I’m not sure what to make of Jonathan Bernstein’s latest post in which he argues that candidate quality doesn’t really matter, particularly in presidential elections.To make his case, he relies heavily on research from Adam Bonica and Jake Grumbach, but their study is concerned with ideology rather than quality. It’s still interesting research. For example, they find that moderate candidates perform better than more progressive or conservative candidates, but that the advantage is almost imperceptibly small. This contrasts with the story the folks at the centrist Split Ticket are selling, which argues that progressives have a big disadvantage and are hurting the Democrats’ electoral chances.

I think where Bonica and Grumbach agree with Bernstein is that where a candidate puts his or herself on the ideological spectrum is far less predictive of political success or failure than outside factors like the state of the economy or the overall zeitgeist of a particular election season. But I don’t think this supports the idea that “candidates just aren’t all that important to general election outcomes.” It possibly supports the opposite conclusion.

This could be a case where it’s just very hard to tease out what you’re looking for in any kind of rigorously scientific way. In the most basic sense, we want to find cases where a politician won an election that the vast majority of candidates from their same party would have lost. And then we want to figure out why they were successful so this can be distilled and transferred to other races in the future.

If you don’t keep the focus this narrow, you wind up with silly or contentious conclusions. For example, if a progressive candidate wins an election by ten points that a centrist would have won by fifteen, does that matter at all? If you’re a progressive, you’re happy even if there might be coattail effects that hurt other candidates on the ticket. If you’re a moderate, you feel the opposite.

But the name of the game is capturing difficult-to-win districts in order to make or maintain majorities. Or, if we’re talking about a presidential election, the goal is to outperform the election cycle, to win a losable election, and to boost the ticket overall.

The truth is, there could be election cycles where being in the moderate lane is a big advantage, and others where it’s a big disadvantage. What the data demonstrate, however, is that it generally is far less important than other factors.

Still, even if the ideology of candidates doesn’t much matter, that doesn’t mean the candidates themselves don’t matter. If the variables that control election outcomes are largely outside of the candidates’ control, that means selecting candidates is all the more important. After all, we’re talking about politicians who greatly outperform what’s expected. The question is not whether these politicians are possible, because they obviously exist in every election cycle. The question is whether we can benefit from researching them and transfer their success to others.

And maybe we can’t. Maybe you can’t take the charisma of Bill Clinton and put it in Al Gore’s body. You can take the organic excitement of Barack Obama’s campaign and put it into Hillary Clinton’s political organization. J.D. Vance might be completely incapable of holding onto whatever it is that binds so many people to Donald Trump.

If the things that make some politicians defy gravity and outperform expectations are largely intangible and non-transferrable, then it could be that attempts to study them to get a political advantage are bound to fail. But you can select for intangibles like charisma. This tells you not to choose Bob Dole, Al Gore or Mitt Romney as your nominee. This isn’t because of their record or their policy positions or the talent of their staffs, but simply because they quite obviously don’t have any kind of ‘it’ factor that excites people.

I don’t think charismatic politicians grow on trees, so I don’t think this is a winning strategy for a political party, although the folks in charge of candidate recruitment should take these intangible qualities into consideration. As a general matter, I think excitement is a key ingredient, and that moderates are the least likely people to create it. But there’s an age-old debate over whether to focus on mobilizing the base with red meat or to fight over persuadable voters in the middle.

I believe there isn’t a correct answer to that question that covers all races and all election cycles. What all the researchers seem to agree on is that there is at least a small, if tiny, advantage to being a moderate. If that’s true, I suspect it’s because of the examples where candidates are simply so extreme that they lose races that a generic candidate would have won. In my recollection, this is way more common on the right than on the left. But I suppose there’s probably a political pendulum on that phenomenon, too.

If I had to make a rule of thumb, it would be that the ideal candidate has some magnetic qualities and some out-of-the-mainstream ideas that worry centrists, but that they aren’t so fucking crazy as to scare off normal people. Oh, and incumbents should have excellent constituent service because in many election cycles, being an incumbent is actually a big disadvantage.

Put another way, being a moderate or centrist works fine if your party is having a good election cycle anyway, but it’s also boring. The stats may show it pays off in the end, but I think that’s just because people who take chances have a greater chance of failure.