November 7, 2023
Nate Silver, the other day, tweeted (or “Xed?”) that the Democratic Party should be very very worried.
“You have the whole electorate basically screaming ‘BIDEN’S TOO OLD.’ There’s a year’s worth of campaign to go, very likely some reversion to the fundamentals, Trump’s legal issues probably a larger liability than they seem now. But still, Democrats can’t say they weren’t warned.”
Nate Silver is as good as it gets when it comes to Bayesian forecasting. But eight years ago, in mid-2015, he had this to say about a candidate that was leading the GOP primary field prior to the first primaries:
“Trump’s campaign will fail by one means or another. Like Cain, Bachmann and Gingrich, Buchanan, Huckabee and Forbes came nowhere close to winning the Republican nomination. …If you want absurd specificity, I recently estimated Trump’s chance of becoming the GOP nominee at 2 percent. How did I get there? By considering the gantlet he’ll face over the next 11 months — Donald Trump’s Six Stages of Doom….”
I’ve mentioned this example before, and I mention it in my forthcoming book Fatal Certainty: How a Cult of Prediction Made the Twenty-First Century an Era of Strategic Shock – and How Rigorous Imagination Could Bring Us Back. To his immense credit, Silver has not scrubbed the ABC/Five Thirty Eight site that contained that essay; indeed, he has addressed it forthrightly in many forums. As an honest man, he has sought to examine his errors (in the sense of “scientific trial and error”) and to take them into account in his future attempts at prognostication.
Silver’s post-election explanation, which is fairly convincing: “The New York Times…suggested that …‘For a candidate who once seemed like an electoral phenomenon, with an unshakable following and a celebrity appeal that crossed party lines, Mr. Trump now faces the possibility that his missteps have erected a ceiling over his support….’ The flaw in the analysis is that by this logic, Clinton also had a ceiling.”
So there was a basic flaw in Silver’s model of the entire 2016 election, which he appreciated only ex post facto.
My question, as a self-nominated scourge of the Cult of Prediction, of which (notwithstanding his admirable honesty and integrity), Nate Silver is one of the leading gurus, would be: How might Nate’s model of the 2024 presidential election be similarly fundamentally flawed?
Off the top of my head, I can think of a number of scenarios that would throw his Trump vs. Biden, current-status-quo model far out of whack:
- Biden is not the nominee, for health or other reasons (Hunter indictment? Actual corruption found? Rebellion of Democrats?)
- Trump is not the nominee, for any number of reasons (Health? Criminal conviction? Rise of a suddenly popular alternative?)
- National emergency of some kind (Terrorist incident? War? Serious resurgence of COVID?)
- Significant increase in AI-driven disinformation and deepfakes just before the election?
- Economic boom wipes out any chance of Biden losing?
- Trump age becomes an issue?
- Six Stages of Doom actually applies to Trump this time, as his loud but shrinking base is overwhelmed by an electorate that, when it finally pays attention to the election (as it proverbially never does until it is actually upon them), decides, as it seems to have done in 2020, that Trump is too extreme?
- Some other trend reverses in a different way, guaranteeing an easy win for Trump (Economy tanks? Terror attack blamed on Biden? China invades Taiwan? Russia conquers Ukraine despite heavy spending by the U.S.?)
- And last but hardly least: Biden drops out… and his substitute loses by a landslide.
The point here is, predictive models will predict just fine – until they don’t anymore. Then and only then will the Nates of the world be able to explain why their model was flawed. That’s the only way prediction works – by correcting based on past experience. And by then it is often too late. All data is about the past; there is no data about the future. Prediction is seductive, fun… and the wrong thing to be doing, in most important cases.
The journalist James Fallows noted the futility of prediction on CBS Sunday Morning this past weekend: “Given the unknowability of elections, we’re always surprised. Maybe it means that we collectively should spend less time trying to predict what’s going to turn out, because we are bad at it. You know, bookies in the sports world have to pay off, and we [journalists] don’t really have to pay off that often. This is basically an impossible task of predicting exactly how people are going to vote. Maybe this means we recognize its impossibility and spend more times on other things than forecasting.”
And I personally would recommend that among the “other things than forecasting” that journalists should be giving us should be the actual potential ramifications of Biden getting re-elected, or Trump winning, or the other plausible outcomes I mentioned above.
Back in 2016 and 2020, there were very few deeply thought-out analyses of what a Clinton, Trump, or Biden victory would actually mean for the nation and the world. Even now, almost the only analyses of, e.g., a potential Trump victory, have been about nightmare scenarios (written by his opponents) in which he guts the Executive branch and installs loyalists in previously nonpartisan positions. The only people who appear to be doing extensive thinking about the consequences of various outcomes appear to be the extreme partisans of both sides, whose judgment clearly is questionable.
Journalists, unfortunately, even after the past two elections, still seem to be all about the horserace. Even the PBS NewsHour is all about the horserace – “Politics Monday” often seems to be completely about counting votes, polling, and inside-baseball, short-term tactics.
Their recent coverage of the Republicans’ struggles to choose a Speaker of the House almost never ventured beyond “The Math,” as both panel members called it – who might get enough votes to win the immediate race. There was brief mention of the unprecedented nature of the intraparty squabble; there was, as far as I could see, almost no attempt to step back and give more than a one-dimensional journalistic appreciation of the impacts of this imbroglio.
Ukraine funding was mentioned. The debt ceiling was mentioned. The longer-term damage to American interests resulting from this disarray of our government was hardly touched upon. The broader, longer-term, absolutely unprecedented fact of increasing power wielded by the most extreme members of the GOP caucus over, not only that caucus, but the entire House, the entire country, and the whole world – this was not conveyed.
“Politics Monday,” to me, increasingly begs the question: What are politics FOR? MUST journalists confine themselves to “The Math?” If they do not provide the broader, longer-term context, WHO WILL? Are we all blithely to follow “The Math” straight down the tubes, in the name of “objectivity?”
James Fallows, by the way, responded to Nate Silver’s worries about Biden’s age on Twix, or Xter, or whatever it is these days, with the following:
“Yeah, and he [i.e., Biden] is likely to be running against DONALD [91 felony counts] TRUMP. Who has a very loyal following. As Barry Goldwater, George Wallace, and (different base) George McGovern all did. …C’mon.”
Fallows may be right that Trump is doomed. And Nate might be right that Biden is doomed. But they are both basing their judgments on the past.
Better if journalists, and we all, admit, as Fallows said in his CBS Sunday Morning piece, that we are terrible at predicting.
Hell, even BOOKIES are terrible at predicting. I actually have a friend who is a bookie, and I think it is worth pointing out that he actually doesn’t make his money by predicting; he makes his money off people – his customers – vainly trying to predict.
Better that journalists start having a broader conception of their jobs than just “The Math.” Rigorous imagination is something we should require of our press corps.
