Political violence is shocking. Everyone should be grateful for this fact, for a world in which political violence is not shocking is a world in which political violence is common. It is not by accident that most people alive today do not live in such a world. We must remember that political violence is rare today only because humanity has discovered, after long processes of trial and error, certain strategies for preventing it.
The strategies we have discovered rest on shared assumptions about the state of the world. If the world changes in particular ways, the strategies we currently depend on will lose their efficacy, and political violence may drastically increase.
The Old Tools
All successful strategies for preventing political violence necessarily fall into one of two categories.
The first category contains strategies intended to raise, from the perspective of would-be perpetrators, the costs associated with committing an act of political violence. All strategies belonging to this category take the form of a credible threat. A state that can credibly impose a cost of death or imprisonment on individuals who commit acts of political violence will prevent all individuals unwilling to pay such a cost from committing such acts. The threats made by states today are more credible than ever, given the extent of modern surveillance technologies. As a result, political violence is far less common than it was even fifty years ago.
The second category contains strategies intended to reduce the benefits accrued by perpetrators of political violence. In the United States, where I live, three such strategies have enjoyed great success:
1) Unlike other global superpowers, the U.S. has succession protocols which align the ideological commitments of its head of state with those of their potential successors.[1] As a result of this, assassination (the quintessential form of political violence) is generally ineffective at changing the direction of governmental power.
2) The American electorate is sympathetic to victims of violent acts, be they individuals or political parties. This means that political violence aimed at swaying public opinion one way will often have the opposite of its intended effect.
3) Citizens of the U.S. rarely have reason to fear imminent persecution from the government. Consequently, Americans generally are not impelled to commit acts of political violence aimed at preventing the government from harming themselves or their loved ones.
These three factors are very effective at eliminating the potential strategic gains from political violence. To see why, consider that there are people living in the U.S. who would be eager to pay the costs of political violence if doing so also meant accruing certain corresponding benefits. Thousands of Americans would gladly give up their lives to replace the president with a member of their preferred political party, or shift public opinion on a particular issue by five percentage points, or protest against a system of political apartheid. In some countries, political violence provides a mechanism to reliably execute an exchange of this sort. It is thanks to the strategies laid out above that no such mechanism exists in the United States.
If the costs of political violence in America are so high, and the corresponding benefits so low, why does anyone do it? Generally, the answer to this question is that they are an outlier in at least one dimension. They may be deeply religious, or hold extreme political views, or have relatives in foreign countries.[2] Most commonly, though, people commit acts of political violence simply because they have made an enormous miscalculation.
As a result of this dynamic, perpetrators of political violence are selected for irrationality, at least in the United States. This is, of course, a verygood thing. The more irrational a would-be assassin, the more likely they are to make a tactical error preventing them from assassinating someone. This selection effect constrains the aggregate damage caused by acts of political violence. We cannot know what would happen if every aspiring terrorist were as competent and pragmatic as the average middle-school principal, but I suspect that such a circumstance would produce very bad outcomes.
Looking Ahead
This final thought experiment is particularly relevant to the new world we are entering, one in which a large proportion of Americans ostensibly believe that A.I. will likely bring about the end of humanity. I should first note that this polling is not at all reflective of my experience; perhaps 10% of the people I talk to on a daily basis are even mildly concerned about A.I.-caused human extinction, and I live in the Bay Area. As such, I would (very roughly) place the share of the U.S. population that believes A.I. will cause human extinction within a decade at less than 1%. But I expect this number to grow, perhaps radically, and perhaps very soon.[3]
The strategies we have developed for controlling political violence were not designed for a world in which 10% of people believe viscerally that the extinction of humanity is imminent. As a result, these strategies will cease to function in such a world.
Understand that people who believe they are highly likely to die soon will have little reason to fear being killed or imprisoned.[4] Such people will gamble their lives at long odds, and even worse, they will gamble with the lives of those around them. To these people, living in a doomed world, life will be cheap.
Our systems of political succession were not designed with these people in mind. Yes, the president and vice president are closely aligned on questions of sexual politics and immigration policy; are their positions equally fungible with regard to A.I.? Perhaps not. And for these desperate people, “perhaps” will go a long way.
The labs themselves pose maybe the gravest challenge. Chimeras of the startup, the nonprofit, and the university, these companies lack governance structures designed with the possibility of political violence in mind. Their succession protocols are not even deterministic, meaning any current C-suite executive could plausibly be placed in control of a lab during an interregnum. It is inconceivable, of course, that all these executives hold identical opinions on all matters relevant to existential A.I. risk.
Unfortunately, this makes frontier lab executives particularly tempting targets for assassination. Prospective assassins will consider the situation as follows: if a power vacuum were to arise at one of these labs, who would be most likely to fill it? How appealing are their publicly available beliefs on A.I. safety compared to the existing alternative? Who is second most likely to fill the vacuum? Third? Fifteenth? When your life is cheap, it may well be worth spending on a sample from this distribution.
These frantic, terrified people will likely have some interest in cultivating public support, but this will do little to restrain them. Reacting against a state they perceive as indifferent or oblivious to their impending death, they will take any action they deem necessary to save the lives of those they love. If there are tens of millions of such people, many of them will not be lacking in competence. We are not prepared for millions of smart, competent, terrified people who feel they have nothing to lose.
Understand that it is easy to make people afraid, especially when their lives are changing rapidly. Existing strategies will be incapable of containing this new fear. Perhaps eventually, with some luck, we will develop new strategies. Expect shocks in the interim.
I.e., if the president dies, their office passes to the vice president, who was presumably selected for that job because the president wanted them as their successor.
One is reminded of Gavrilo Princip, whose decision to assassinate Archduke Franz Ferdinand was informed by his belief that he would soon perish from tuberculosis.
Political violence is shocking. Everyone should be grateful for this fact, for a world in which political violence is not shocking is a world in which political violence is common. It is not by accident that most people alive today do not live in such a world. We must remember that political violence is rare today only because humanity has discovered, after long processes of trial and error, certain strategies for preventing it.
The strategies we have discovered rest on shared assumptions about the state of the world. If the world changes in particular ways, the strategies we currently depend on will lose their efficacy, and political violence may drastically increase.
The Old Tools
All successful strategies for preventing political violence necessarily fall into one of two categories.
The first category contains strategies intended to raise, from the perspective of would-be perpetrators, the costs associated with committing an act of political violence. All strategies belonging to this category take the form of a credible threat. A state that can credibly impose a cost of death or imprisonment on individuals who commit acts of political violence will prevent all individuals unwilling to pay such a cost from committing such acts. The threats made by states today are more credible than ever, given the extent of modern surveillance technologies. As a result, political violence is far less common than it was even fifty years ago.
The second category contains strategies intended to reduce the benefits accrued by perpetrators of political violence. In the United States, where I live, three such strategies have enjoyed great success:
1) Unlike other global superpowers, the U.S. has succession protocols which align the ideological commitments of its head of state with those of their potential successors.[1] As a result of this, assassination (the quintessential form of political violence) is generally ineffective at changing the direction of governmental power.
2) The American electorate is sympathetic to victims of violent acts, be they individuals or political parties. This means that political violence aimed at swaying public opinion one way will often have the opposite of its intended effect.
3) Citizens of the U.S. rarely have reason to fear imminent persecution from the government. Consequently, Americans generally are not impelled to commit acts of political violence aimed at preventing the government from harming themselves or their loved ones.
These three factors are very effective at eliminating the potential strategic gains from political violence. To see why, consider that there are people living in the U.S. who would be eager to pay the costs of political violence if doing so also meant accruing certain corresponding benefits. Thousands of Americans would gladly give up their lives to replace the president with a member of their preferred political party, or shift public opinion on a particular issue by five percentage points, or protest against a system of political apartheid. In some countries, political violence provides a mechanism to reliably execute an exchange of this sort. It is thanks to the strategies laid out above that no such mechanism exists in the United States.
If the costs of political violence in America are so high, and the corresponding benefits so low, why does anyone do it? Generally, the answer to this question is that they are an outlier in at least one dimension. They may be deeply religious, or hold extreme political views, or have relatives in foreign countries.[2] Most commonly, though, people commit acts of political violence simply because they have made an enormous miscalculation.
A Fool’s Errand
Many perpetrators of political violence believe false things about the world. People who commit acts of political violence often fail to understand that violence will make their fellow citizens less sympathetic towards their ideology; they suffer from delusions of persecution and conspiracy; and commonly, they are insane. These false beliefs prevent the strategies laid out above from functioning properly. Specifically, they make it impossible for perpetrators to correctly assess their prospects for success, which in turn prevents them from understanding why political violence is such a bad idea.
As a result of this dynamic, perpetrators of political violence are selected for irrationality, at least in the United States. This is, of course, a very good thing. The more irrational a would-be assassin, the more likely they are to make a tactical error preventing them from assassinating someone. This selection effect constrains the aggregate damage caused by acts of political violence. We cannot know what would happen if every aspiring terrorist were as competent and pragmatic as the average middle-school principal, but I suspect that such a circumstance would produce very bad outcomes.
Looking Ahead
This final thought experiment is particularly relevant to the new world we are entering, one in which a large proportion of Americans ostensibly believe that A.I. will likely bring about the end of humanity. I should first note that this polling is not at all reflective of my experience; perhaps 10% of the people I talk to on a daily basis are even mildly concerned about A.I.-caused human extinction, and I live in the Bay Area. As such, I would (very roughly) place the share of the U.S. population that believes A.I. will cause human extinction within a decade at less than 1%. But I expect this number to grow, perhaps radically, and perhaps very soon.[3]
The strategies we have developed for controlling political violence were not designed for a world in which 10% of people believe viscerally that the extinction of humanity is imminent. As a result, these strategies will cease to function in such a world.
Understand that people who believe they are highly likely to die soon will have little reason to fear being killed or imprisoned.[4] Such people will gamble their lives at long odds, and even worse, they will gamble with the lives of those around them. To these people, living in a doomed world, life will be cheap.
Our systems of political succession were not designed with these people in mind. Yes, the president and vice president are closely aligned on questions of sexual politics and immigration policy; are their positions equally fungible with regard to A.I.? Perhaps not. And for these desperate people, “perhaps” will go a long way.
The labs themselves pose maybe the gravest challenge. Chimeras of the startup, the nonprofit, and the university, these companies lack governance structures designed with the possibility of political violence in mind. Their succession protocols are not even deterministic, meaning any current C-suite executive could plausibly be placed in control of a lab during an interregnum. It is inconceivable, of course, that all these executives hold identical opinions on all matters relevant to existential A.I. risk.
Unfortunately, this makes frontier lab executives particularly tempting targets for assassination. Prospective assassins will consider the situation as follows: if a power vacuum were to arise at one of these labs, who would be most likely to fill it? How appealing are their publicly available beliefs on A.I. safety compared to the existing alternative? Who is second most likely to fill the vacuum? Third? Fifteenth? When your life is cheap, it may well be worth spending on a sample from this distribution.
These frantic, terrified people will likely have some interest in cultivating public support, but this will do little to restrain them. Reacting against a state they perceive as indifferent or oblivious to their impending death, they will take any action they deem necessary to save the lives of those they love. If there are tens of millions of such people, many of them will not be lacking in competence. We are not prepared for millions of smart, competent, terrified people who feel they have nothing to lose.
Understand that it is easy to make people afraid, especially when their lives are changing rapidly. Existing strategies will be incapable of containing this new fear. Perhaps eventually, with some luck, we will develop new strategies. Expect shocks in the interim.
I.e., if the president dies, their office passes to the vice president, who was presumably selected for that job because the president wanted them as their successor.
If you believe, for example, that life begins at conception, or that the government is sliding into totalitarianism, or if you have family in a place which America is bombing, then the logic in (3) does not apply to you.
Of course, current members of the 1% have a natural interest in bringing this change about.
One is reminded of Gavrilo Princip, whose decision to assassinate Archduke Franz Ferdinand was informed by his belief that he would soon perish from tuberculosis.