In the twenty-four hours before I wrote this article, three candidates dropped out of the democratic presidential primary: Tom Steyer, Mayor Pete Buttigieg and Senator Amy Klobuchar. Shortly after this article will be printed, the polls will open in Massachusetts and a basket of other ‘Super Tuesday’ states. The consensus among the pundits is that this is now basically a two-person race between Joe Biden and Bernie Sanders. Bloomberg, my former mayor, is still dumping obscene amounts of money into advertising, but I don’t think people expect him to go very far. Tulsi Gabbard remains a darkly interesting footnote, pulling reassuringly negligible numbers as she continues to apologize for far-right leaders around the world. This leaves us with just one other candidate: our own senator, Elizabeth Warren. She’s been the focus of much pressure in the last few hours to follow her recent competitors and drop out, and, if she wishes to remain influential, cast her lot with someone who can actually win.

The trouble with this pressure is that Warren is not purporting to have a shot at winning an outright majority of delegates: she seems, from what I’ve read at least, to be angling for a contested convention in Milwaukee, depriving Biden and Sanders of enough delegates and winning the few needed herself, to have a shot at convincing their own delegates to vote, on a second ballot, for Senator Warren — making her the nominee rather than either of her competitors, despite her having received far fewer votes than either one of them. This poses an obvious democratic (with a small ‘d’) dilemma.

A comparison might be made to the electoral college in a general election: surely it’s unjust, the argument goes, that Donald Trump became president even though Hillary Clinton won millions more votes. And, certainly, that’s problematic. But the silliness of electing presidents via the electoral college shouldn’t make us assume that the alternative, though better, is very reasonable. In the last election, Trump won about 53 million votes while Clinton won about 56 million votes. Both those numbers are so big they’re basically meaningless to me, and the difference between them, though also enormous, is proportionally very slim — about 2%.

The roughly half of the country that voted for Clinton was devastated in 2016 not just by the result, but by the arbitrary way it was obtained. It wasn’t just that they lost — it was that they won, and still they lost. But suppose the popular vote were reversed — that Trump had won 56 million votes and Hillary only 53 million. Or suppose that Hillary had won both the popular vote and the electoral college. In any of these scenarios, half the country is governed by someone they have no faith in. And a country whose people are faithless can easily become a state that is perceived as illegitimate. If November yields a narrow Democratic victory, this legitimacy problem could be hard to ignore, especially if the current President conjures up doubt, as he often does, about the facts.

Now back to the Democratic primary. Imagine the following contested convention scenario: Nobody receives an outright majority on the first ballot. Bernie Sanders receives a plurality, Biden follows close behind and Warren is way below them, with just enough to muck up the system. What is the democratic thing to do here? There is an easy argument that Sanders should receive the nomination: he got the most votes. But, for the reason I outlined above with regard to the general election, I don’t think that’s as compelling as it might seem at first glance — whether it’s Sanders or Biden, half the people lose. And that’s reasonable enough when there is no other option.

But what about when there is the option to nominate someone who, despite having received very few votes the first time around, may be more acceptable to more people than either of the other two candidates — a mutual-second-choice scenario? This is the democratic argument for nominating Elizabeth Warren in such a scenario, and it is also the argument, put forth by the New York Times editorial board recently, for abolishing the current convoluted system altogether and replacing it with a national ranked-choice primary in which the person with the broadest base of support, rather than the best consolidated ‘lane,’ would walk away the nominee. (Since the last election, a number of states have adopted ranked-choice voting systems; the Voter Choice MA movement is working to get it here in Massachusetts.)

But in the absence of a better system, we have to make principled decisions about how to use the one we have. That requires thinking not just about numbers, but about where the legitimacy of a government actually comes from — this is a task that all states have to reckon with in this particular historical moment, when the whole world order is being challenged, but it is one with which the opposition to Donald Trump, in particular, has to grapple. The Republican base believes in Trump — often fanatically. At the same time, a vote for Trump in 2016, like a vote for any demagogue in a free election, was not so different from a vote-of-no-confidence in the existing system.

Eking out a Democratic victory in the electoral college in November, even if it succeeds in effecting a peaceful transfer of power back to a competent, reasonably moral adult, will not automatically fill the legitimacy-void that Trump now claims squatter’s rights to occupy. Neither will a numerical majority in the popular vote. Indeed, it will be hard to achieve that numerical majority, or that electoral victory, unless the candidate who runs against Trump can make a claim whose democratic nature doesn’t begin and end at the ballot box that Tuesday afternoon.

And so, having defended Warren’s pursuit of a contested convention, I want to close with what I think is a better argument for awarding the nomination to a Sanders plurality than one that is based on a delegate count: the Sanders campaign has a credible assertion of legitimacy that no other candidate, absent a majority of delegates, can overcome. This legitimacy begins, but doesn’t end, with Sanders’ delegates: it comes from the record number of small donations to his campaign. It comes from all the people who have used that campaign as an organizing hub for countless local causes apart from the presidential race — a subscriber to campaign texts receives updates, for example, on local labor strikes. If Warren or Biden can make a better claim to this sort of legitimacy, let them.

But as we march toward the convention, the Democratic Party ought to take democracy — not just delegate counts — seriously. That’s what’s going to matter in November, no matter who ends up as the nominee.