.
China faces the opposite risk. If Beijing responds maximally to both small and large perceived provocations, it reduces its ability to signal which actions it considers especially serious. A state possessing ten levels of escalation but using level nine repeatedly makes level nine less informative.
This is a problem of signal saturation. Threats communicate information only when variation in behaviour corresponds to variation in concern. If every action is described as crossing a red line, eventually outsiders struggle to determine which red lines genuinely matter.
The United States experiences the same danger when officials use increasingly strong language over routine disputes. Credibility does not come from maximising rhetorical intensity. It comes from maintaining a relationship between words and actions.
This is why restraint can itself become a costly signal. A powerful state capable of escalating but deliberately choosing a lower level communicates that it wants to preserve limits. The difficulty is ensuring the opponent interprets restraint as intentional rather than weakness.
That ambiguity sits at the centre of almost every deterrence problem.
Domestic politics makes restraint especially difficult because politicians can be punished for appearing weak much more visibly than they are rewarded for preventing hypothetical crises. McCarthy travelling to Taiwan would produce photographs demonstrating solidarity. Choosing not to travel produces no comparable image of the confrontation that did not occur.
Prevented crises are politically invisible.
This creates systematic incentives towards over-signalling resolve.
The same mechanism operates in Beijing. Chinese leaders face nationalist audiences that have been told Taiwan is an inseparable part of China. Responding weakly to high-level US-Taiwan engagement could create domestic criticism that the government tolerates violations of national sovereignty.
Authoritarianism does not eliminate audience costs. It changes how they operate. Xi does not face competitive national elections, but regime legitimacy, elite politics, military credibility and nationalism still constrain policy.
This is one reason nationalist rhetoric can become strategically dangerous even for governments that created it. Leaders use nationalism to strengthen domestic legitimacy. The stronger the nationalist expectations become, the more politically expensive compromise becomes later.
Governments can become prisoners of their own rhetoric.
Taiwan’s democracy creates a different set of audience costs. Taiwanese politicians have to respond to voters whose political identity increasingly centres on Taiwan itself rather than a future political relationship with mainland China. A leader perceived as yielding excessively to Beijing can pay an electoral price.
This makes any negotiated settlement more difficult because concessions need domestic legitimacy on both sides.
International bargaining therefore cannot be separated from domestic institutions.
Robert Putnam described international negotiations as two-level games: leaders bargain simultaneously with foreign counterparts and domestic constituencies. An agreement feasible internationally may be impossible domestically, while a position popular domestically may make international agreement impossible.
Taiwan may be an extreme version because the fundamental sovereignty dispute touches national identity itself.
Economic interdependence complicates the situation further. China remains an enormously important economic partner for Taiwan, while Taiwanese companies have extensive commercial relationships across the strait. Taiwan also occupies a central position in advanced semiconductor manufacturing, giving instability around the island global economic consequences.
Interdependence should raise the cost of conflict.
It does not eliminate the possibility.
Europe’s relationship with Russia before Ukraine demonstrated why. Extensive trade can create incentives for peace while also creating asymmetric dependencies and political leverage. Governments can choose geopolitical objectives over economic welfare when leaders value the strategic objective sufficiently.
The Taiwan Strait therefore cannot rely on semiconductor supply chains as a permanent peace guarantee. The economic cost of war would be enormous for China, Taiwan, the United States and the global economy. That raises the threshold for conflict. It does not make the threshold infinite.
This is why deterrence ultimately needs several layers: Taiwan’s own defensive capacity, uncertainty over American intervention, economic consequences, regional alliances and diplomatic communication capable of preventing accidents from becoming deliberate escalation.
No single layer is sufficient.
Again, resilience comes through redundancy.
Taiwan’s military cannot guarantee deterrence alone because China is vastly larger. American guarantees alone could create moral hazard or increase Beijing’s threat perception. Economic sanctions alone may not deter a government that considers sovereignty existential. Diplomacy alone cannot compensate for an absence of credible defensive capability.
The objective is to create enough overlapping costs and enough communication that every actor continues preferring the status quo.
That sounds unsatisfying because the status quo resolves nothing.
Perhaps that is its purpose.
International politics sometimes treats unresolved disputes as failures waiting for courageous leaders to solve them. Some disputes are dangerous precisely because every available final resolution is unacceptable to at least one powerful actor.
In those cases, management can be an achievement.
Northern Ireland required a negotiated settlement because both communities eventually accepted institutional compromise. Taiwan currently lacks anything resembling that shared political foundation between Beijing and Taipei. Attempting to force final status could therefore destroy the ambiguity preventing war.
This does not mean ambiguity should last forever. It means timing and consent matter.
The people of Taiwan cannot simply become variables in a US-China strategic equation. More than twenty-three million people live under Taiwan’s democratic institutions, and their preferences deserve independent political weight. Likewise, understanding China’s historical and nationalist claims does not require accepting that those claims automatically override the political choices of people currently living on the island.
This is where realism and liberal principles collide. Realism tells us that China’s power and geographical proximity cannot be wished away. Liberalism tells us that citizens’ political preferences and democratic institutions should matter. Responsible policy has to operate inside both constraints.
Ignoring power produces dangerous idealism.
Ignoring consent produces dangerous cynicism.
Strategic ambiguity is partly an attempt to avoid choosing completely between them.
The current military exercises demonstrate how fragile that arrangement has become. Every high-level meeting generates pressure. Every Chinese exercise normalises a greater military presence around Taiwan. Every American response creates another precedent. Incremental actions can gradually change the status quo without anyone formally announcing that it changed.
This is the salami-slicing problem from last August.
The median line of the Taiwan Strait historically functioned as an informal buffer even though Beijing never formally recognised it. Increasing PLA crossings can make the boundary progressively less meaningful. Once repeated military activity becomes normal, Taiwan receives less warning when genuinely hostile preparations begin.
Normalisation therefore has operational consequences.
If aircraft appear every day, distinguishing exercise from attack becomes harder.
That increases the risk of miscalculation.
The same applies to naval operations. A blockade could begin with movements resembling exercises. Taiwan and the United States then face a difficult intelligence problem: respond early and risk escalating an exercise; respond late and allow a blockade to become established.
Military signalling therefore suffers from the same ambiguity that makes diplomatic signalling useful.
The difference is that military ambiguity can kill people.
This is why communication channels between Washington and Beijing remain important even when the broader relationship deteriorates. Hotlines and military-to-military contacts are not rewards for good behaviour. They are risk-management mechanisms between adversaries.
During the Cold War, the United States and Soviet Union eventually developed increasingly sophisticated mechanisms for managing crises precisely because neither trusted the other.
Communication becomes more valuable as trust falls.
Refusing dialogue to punish an opponent can therefore be strategically backwards. The moment when leaders most want to stop talking may be the moment when misunderstanding becomes most dangerous.
The Tsai-McCarthy meeting and China’s response do not mean war is imminent. Predicting imminent conflict every time the PLA exercises around Taiwan would exaggerate the evidence and eventually make genuine warnings less credible.
The more important concern is cumulative.
How many boundaries can move before the underlying equilibrium changes?
How often can military exercises occur before they become ordinary?
How explicit can American commitments become before strategic ambiguity disappears?
How much coercion can Taiwan absorb before its government concludes the status quo is already being altered against it?
Stability is not a fixed condition.
It is an equilibrium continually reproduced through expectations.
That makes seemingly symbolic diplomatic choices matter.
Meeting in California rather than Taipei may look like a minor detail compared with aircraft carriers and missiles. But diplomacy operates partly through exactly those details because states need ways to communicate restraint without surrendering substantive interests.
Last August, Pelosi’s visit demonstrated the danger of a credibility trap in which both Washington and Beijing believed backing down would reward the other’s coercion. This week’s meeting may represent an attempt to create more room between backing down and escalating.
The fact that China still responded militarily does not prove the attempt was pointless.
It tells us how narrow that room has become.
The long-term objective should not be proving which side can tolerate more provocation. Great powers can continue escalating that contest until somebody miscalculates.
The objective should be preserving enough deterrence that nobody believes force will succeed and enough reassurance that nobody believes force is becoming necessary.
Those goals sound contradictory.
That is what makes deterrence difficult.
Strength without reassurance looks like preparation for aggression.
Reassurance without strength looks like vulnerability.
Taiwan needs both.
China needs both.
The United States needs to communicate both.
The most successful signal may therefore be one that satisfies nobody completely: strong enough that allies do not feel abandoned, restrained enough that adversaries still possess an exit.
Tsai met McCarthy.
McCarthy did not travel to Taipei.
China protested and exercised militarily.
Nobody has yet crossed into direct confrontation.
That is not a solution to the Taiwan question.
For now, avoiding the need for a final solution may be part of the strategy.
