Barbara O’Brien quoted me in support of an argument that I wouldn’t have made in precisely the same way, so I thought I’d offer a brief counterpoint. Her thesis is that no matter who the Democrats nominate for president, that person will be a pariah on the right by the time the right-wing media Wurlitzer gets done with them. For that reason, the Democratic voters should not be concerned about electability because no candidate will have any electability whatsoever.

She uses this to advocate for voting with your heart and for the candidate that best matches your policy preferences. If you can get excited about a candidate, it’s likely that others will feel the same way, and ultimately a little genuine enthusiasm for the candidate is going to go a lot further than any illusory crossover appeal.

I just want to push back on this a little, because she’s not wrong about how Republicans will behave or even about how they will feel.

The first thing I want to address is a logical fallacy in her argument. It’s simply not true that people don’t change which party’s candidate they support for president from one election cycle to the next. We can get into a debate about what makes a person a Democrat or a Republican, but Hillary Clinton lost Pennsylvania because a lot of people who voted for Barack Obama decided to cast a vote for Donald Trump. In fact, I’d describe this as a ton of people. There were so many, in fact, that it overwhelmed a huge number of people who voted for Clinton after having voted for Mitt Romney.

Not only do these people exist, but they can easily decide the outcome of the election.

I’d argue that these crossover voters come in three main flavors. The first flavor are people who respond to events. They were mad about Watergate so they gave Jimmy Carter a chance but then were upset about interest rates, stagflation, and the Iranian hostage crisis and voted for Reagan. They didn’t like how Bill Clinton conducted himself, so they voted for Dubya. They didn’t like the Iraq War so they voted for Kerry and Obama. Related to these folks, are people who are just antiestablishment by nature, So, the second flavor are folks who just generally vote against the party in power, on the theory that change is a better risk than the status quo. If you’re a candidate challenging an incumbent, you don’t have to do a whole lot to win over either of these flavors of voters. As long as you don’t scare them more than what they’ve already got, you’re going to get their support.

The third flavor are people who don’t really care about policy at all, or really even performance. They respond to the person. Do they like them? Do they trust them? Do they think they are pretty or handsome or that they look like a president should? Basically, they respond to charisma. You’ll often hear people say that the taller candidate almost always wins, or that the more likable candidate always wins, and to the extent that this is true, it’s because of this charisma factor.

It’s hard to define charisma, but I’m sure anyone who has been in the same room with both Bill Clinton and Bob Dole could tell you who had more of it. Barack Obama and Bill Clinton’s charisma is off the charts. Whatever their charms, you cannot say the same about Al Gore, John Kerry and Hillary Clinton.

Now, one legitimate critique about electability is that it’s hard to predict, especially a half a year or more before an election, which is when we choose the nominee. But it wasn’t hard to predict that Bill Clinton would have more charisma than Paul Tsongas or that Barack Obama would have more crossover appeal than Christopher Dodd or Bill Richardson.

The main point here is that it isn’t impossible to correctly perceive in advance which candidates will have more potential to win over people who supported Donald Trump. The error comes in thinking of every Trump supporter as some irredeemable troglodyte whose support is a lost cause.

But it’s also an error to think that ideological positioning plays much of a role in this. In our polarized world, people who care about policy are already sorted by party, so a person who is strongly anti-choice is going to be nearly impossible to win over even if they like your moderate position on guns or health care. Trimming on policy isn’t a very good way to win over voters from the other side. However, it is possible to lose your own party’s supporters if you position yourself so far to the left that some find it threatening. Democrats are more reliant on affluent, well-educated professionals than ever, and they can only push them so far before they’ll flee. Fortunately, Trump is so distasteful to this contingent that the Democratic nominee has a lot more ideological room to maneuver than would normally be the case.

This is a long way of saying that electability should inform your decision. The key is to properly understand which factors matter and which do not. If people like and trust the nominee, there will be a big crossover of votes. If suburbanites aren’t overly threatened by the nominee, they will stick with their strong inclination to vote Trump out. But if the nominee only appeals to partisan Democrats and no one else, they’re going to be at risk of losing no matter who they’re running against.