Digital authoritarianism causes extinction and nuclear miscalc

While digital authoritarianism can enhance regime durability and national power, it also introduces deep-seated vulnerabilities, eight of which are considered below. Significantly, digital authoritarians may find themselves in a state of constant contest with other regime types, trapped in cycles of overreach and backlash, and prone to strategic miscalculations that pull them into interstate conflict. The current turn to digital authoritarianism therefore also has broader implications for international peace and stability.

Brittle Legitimacy

Reliance on information control makes authoritarians brittle. Small chinks in their information control armor could have existential consequences, particularly during political or economic crises (i.e. when the regime needs to rely on control for legitimacy because it is not delivering for citizens). The information and ideas most dangerous to authoritarians include:

• the identity of opposition groups and leaders and their levels of support; 17

• technical means for subverting control of communications and surveillance technologies;18

• ideas about values that transcend state sovereignty, such as liberalism and human rights;19

• evidence that the central government is not delivering efficient outcomes;20 and

• ideas that undermine the myths and narratives used to legitimize authoritarian rule or the power of the ruling elite.21

Constant Contest

Since technologies and ideas are dynamic, the battle for information control is a constant struggle. It can never be ‘won.’ Authoritarians are therefore in a perpetual state of information warfare, inside and outside their regime, and feel perpetually insecure. This dynamic may lead authoritarian governments to assess that it is worth engaging in information or cyberattacks to discredit liberal ideas at their foreign source or to shape or disable systems that jeopardize their information control—despite real risks of conflict escalation and global pushback. Overreach and Backlash The fundamental importance of information control to authoritarians increases the likelihood of overreach, leading to cycles of backlash and reprisal. Many perceive China’s heavy-handed narrative warfare in Hong Kong and confrontational efforts to control narratives about coronavirus to be strategic missteps. For example, CCP efforts to stifle dissent by punishing online gaming company Blizzard and the National Basketball Association (NBA) arguably aided Hong Kong protester narratives;22 while CCP obfuscation about coronavirus has prompted unprecedented diplomatic rebukes from world leaders.23 Despite rising international awareness and condemnation of China’s sharp power tactics,24 China is accelerating, not muting, these behaviors.25 One explanation for this is that the CCP calculates that the risks of international backlash (and occasional overreach by its officials) are acceptable, compared with the risk of letting domestic information control falter. Impaired Feedback Mechanisms Authoritarians embrace technology to increase the legibility of their societies. But legibility requires cooperation from society. It is facilitated by an open information ecosystem, robust civil society, mechanisms of transparency, and protections for political speech.26 Conversely, information control and technology-enabled systems of surveillance and enforcement discourage accurate reporting and punish whistleblowing, while incentivizing officials to conceal failures and exaggerate successes.27 In 2007, Le Keqiang (before he became China’s premier) described China’s national income figures as “man-made” and unreliable, and noted that more objectively verifiable proxies should be preferred to official statistics collected by provinces.28 Without elections, authoritarians can also struggle to understand public sentiment, a problem highlighted by the Chinese government’s mismanagement of massive ongoing protests in Hong Kong. Party leaders wrongly assessed that the protestors’ grievances were primarily economic rather than political and that they did not enjoy broader public support.29 As Zeynep Tufekci has observed, the costs of China’s “authoritarian blindness” have been immense: a solvable issue (demands to withdraw a relatively unimportant extradition treaty) became “a bigger, durable crisis” with ongoing political consequences.30 China’s delayed reaction to coronavirus is a stark example of the authoritarian legibility and feedback problem. Local officials and hospital administrators in Wuhan suppressed information about the outbreak and punished doctor whistleblowers—depriving other provinces and the central government (not to mention international authorities) of vital signals that would have allowed swifter action to control the pandemic.31 Once authorities acknowledged the pandemic, China deployed the full weight of its digital surveillance capabilities. It was able to implement top-down lockdowns quickly; marshal its tech sector to build health apps; force citizens to download these apps; and access vast commercial holdings of personal data to cross-check compliance. However, it lacked critical bottom-up feedback systems that may have obviated the need for such draconian measures in the first place.32 Indeed, controlling for income and population size, authoritarian regimes appear to be more lethal than democracies during epidemics, arguably because of their closed information ecosystems.33 Overreliance on Technological Systems which ‘Fail Hard’ Many authoritarian governments are embracing AI-driven surveillance and control methods—from ‘smart cities’ to digital currencies, e-payment platforms and social apps. However, when AI systems fail, they tend to fail in unpredictable, often catastrophic ways. While citizens in democracies lament slow adoption of digital governance, authoritarians’ speed comes with the risk that authorities roll out unsafe or vulnerable systems.34 Imagine a critical failure of China’s social credit system—whether by accident or sabotage—which affected the integrity of records. The implications for regime stability could be significant. AI systems do not need to fail to produce problematic results. They draw insights and make predictions based on correlations in vast datasets but are not good at identifying causal mechanisms. This means that AI systems often produce outcomes which humans cannot reverse engineer or routinely evaluate. Like using asbestos to build a city, AI governance systems might produce good results in the short-term, but inconsistencies or oversights in their approaches could lead to cascading failures that humans struggle to identify, let alone rectify.35 Unintended Consequences from High-Tech Modernism Fixation by central governments on achieving targets or deploying certain technologies creates incentives for local officials to deploy “technology placebos” that do little to address underlying economic and social concerns. For example, many so-called smart city projects in authoritarian societies have failed to meet development and economic goals. They are fraught with issues such as “unclear strategic goals” (e.g. they often optimize for surveillance, not development) and “inadequate implementation.”36 This problem may be particularly pronounced for less-developed authoritarian governments which have been persuaded, for strategic reasons, to buy Chinese-exported digital surveillance tools that are not customized to local circumstances. These cities may also become locked into unstable or insecure technical architectures37 and economic dependence on China.38 Commitments to targets, and ideological fervor about technology, can also distort commercial decisions and raise unrealistic public expectations. Analysis of China’s AI industry, for example, suggests that companies are eschewing investment in basic research and focusing on quick wins in applied research.39 Additionally, China is already behind on meeting a number of its technology targets40—a lag that will likely be exacerbated by the global economic downturn following the coronavirus pandemic, and rising security fears in foreign markets about the security of Chinese technology and IP theft by its companies. From a strategic perspective, there are risks that authoritarian governments’ fixation on technology-centric strategies will lead them to overestimate what technology can in fact achieve. For example, Chinese military strategists have posited that AI could lift the ‘fog’ of war and eliminate uncertainty and confusion on the battlefield. This is an ahistorical and unlikely prediction that could inspire miscalculation.41 Russian strategists theorize about how psychological operations might subdue adversaries without a shot being fired—an approach that may overestimate what cognitive warfare can achieve, at least without being combined with other elements of national power.42 Challenges to Social Cohesion The medium- and long-term social consequences of digital authoritarianism are yet untested. Overreliance on surveillance and enforcement systems could attenuate relationships within a society, exacerbating authoritarians’ underlying low trust problems. Since they tend to reduce citizens to data inputs, these systems may deny citizens’ intrinsic desire for dignity and identity—with unexpected results.43 Information control tactics—such as flooding—can repress opposition, but long-term may exacerbate public uncertainty and decrease business confidence and trust in official information, with implications for social cohesion and economic progress.44 Dysfunctional Innovation Ecosystems Information control and state-led pushes for technology dominance risk hampering innovation. For example, to achieve Xi Jinping’s ‘Made in China 2025’ goals, the CCP is supporting high-tech monopolies, restricting international collaboration, and yoking the state and market together.45 However, monopolies are notoriously inefficient and cross-border collaboration is an important driver of innovation. Further, innovation works best under free market conditions and in open societies.46 Some analysts argue that China’s success in deploying AI applications is an exception to this rule. However, there is a risk that Chinese companies are prioritizing shortterm breakthroughs (e.g. analyzing existing datasets to find new insights) at the expense of long-term investment in basic research.47 While authoritarians may excel at developing and deploying AI applications, conceptual research is arguably the real engine of AI advancement—and something that will continue to thrive in open societies. Summary and Further Research All states face risks in the information age, but the extent to which regime type affects the relative likelihood of these risks materializing, and their magnitude, is understudied. For example, much has been written about liberal democracies’ vulnerabilities to propaganda and foreign interference via social media.48 But while information warfare against open societies is more likely, arguably it is a higher magnitude threat for authoritarians, where control of information is core to regime survival. Similarly, analysts often lament that democratic governments have been slow to digitize governance systems and craft forward-looking technology policy.49 But while digital authoritarians might outcompete democracies in the roll-out of advanced technologies, this creates new vulnerabilities and risks. Inappropriate safeguards and accidents may result in cascading failures, while heavily digitized governance systems may be susceptible to foreign attack. Regime type may also affect the relative ability of authoritarians and democracies to mitigate their information age risks. For example, a democracy can build resilience to cyber and information threats through a variety of civil society and market-based interventions. Digital authoritarians must rely on a more limited set of top-down policy tools. Ultimately, a more systematic effort to map the comparative strengths and vulnerabilities of authoritarians and democracies in the information age could help both to better understand the other’s threat perceptions and manage escalation risks. It might also highlight ways in which democracies can hold digital authoritarians’ core interests at risk, in order to deter authoritarian interference in their own digital environments.