160 million Americans are eligible to be president. How did we end up with these two dudes again?

My head says it's irrational to talk about replacing Biden. But my gut won't stop asking questions.

Sometime this spring, I got dinner with a friend in Miami and the conversation turned to the presidential election. This was right about the time that Ron DeSantis’s campaign was manifestly starting to flounder. I told him that the nominees were probably going to be Donald Trump and Joe Biden again.

My friend, a sharp guy but not someone who intensively follows politics, was taken aback. I don’t want to put words in his mouth, so I’ll just report my impression of what he was feeling. Really? They’re just going to run it back? In a country where 160 million people¹ are eligible to be president, they’re just going to nominate these two guys again?

Trump and Biden are unpopular — if you’re a regular reader of this newsletter, you probably don’t need me to tell you that.² But if they run against each other then one of them has to win — barring an unlikely third-party victory. A New York Times/Siena College poll this weekend suggested that Trump might have the edge. In that poll, Trump led by 4 points across six swing states.

Although this result isn’t all that newsworthy unto itself — we’re a year away from the election, and many other polls have already shown Biden trailing or in a tight race — the Times did something clever by also asking voters also asking voters how they’d vote if there were different candidates in place:

• In a matchup between Trump and an unnamed, different “Democratic candidate”, the Democrat led by 8 points instead of Biden trailing by 4.

• Meanwhile, in a matchup between Biden and an unnamed, different “Republican candidate”, the Republican led by 16 points instead of Trump’s 4.

Here’s where I’m supposed to play the role of the responsible psephologist and tell you that a generic candidate isn’t really a thing. You can project your ideal version of “a Democrat” or “a Republican” all you want. It almost certainly won’t match the flawed model you’ll actually wind up with, especially after a year’s worth of this candidate being maligned by the opposing campaign.

When the Times tested some named opponents instead, the results were more mixed. Although Nikki Haley led Biden by 8 points, exceeding Trump’s margin, DeSantis had only a 1-point edge on Biden — actually less than Trump. Meanwhile, Kamala Harris trailed Trump by 3 points. That’s much worse than the generic Democrat, although it’s slightly better than Biden — significant since the last time I checked in on this, I literally couldn’t find a single poll showing Harris outperforming Biden. Whether that means the Times poll is a fluke or a part of a new trend, we’ll have to see.

What type of voters are more into a theoretical “other” Democrat than Biden? Mostly it’s younger voters. Generic Democrat leads Trump by 21 points among voters aged 18-29 — and Harris leads by 9 points — versus just a 1-point lead for Biden. This isn’t reciprocated for Generic Republican, who does better than Trump across the board but not in any particular age-related pattern:

Although you can speculate on why Biden’s numbers are so mediocre with younger Americans, one explanation is staring us in the face, which is that 78 percent of under-30 voters in the poll say Biden is “just too old to be an effective president” (48 percent say the same of Trump). And of course, voter concern about Biden’s is a pattern that has manifested itself across a lot of polls. It’s not an issue that will go away; Biden is 80 and will turn 81 later this month. He’s a big historical outlier; Biden was older at the start of his term than any prior president had been at the end of their presidency.

You can ask why there’s such a large gap in voter perceptions of Biden and Trump, who is 77. A three-and-a-half-year age difference actually isn’t trivial actuarily speaking. But obviously some of this is subjective as well. The media tends to focus on Biden’s verbal stumbles more than Trump’s, and Biden generally projects a more old-timey affect.

I also wonder, though, about the extent to which Biden feels like a little bit of a forced choice, at least to younger voters — that his age is a proxy for being a stale, default option. The Democratic nomination in 2020 was weird, with a wide-open field consolidating itself very quickly around Biden after South Carolina in the face of a looming pandemic. This year, there hasn’t really been a race at all, with RFK Jr. taking positions on many issues that are closer to Trump than Biden and the only other options being Marianne Williamson and now the relatively obscure congressman Dean Phillips.

Isn’t it too early to be looking at polls in the first place? I actually think there’s a lot of truth to that. There are plenty of people like my friend who aren’t thinking about the campaign much. Major issues like the legal jeopardy that Trump faces might not be fully priced in to voters’ evaluation of the candidates. Only 38 percent of voters in the Times poll thought Trump will be convicted for his actions after the 2020 election. However, prediction markets think it’s more likely than not that Trump will be convicted on at least one charge before November.

Plus more generally — again, the responsible psephologist in me is insisting that I tell you this — if you look at historical head-to-head polling a year ahead of the election, you’ll find it has little if any correlation with the eventual results.

But the political scientist Dan Hopkins raises some good counterarguments in this Twitter/X thread. In particular, given that both Biden and Trump are extremely well-known, you might tend to take the polling more at face value. Moreover, Biden’s approval rating is only 39 percent, lower than any president who later managed to be re-elected. Is it really so surprising that he’s struggling against Trump?

And on the specific question of Biden’s age, voters are sending about as clear a message as they can. To explain it away or to say “that’s not fair, Trump is almost as old too!” reeks of the midwit meme to me.

Plus, if we’re not looking at polls, then how else are we supposed to forecast next year’s race? The stock answer among us data-nerd types is that you’re supposed to look at the “fundamentals”, usually meaning some combination of incumbency and economic performance. Although my gut says that Biden is in trouble, I learned a long time ago that my gut isn’t particularly trustworthy when it comes to electoral politics. Responsible Psephologist Nate, considering those models, says Biden is at least a toss-up against Trump if not a modest favorite, although that leaves open the question of whether a different Democrat would be a bigger favorite.

Still, those fundamentals-based models have their issues too:

• They haven’t actually performed very well when making predictions out of sample (as opposed to being backtested). That’s because a lot of them are poorly-designed, with results that were p-hacked on a small sample of elections.

• The incumbency advantage baked into these models is probably decreasing. The last five incumbent presidents (both Bushes, Bill Clinton, Obama and Trump) won the popular vote by an average of only 1 percentage point.

• The question of how to specify “the economy” is deeply ambiguous, especially at a time when subjective voter perceptions about the economy don’t seem to line up with the hard data.

• If Trump has some sort of durable Electoral College edge — actually an open question and not a given — then a model that predicted Biden to win the popular vote wouldn’t necessarily imply that he’d stay in office.

• Most importantly, I’d argue that both Biden’s age and the possibility of Trump’s criminal convictions are situations that a model isn’t well-equipped to handle since they have no historic precedent. Maybe if we had 1000s of historical presidential elections, we could say “ahh, see, candidates pay a 1-point penalty for every year of age beyond 78, so that’s why Biden lost by 1 point instead of winning by 2, it’s right there in the data!”. We don’t have that, so we’re just left to speculate — but just because something is hard to quantify doesn’t mean you should treat the impact as zero.

Six weeks ago, when I considered whether Democrats would be better off nominating someone other than Biden, my conclusion was complicated:

• With medium confidence, I think the risks of a serious primary challenge to Biden at this point in time would outweigh the benefits for Democrats.

• With low confidence, I think the risks of Biden volunteering not to run for a second would also outweigh the benefits for Democrats, but this is closer.

• With low confidence, and taking full advantage of hindsight bias, I think Democrats probably would have been better off if Biden had announced 6-12 months ago that he wouldn’t seek a second term.

I don’t know that my views on this question should change all that much on the basis of six more weeks of poor polling for Biden — plus, now it’s even later for someone else (especially someone other than Harris) to run in his place. What I would say, though, is that my head and gut have begun to diverge. Responsible Psephologist Nate says that it’s at least six months too late to have this conversation — that the most likely outcome of a serious primary challenge to Biden would be Biden winning the nomination anyway but being even worse for wear, or perhaps stepping down and endorsing Harris. You can’t call down to the bullpen and retrieve Gretchen Whitmer ready to throw 99-mph heat against Trump without voters asking what the hell happened to Harris or whether Whitmer is even in game shape.

But my gut — which I must reiterate, you should not trust — is with my friend in Miami. It’s saying this: Voters are telling us in every possible way that they’d really like someone other than these two guys to be on the ballot next year. And since this is a democracy, maybe somebody should listen?

1. There are about 290 million Americans who were born in the United States, of whom 55 percent are aged 35 or older. Ignoring the nuance of what it means to be a “natural born citizen”, that gets you to about 160 million people.

2. Although this might seem like an inevitable feature of the modern, polarized political environment, it didn’t always used to be this way; as recently as 2008, both John McCain and Barack Obama were well-liked by voters.