For most of the history of American democracy, campaign strategy has been driven by geography. Campaigns looked for areas with high concentrations of supporters, and then targeted those areas.

But, as the Economist explores, that has changed with the evolution of the Internet and data science, giving rise to what might be called “data politics.” Its roots lie in the 2002 establishment of computerized statewide voter registration lists, but things really took off after the data-driven 2008 victory of Barack Obama.

Get Data Sheet, Fortune’s technology newsletter.

Now, databases and algorithmic profiling allow canvassers to target not just strong regions, but the front doors of likely supporters. Email and social media campaigns can relentlessly test response rates and hone in on their audiences. And more companies are looking to turn voter data into revenue, such as Civis Analytics, which emerged from the 2012 Obama campaign, and i360, a database firm on the conservative side.

But who does the data drive benefit overall? Clearly it’s the campaigns that make the best use of it—and for years, that battle has gone to Democrats. In the early days of data politics, the Economist finds, left-leaning campaigns found it easier to attract data scientists. More recently, that talent gap has closed, but Republicans have proven less able to work together to build a unified voter database.

For more on tech and politics, watch our video:

But data politics also gives a bigger edge when dealing with particular kinds of voters. Matt Hindman, an academic expert on political communication, suggests that the Democrats might benefit most here too, because their voters are traditionally harder to mobilize. Age may also play a crucial role: Voters under 33 lean 51% Democratic vs. 35% Republican, and that age group is more likely to use closely-targetable digital channels than more conservative, older voters.

Of course, those realities don’t automatically translate to getting people to the polls. That’s something younger voters are notoriously bad at, and it’ll take more than A/B testing to overcome likely Democratic nominee Hillary Clinton’s widening enthusiasm gap.