It’s often tempting to analyze politics as a process of promising stuff in exchange for votes, but this ancient gambit of transactionalizing democracy has some drawbacks: most obviously, delivering on even a fraction of the guaranteed amelioration—even if you possessed clues from an omniscient being about the perfect policy—would be impossible in a modern government that’s overgrown, gridlocked, and packed with conflicting ambitions and Machiavellian careerists. Ultimately, it's much cheaper to just promise that your opponent will make things worse; thus, you can restrict your political energies to polemics and takedowns, which is invariably simpler and less divisive among your coalition than manufacturing solutions would be. Also, fear is a great motivator.
And while any reward for trafficking in this stuff is probably corrosive, the most recent vintage of political fear-mongering—this hysterical doomsday prophesying, where pundits cluster around glassy tables and shout nerve-rattling ghost stories to each other about a future timeline wherein their political adversary is sworn into office, is especially deleterious—it’s like a society-eating virus that’s obliterating whatever crumbs of decency it encounters.
Conservatives are, by temperament and philosophical outlook, naturally more predisposed to be suckered by this and get panicky about social evolution potentially being ruinous—that’s partly what conservatism is all about—but even lefties are struggling to differentiate between plausible and implausible nightmares in post-Trump America, given his unique, erratic, and unprecedented character. Pretty much everyone reading this enjoys some immunity to that stuff, though: it’s low-rent pablum for the crackpots and low-decoupling partisans, right? Well, there’s a somewhat parallel (but more refined and upmarket) method of politics envenoming Substackers, and I’m unsure whether it’s wrongheaded.
Expected Value Politics
For example,
supports voting for Trump even though he believes that “many conservatives are stupid, intellectually lazy, conspiratorial, bigoted, anti-democracy, have awful views on abortion and euthanasia, and are in many ways largely motivated by ugly instincts,”1 and he isn’t a Trump fanboy, either:It sounds funny to say Trump should be jailed for attempting a coup but since he hasn’t been he should be elected president since he’s the more pro-market candidate, but this is pretty much where I am at.
Hanania’s view is that economic freedom/growth should be accorded primacy when grading candidates and that the increasingly wobbly Republican allegiance to free markets and capitalism should still outweigh other considerations. It’s not the unhinged talk about Kamala Harris being a devout communist that inferior commentators are peddling, but—like those other Trump apologetics (and the Flight 93 essay that jump-started this trend)—Hanania is looking to transcend Trump’s glaring flaws and megalomania with the (supposedly) unique gravity of a particular outcome.
It’s an intriguing style of semi-single-issue voting: it’s not like other issues are beyond his ken, but given the effects of compounding, even a minor change in economic trajectory is amplified over time. It’s a tunnel vision derived from number-crunching rather than an arbitrary fascination. But can/should everything outside of econ really be eschewed as immaterial over the long run—the irrational preoccupations of a fussbudget? To anyone minimally pliable brain, his point is graspable, even if it requires momentarily setting aside some objections:
offered a rebuttal to Hanania, but primarily does it on shared turf, rolling with the idea that the lion’s share of political issues ought to be predominated by a clutch of expected-value heavyweights. And BB admits that economic growth is crucial (though he questions whether Republicans are even the superior choice in that respect), but he assigns a nonzero value to the chance that Trump tries fabricating some excuse to retain power beyond his term or attempts to convert the presidency into a dictatorship on the fly.2 BB also claims that Trump is the inferior selection when considering animal welfare and existential risks like pandemics and AI. So, he’s kind of looking to beat Hanania at his own game: economic growth outweighs most issues, but can it really outweigh the potential of Skynet?Using similar utilitarian/EA logic, BB has also used the back of an envelope to prove that pro-lifers should relinquish their obsession with abortion and focus on policies with better ROI, because once you indulge his frenzied, ultra-casual utilitarian mathematics, the magnitude of other issues (e.g., nuclear war) dwarfs the impact from the illegality of abortion, even if you suppose that outlawing abortions translates to rescued lives.
Setting aside whether Hanania or BB is correct, their overlapping style of analysis is worth scrutinizing: must we cede the presupposition that oodles of political issues are too trivial to factor into responsible voting? Can powder kegs like abortion be excised from political discussions altogether, and should we stop fussing with the technicalities and merits of various policies and concentrate on gauging their maximum possible impact—redirect our attention to sizing up issues rather than hashing them out?
Their maneuver intertwines politics with the rationalist milieu that’s been infiltrating public discourse over the past decade. That mindset incurred some backlash during the SBF fallout and the OpenAI-Altman palace intrigue last year, but the pushback was never particularly coherent (or devastating). Even if some hardcore EA folks too readily unpack their calculator when confronted with a moral quandary, I’m sympathetic to its key components; otherwise, humans are suckers for compelling narratives and too susceptible to empathy derailing our ratiocination. But whether this hyper-rationalist paradigm should displace conventional political disagreement between citizens is a different question.
In fairness, it’s unclear how democracy could avoid malfunctioning under the current setup: voters can’t be adequately knowledgeable about the numerous issues at play, and trying to squeeze judgment across myriad topics into votes that are also driven by personality and vibes is foolish. Political platforms are just too multifaceted to profitably interpret election outcomes unless one or two issues predominate the others. Shrinking voters’ purview down to a handful of existential risks mitigates these challenges, but it also permanently severs any whisper of feedback about non-doomsday-level policy matters. You could still contest lesser matters among friends, but it’d be wastefully misappropriating energy, since nothing but a handful of enormous gambles will ever merit decisive attention in a political contest.
Trump as Pascal’s Mugger
Most people are familiar with an argument by Blaise Pascal known as Pascal’s Wager: basically, the idea is that you should believe in God irrespective of how remote the chances are that he exists because the potential consequences are so immense and being religious is only mildly inconvenient by comparison. Pascal’s Mugging is a negative variation of this: someone randomly approaches you and demands that you forfeit your wallet; they claim to be an omnipotent demon that will inflict infinite suffering upon you unless you comply. According to Pascal’s logic, you’d be impelled to cough up your wallet to any marauders who tried this. Even if you’re confident that it’s untrue, the prospect of infinite suffering overwhelms the long odds.
Some accuse AI safety as essentially amounting to a Pascal Mugging—let’s call it a quasi-mugging (since the setup doesn’t really include infinities).3 That same framework of intermixing long shot probabilities with high-stakes outcomes to override less fantastical interests applies to many other existential threats. In fact, the main driver of the insoluble hostility between Americans over Trump is disagreement over just this sort of thing—only a fraction of the anti-Trump pugnacity derives from convictions about his policy ideas; much of it relates to inconsistent readings of his impact on existential risk.
Maybe some extremists, hungering for unrest or soft-revolution, view Trump’s erraticism as an asset, but most right-wingers like Hanania downplay the potential for Trump to fulfill his promises of overturning the established order. The rest of us concede to Pascal-style thinking and maintain that even a scintilla of risk that Trump, either by appointing enough servile goons or somehow evading the dogged subversion of his own staff, succeeds in one of his many harebrained ambitions and undoes American democracy or effectuates some other disaster, is intolerable. Thus, the fractious state of the American political scene can partly be understood as voters differing over the proper reaction to the Pascal Mugging of America by Trump. Here’s part of an exchange I had in BB’s comments:
I hope that
is correct, but my own Pascal-ish reflex is to point out that being wrong in only a tiny fraction of possible worlds is sufficient to justify alarm. Another commenter argued that BB should adjust his projections because nuclear war probably wouldn’t even destroy half of the planet’s population. Again, is this what politics should be now?Hard Times
Moreover, should politics have been like this all along? There’s reason to think not. The idea that we shouldn’t expect anything so anomalous to transpire in our lifetimes is something Freddie deBoer calls “Temporal Copernicanism,” and it’s been a hot topic on Substack lately, but also effectively refuted here by Scott Alexander:
“Temporal Copernicanism”, as described, fails basic sanity checks. But we shouldn’t have even needed sanity checks as specific as these. Common sense already tells us that new apocalyptic weapons and environmental disasters were more likely to arise during the 20th century than, say, the century between 184,500 BC and 184,400 BC!
Applying this same thinking to the issue at hand, maybe Hanania’s single-minded focus on economic growth made more sense in 1900 than it does in 2024. It’s still a potent consideration, but humanity has racked up mega-scale threats that were nonexistent in the less developed world of the past and now they could easily nullify whatever nudges to growth rates can be delivered by astute policy. The future is unknown; however, with respect to the past, it’s arguable that we’re living in a high-variance era and are especially vulnerable to these catastrophic possibilities.
So, should our intra-familial feuds and boozy, late-night contretemps between politically unaligned friends all revolve around which candidate escalates the probability of an unlikely apocalypse by a (disputable) smidgen? Even during the Cold War, when the specter of nuclear annihilation was nontrivial, it’s not like everyone agreed to table the less apocalyptic policy disputes until the Berlin Wall was toppled.
Analysis
The most promising counterargument to all of this expected-value politics is that disagreements over anomalous events are the least tractable, even if they’re the most important. When political allegiances are supposed to be decided by adjustments to minuscule odds and murky guesswork, motivated reasoning can easily disguise partisanship and smuggle it into a facially neutral analysis. Political discussion is already pigheaded and vulnerable to bias. Conversations about speculative and remote possibilities, even if their outcome is theoretically weighty, could yield comparatively little benefit: the delta between expected values of a policy change must be adjusted by the odds of fruitful interactions to rightly identify the proper subject matters for political discourse.
Maybe a civilization of unbiased automata should reorient their politics towards existential risks when their society surpasses a certain technological threshold, but it’s unlikely that humans can usefully engage in such a fragile mode of analysis. It’s even arguable we should do exactly the opposite of what some Vulcan-like society ought to: abandon any goal of politicking about existential risk because the process is too easily spoiled and rededicate ourselves towards the project of deliberating the ordinary, mundane challenges of our communities.
But that doesn’t necessarily alter the thesis that people should vote on the most critical issues, just whether we should discuss them. So accepting this counterargument leaves us in a different awkward situation: viz., the focus of political debate would be wholly disconnected from the motivations for voting. Alternatively, if we connected this dialogic humility with some epistemic humility, we might conclude that we’re also just too ignorant to vote about the most important stuff in addition to being too fallible to debate it. But a political system wherein the citizenry is committed to voting specifically on unimportant issues is a strange destiny of its own.
If, however, this counterargument fails (and no superior one replaces it), we’re saddled with an equally enigmatic situation. In a world of heightened (but still remote) existential risk, adopting utilitarian reasoning could leave us preoccupied with stuff that probably won’t happen. An unbending EA-style logic doesn’t merely recommend including these matters within a broadminded analysis or suggest that they’re undervalued—it implores an oligopoly of the improbable.
A truly enlightened politics would, therefore, be one in which we deploy the entirety of our political energies to debating and voting about the intricacies of stuff that’ll probably never occur. It’d commit us to expend whatever meager supply of political acumen we have towards governing nonexistent circumstances, judiciously overseeing the happenings of possible worlds that are unlikely to actualize, forever hostage to imagined catastrophes. It’s a curious fate, but it’s not obviously wrong.
While this expected-value logic seems able to warp governance and foreclose the possibility of conveying useful information about everyday problems, it can’t be sequestered in the private sector. Existential risks are liable to generate collective action problems, and unless we construct fitting political solutions, the same utilitarian analysis that tells us to devote more brainpower toward existential risk will likewise consign us to Armageddon:
In sum, the game theory of existential risk presents exactly the kind of problem that you’d want to enlist government to solve. Here are two possible solutions: (1) accept expected-value politics but endorse libertarian policies in all other cases, so that government exits the business of managing issues that are too small to make the cut for consideration in the post-EA public square; (2) accept expected-value politics exclusively at the national level and re-localize everything else. This latter option seems ideal and sensible to me, but I like federalism and more localized politics anyway. And per BB’s argument cited above, a preference for localizing issues with smaller expected values would probably entail rejecting something like Roe v. Wade and demoting abortion to the state level, which many people probably view as disqualifying. Ultimately, the complexion of national politics would need to radically transform, but maybe we need that anyway.
See his article for links explaining why he thinks those things.
Hanania admits he’s uncertain Trump would leave office uneventfully: “[A]lthough I don’t believe Trump would try to stay for a third term in 2028, I don’t think the possibility can be completely discounted given the degree to which the Republicans have now become a cult of personality.”