Here’s a quick tip from someone who gets into way too many political conversations: If you find yourself in a room full of committed progressives talking about 2020 strategy, but you don’t really want to engage, just ask: “Are you sure that will play well in Pennsylvania, Wisconsin and Michigan?”

It’s all but guaranteed: At least one person will launch into a soliloquy about how the most progressive version of the Democratic Party is actually the most electable version.

The exact content of the progressive argument might vary. Some say that most Americans want progressive policies and Democrats just need to be louder and contrast more with the GOP on economics. Others think that President Trump is simply too unlikable to win and that Democrats should take the opportunity to nominate the most liberal person possible. But the most interesting response, and the least common one I’ve run into, is that Democrats don’t need all those states.

Some think the best Democratic strategy in the medium term (or even the short-term) is to focus less on Pennsylvania, Wisconsin and Michigan and invest instead in diversifying, suburban Southern and Southwestern states. That approach, the theory goes, would allow Democrats to win without nominating a comparative moderate such as former vice president Joe Biden and trying to appeal to the more conservative blue-collar whites of the Upper Midwest.

That’s not a crazy strategy, but Democrats should be cautious. Purple states are highly sticky from one election to the next, so they’re probably not going to be able to get away from the traditional Midwestern swing states in 2020. But the right candidate could still coax new battlegrounds onto the map.

Recent electoral history and demographics suggest that, despite the quick leftward movement of well-educated suburbs across the country, Democrats probably still have to worry about Wisconsin, Florida, Michigan, Pennsylvania and other currently purple states in 2020.

This series of maps shows the baseline political leaning of every state in every presidential election from 1976 to 2016. These aren’t results maps — they instead show us which states were more or less Republican than the nation as a whole. The light-red and light-blue states fall in the middle — they’re the ones you’d expect to be close if the election had been neck-and-neck that year.

While the marginal states often change in the long term (see the difference between the 1976 and 2016 map), we’re now in an era of swing-state stability.

The 2000 map strongly resembles the 2016 map, and states such Pennsylvania, Ohio, Wisconsin, Minnesota, Florida, New Hampshire and Colorado have been near the political center for much of the 2000s and 2010s. Most of these states are large and just diverse enough to be competitive. They have enough well-educated and non-college-educated white voters to give the Republicans a strong base, as well as enough white liberals, still-blue working-class whites and voters of color to give the Democrats a base. So unless Democrats nominate someone who radically shakes up their coalition (and none of the leading primary candidates seems poised to do that), or some Republican other than Trump is somehow on the ballot in 2020, we should expect both parties to feel forced to compete in familiar, electoral-vote-rich, typically marginal states.

Put simply, the winner of the 2020 election will probably get his or her 270th electoral vote from a familiar state, despite the wishes of some progressives.

But Democrats aren’t powerless over the map. By nominating Trump, Republicans were able to take over Iowa and Ohio and push both Maine and Minnesota closer to the margin. Trump wasn’t enough to move other states with large non-college-educated white populations, such as Wisconsin, Pennsylvania and Michigan off the map, out of play. Similarly, as Democrats have become more urbane and politically liberal (think of the differences between Bill Clinton and Hillary Clinton), they’ve lost strength in the rural South but locked down much of New England and turned Colorado and Nevada light blue.

If Democrats want to make further inroads into Arizona, Georgia and Texas while still winning the quadrennial swing states, their best bet is to find someone who can claim some liberal, upscale cultural credibility without looking like a socialist revolutionary. The last Democrat to win statewide office in one of those three states is Arizona Sen. Krysten Sinema, who could claim liberal cultural and political credibility as a former Green Party activist and anti-Iraq-War protester while using her moderate congressional record to win over formerly Republican suburbanites in Phoenix. The candidate who most clearly fits this profile is former Texas congressman Beto O’Rourke — whose voting record is more moderate than his reputation, who is from the region and who has clearly demonstrated that he plays well in Southwestern suburbs. But it’s possible to imagine Sen. Cory Booker (N.J.) or South Bend, Ind., Mayor Pete Buttigieg managing this kind of strategy.

The major parties have only so much control over which states end up in the battleground. The Republicans nominated dramatically different candidates in 2012 and 2016, but most states and voters stuck with the party they always vote for. But they do, mostly through their presidential primaries, control the cultural and political image of their party. If Democrats nominate a Joe Biden or Sen. Amy Klobuchar (Minn.), they may have the best chance of winning the electoral college through the Midwest. And Sens. Elizabeth Warren (Mass.) and Bernie Sanders (Vt.) are the obvious choice for those who want to nominate a progressive true believer first and figure out the coalitions later.

But if the Democratic dream is to shift the map, the party may need a candidate who contrasts strongly with Trump on cultural issues but maintains the sort of mixed voting record that can pacify Mitt-Romney-loving suburbs. It’s not the most inspiring formula, but it just might work.

Read more: