Richard Pildes is Sudler Family Professor of Constitutional Law at New York University. Jonathan Rodden is political science professor at Stanford University and senior fellow at the Hoover Institution.

Democrats have engaged in a passionate debate leading into the midterms on Nov 6. “Progressives” argue that the path to victory this year and beyond lies in motivating their youthful urban base by moving the party to the left. “Pragmatic” centrists, on the other hand, argue that victory requires ideological moderation that will attract independents.

Paradoxically, both sides might be right, which is why this tension is unavoidable and likely to endure. To understand this, we must grasp how electoral geography shapes politics. President Trump won 230 congressional districts to Hillary Clinton’s 205, even though she outpolled him by more than 3 million votes nationwide. This reflects, in part, the fact that progressive voters are increasingly concentrated in the areas that make up urban congressional and state legislative districts, while moderates and conservatives are more evenly dispersed in exurban and rural districts.

As a result, in competitive states, the decisive voter in a statewide election is to the left of the decisive district. This means that a Democrat with a relatively progressive platform might be able to facilitate high turnout and win the statewide popular vote. But an identical platform would be too far left for the pivotal districts that determine the make-up of the state’s congressional delegation or state legislature.

In other words, even within a specific state, the strategies preferred by progressive leftists and centrist pragmatists can both make sense electorally, depending on what type of election they are most concerned about winning. Relatively progressive candidates can win Senate races in states such as Ohio, but the same liberal reputation drags down Democratic candidates in the decisive districts needed for overall control.

To win control of Congress and state legislatures, Democrats mustcapture relatively conservative districts that support Republicans in presidential elections. Structurally, this is nothing new. Democrats have been relatively concentrated in urban districts since the New Deal, and for decades, their geography made it necessary for them to field congressional candidates who could win on “Republican” turf in the suburbs and countryside.

The Democrats achieved this not by nudging their platform to the left or right, but by avoiding a coherent platform altogether. In the 1950s and 1960s, the party of Northern urban workers — including African Americans — was simultaneously the party of Southern segregationists. When religion and social issues became politicized in the 1970s and 1980s, pro-life, pro-gun Democrats were able to win congressional seats in suburban and rural districts around the country.

These idiosyncratic Democrats with locally tailored platforms often initially found their way into Congress as part of “blue waves” following perceived Republican failures. They managed to sustain themselves in Republican districts with a mix of moderate roll-call votes, pork-barrel politics and constituent service. Democrats controlled Congress for decades, including during Republican presidencies, not because their party chose the right national platform, but because they allowed their candidates to craft local “brands” that differentiated them from the party’s urban candidates.

The Democrats’ geography problem has become far more acute in recent decades. Since the late 1980s, votes for Democrats have become ever more concentrated in cities. In addition, Republicans now control more state legislatures than at any previous time, and partisan gerrymandering in several states has manipulated the underlying political geography to make matters far worse for Democrats. Local races have also become more nationalized as politics has become more polarized, with outside money now playing a much larger role in legislative races.

This makes it harder for individual candidates to detach themselves from the party’s national image. Some Democrats, such as Rep. Conor Lamb (Pa.), have pulled it off, but it’s significant that Lamb was nominated to run in his special election not through a primary race, but by local party officials who were particularly sensitive to these dynamics.

To be sure, the nationalization and polarization of elections might benefit Democrats in this cycle if enough Republicans and independents in pivotal districts cast “anti-Trump” votes for Democrats who would otherwise seem too far left. An exceptionally powerful wave election can overcome Democrats’ underlying geography problem and Republican partisan gerrymandering and thus enable victories in suburban districts that typically vote for Republican presidential candidates.

But any such wave is unlikely to wash the Democrats’ geography problems out to sea. It is not yet clear whether the Republican Party’s reputation has suffered significant damage in pivotal suburbs and, if so, how enduring such damage might be.

Whatever the outcome this fall, the Democrats’ basic geography problem is likely to endure. To maintain control of the House or state legislatures beyond the isolated wave election, self-styled exurban and rural Democrats will feel the pressure to craft local brands that distance themselves from their party’s liberal reputation, even if that reputation serves the party well in winning the national or statewide popular vote.

Political geography — not just ideological conflict on its own — thus makes it likely that tensions between the progressive and centrist wings of the Democratic Party will endure.

Read more: