The End of a Golden Age?

Why elections are increasingly difficult to predict

Dear Readers: We’re excited to feature an essay from a distinguished UVA alumnus, David Peyton, on the challenges election prognosticators face in an unstable and fast-changing geopolitical environment.

The Editors


— American elections are becoming harder to predict.

— Part of this is likely because of the immense changes and disruptions we are seeing not only in the United States, but in the world as a whole.

— If the underlying social phenomena are less stable, then predictions become unavoidably chancier.

The difficulties of forecasting in an unstable world

Readers of the Crystal Ball have become accustomed to predictions of an extremely high level of accuracy. There were years one wondered whether the Center for Politics team could ever do any better.

Predictions in 2022 have not matched previous performances. Numerous sources have discussed the difficulties in polling. A recent phone call for Center supporters included the exceedingly unwelcome news that some universities have abandoned their neutral polling efforts for fear of negative consequences from delivering what some officeholders would consider unacceptable news.

But deeper, underlying reasons indicate that what one might call a Golden Age of political predictions has ended, not to be recovered any time soon.

Scientists must have phenomena with some underlying level of stability so as to make accurate predictions. Lunar eclipses may provide the ultimate example.

Given only so many elections — compare that paucity of evidence with the massive amounts of evidence available in economics or linguistics, for example — political scientists must dig deep into limited data sets. The Crystal Ball has excelled in this regard. Incumbent presidents running for reelection? The primary pattern is reelection; the secondary, defeat by a challenger; and very rarely, reelection by a narrower margin. The Crystal Ball correctly noted the correspondence of Presidents Obama and Wilson in this rare latter phenomenon.

But what happens when American politics loses the stability that makes accurate prediction possible? Indeed, the entire world?

— The world’s climate used to be more predictable. Now, human-induced climate change has produced volatility that will last for the lifetime of any reader — and inevitably yielding some accompanying political instability around the world as a corollary.

— The world economy likewise used to be more predictable, even with big swings in commodity markets, especially oil. Now, cryptocurrency has produced an unpredictability never before observed, going beyond previous wild swings. A firm is now supposedly worth $32 billion, but next week nothing? Who ever heard of anything like that before?

— The rise of social media has likewise proved destabilizing. Error travels several times faster than truth or fact. Social media have enabled neo-Soviet and other forms of disinformation and has enabled fringe hate groups. Facebook has announced mass layoffs, as has Elon Musk as the new owner of Twitter, and some users have been abandoning Twitter for a new open-source platform called Mastodon.

— The rise of authoritarian or strongman rule also makes world politics less predictable. Elected leaders face at least some constraints of moving against public opinion. Unconstrainted in that way, strongmen are harder to read. What will Putin do next? Or Xi? Or any other autocrat, even given a presumption that using his power to keep his power is primal? Venezuela and Turkey, for example, used to be more predictable than today. When that level of predictability will return, no one can say.

— Global public health has increasingly intruded into politics in recent years, with several episodes of virus spread and virus scares. Covid brought the world to a halt like nothing since the “Spanish” flu a century ago. No one could have predicted the timing or origin of the latest damaging virus, although the outlines of the different kinds of national responses may have been dimly visible. For a public health matter to have become a politicized partisan matter in the U.S. was something new. All one can say safely is that some other global pandemic will occur sometime in the future.

— The world already had 100 million refugees or displaced people before the staggering flood in Pakistan — the size of Virginia — displaced 30 million more. Where are these people going to go? Few are going home any time soon. The continuing “crisis” at the southern U.S. border painfully reflects the global reality. Sober forecasts about the inevitable effects of global warming project migration of people away from hotter and less habitable equatorial regions to cooler places. That’s even more refugees.

Against rising global unpredictability, some features of American politics do remain stable (lamentably, some of the worst ones):

— The chaotic presidential primary system — indeed, the nonstop campaigning — seems immovable. (Center Director Sabato proposed a much preferable regional primary system, sequencing chosen by lot, in his book A More Perfect Constitution, but there is no visible progress toward that end.)

— The Electoral College shows the value of Peter Drucker’s insight, in the 1960s, that all decisions become obsolete over time. To employ the old ship of state metaphor, this ship now has a permanent list to starboard, a built-in favoritism to Republicans, as demonstrated twice in this century. But there is no known way to amend the Constitution in such partisan circumstances. States can apportion their votes rather than give them all to the state winner, but only Maine and Nebraska do. The little-noticed proposed interstate compact, by which all participating states would give all their votes to the national popular vote winner, offers a workaround. Yet only 15 states plus D.C. have ratified it, and at 195 electoral votes it remains short of the 270 needed for it to come into force. Moreover, ratification in today’s world takes Democratic control of both houses and the governorship, and it would be subjected to legal challenge.

— Partisan gerrymandering, which dates back to Massachusetts 200 years ago, seems essentially as entrenched as ever — notwithstanding a new, neutral process in Virginia for this cycle — and the Supreme Court now refuses to adopt any remedy. Wisconsin Republicans, for example, appear to have been able to convert even a minority of votes for state legislative candidates into a permanent legislative majority. Alternate approaches such as proportional representation from statewide slates of candidates, or multimember districts, get next to no attention. At least certain state Supreme Courts — notably Pennsylvania, North Carolina, and New York — have provided some boundaries to gerrymandering, although even this is now in question.

— Huge private donations to finance TV ads and social media continue unabated. Indeed, Ken Bensinger and Alyce McFadden recently reported in the New York Times that fully $1.4 billion has been spent on just 4 races in Georgia starting in January 2020. Most major democracies feature public financing of campaigns, but even the limited form employed in the U.S. — the federal income tax checkoff, which actually worked for several cycles — has been overwhelmed by private money starting in 2008.

— The demographic sorting of Americans into Democratic- or Republican-dominated areas also appears quite stable. Of major city areas, essentially all of them are blue to some degree. Conversely, rural areas almost all tend to be red, which the exception of some rural areas that contain a lot of nonwhite voters. Suburbs/exurbs today form the only generally competitive areas. There is the old adage that political parties serve their countries well when they manage to cut across other social divisions: urban/rural, old/young, rich/poor, religious/irreligious. In their respective curious ways, each party has both wealthy and poor adherents. Otherwise, the parties tend to reinforce social divisions.

On the other side of the ledger are factors weighing against predictability:

— The parties themselves, as in other countries, normally provide bedrock predictors of interest aggregation and articulation, and policies conducted in office. Yet 2015-16 witnessed the hostile takeover of the Republican Party by an outsider that establishment leaders did not want — yet to which almost all have acquiesced. The great drama of the 2024 Republican primaries, and the lead-in, will be the extent to which the hostile takeover continues in effect.

— “Soft on Communism” used to be that the most withering accusation that a Republican candidate could wield against a Democrat. Today? Many Republicans are talking about cutting back or ending support to Ukraine against a neo-Soviet (and also neo-Tsarist) invasion. Such an about-face was inconceivable before 2016, when candidate Donald Trump openly invited Russian hacking of Democratic servers — carried out only hours later. Only 4 years before, candidate Mitt Romney, when questioned, had called Russia the biggest threat to national security, surprising some people who thought he would or should have said China instead. But it was a solid traditional Republican answer.

— Some key traditional prevailing voting patterns are losing their force. An incoming president carrying senators and congressional candidates along with him on his coattails? Not much in 2020. Midterm blowback? Not much in 2022. What pattern if any will emerge in 2024? No one can say at this point.

— Random disturbances have always presented a huge problem in forecasting. But with increased general social volatility, this factor is rising. Two years ago, Center supporters were given a state-of-the-art inside look at Wisconsin. The Kenosha disaster followed shortly thereafter. Who could have predicted that? Or what consequences it would have?

— Mass murders by gunfire — a matter of recent pain at UVA — are both stochastic in their local manifestation, as at UVA, and grimly predictable in their totality across the U.S. The deaths in Charlottesville were followed only days later by worse events in Colorado Springs and Chesapeake. Perhaps the bleakest aspect of U.S. politics has been the predictable prevalence of death by gunfire — unlike any other developed country — and the lack of political will by officeholders to overcome the militant minority opposing any effective measures. Finally in 2022 Congress passed a law most charitably called much too little, much too late. Whether any further action will be taken may be called very doubtful, but then a year ago no one could safely have ventured the proposition that Congress would do anything in 2022.

— Election time in the U.S. this year, in a way no one could have foreseen, brought together developments trending toward stability. Brazilian voters, albeit by a slim majority, decided to replace a largely irrational president with a largely rational one. The FTX cryptocurrency crash amounted to a huge punishment for reckless speculation and self-dealing. Troubles at Facebook may hold hope that the worst excesses of social media will not continue unabated indefinitely. Election deniers lost key races for Secretary of State positions across the country.

But even with some signs of damage control or self-correction, it still appears beyond the capability of even the most acute political analysts to deliver the same level of predictive performance as previously. They are simply operating in a less predictable world. As the Economist’s Tom Standage has recently wrote, “Unpredictability is the new normal. There is no getting away from it.” Still, one may remain confident that the Center for Politics will continue to produce the best available analysis and predictions for any interested citizen to access at no charge.

David Peyton, now retired from a career in policy analysis and lobbying in Washington, was a Jefferson Scholar in the UVA pioneer coed class of 1974. His soccer exploits — while decidedly below World Cup level — included winning the ACC with the 1970 team.