Kiss.com—the first modern dating website—launched in 1994, followed by Match.com, one of the heavyweights of the internet love industry, a year later. In 2007 the iPhone allowed us to swipe right and pinpoint locations, paving the way for apps like Tinder and Hinge, which both launched in 2012, to profoundly and permanently change the way we meet people. Since then, the number of people willing to share personal data online to find love and relationships, and the number of services helping them do that has blossomed. It’s worth asking on Valentines Day, 30 years after the advent of online dating, if data-based love is the best we can do?

Dating by the numbers

Overall, 30 percent of U.S. adults say they have ever used a dating site or app, according to Pew research, and that share rises to more than half (53 percent) for those under 30 years old. Ten percent of partnered adults (i.e., married or living together in a committed relationship) met that person through a dating site or app. And Tinder is the app of choice: Its 75 million active users represent 46 precent of overall dating app users (the popular app accounts for 79 percent of dating users under 30). So online dating is pervasive, especially with young adults. But are they satisfied?

According to the Pew research, more respondents indicated their experience was positive (53 percent) than negative (46 percent), although women report being slightly less satisfied (48 percent positive vs. 51 percent negative) than men (57 percent vs. 42 percent) while LGB users were more positive than both (61 percent vs. 53 percent).

Making the match

With so many people using dating apps, but only 10 percent of committed relationships resulting from matches set up by them and many users dissatisfied with the experience overall, the apps don’t seem to be finding the best partners for users most of the time. But the two-sided nature of relationships, the fact that users aren’t always honest in their profiles and preferences, and other factors mean the algorithms developed for dating apps must be more complex and might not produce optimal results.

“Algorithms are used to recommend people to you on a dating app a lot like how Amazon would recommend a product or Netflix would recommend a movie,” Liesel Sharabi, the director of the Relationships and Technology Lab at Arizona State University told the Wall Street Journal in a recent interview. “But unlike products, people have to reciprocate your interest. The algorithms have to take that reciprocity into account when they’re forming matches between people.”

Sharabi draws a significant distinction between the services that require users to self-report preferences vs. the ones that gather behavioral data by analyzing how the user interacts with the app itself. She notes that what people think or admit to themselves they find attractive doesn’t align with what they actually find attractive.

“You might think you wouldn’t date someone under 6’1”, but if they were charming or you really connected you wouldn’t pay attention to something like that,” she said.

The apps are moving away from self-reporting and have turned to figuring out preferences by how they engage with what profiles (what they swipe on, what they linger on, etc.).

Increasingly, users are trying to bypass the algorithms altogether. A recent New York Times article reported growing dissatisfaction with dating sites users feel are not producing quality—or even enough—matches. They are “burned out from swiping” and a dating coach interviewed for the piece notes app users feel “I’m doing everything right; it must be the app’s fault.” In response, they are seeking to game the apps, which they are viewing with increasing distrust since many have paid tiers now and the algorithms are opaque. Most are disappointed, the author concludes.

Is this the best we can do?

Unfortunately, it might be. Despite the disenchantment of many, the apps still are possibly the best place to meet people. Traditional places where people would meet in the past are dwindling. Remote work has rendered the office and the after-work social scene less relevant. Community engagement at places like church is becoming less central to Americans’ lives. Despite the health benefits of face-to-face engagement, we are spending more time online and less engaging physically with others (though that is the goal of using the apps in the first place).

Perhaps the goal should be making the apps, themselves, better. Introducing more AI into the mix could help, but that requires lots of data. Despite Sharabi’s contention that apps are moving away from self-reporting, some are going the other way and asking for even more in-depth information to feed AI analysis. Hinge, an example of this, has a “Most Compatible” feature, which analyzes a user’s stated preferences and sends recommendations of matches that it thinks will be a particularly good fit.

Getting at a prospective match’s personality first rather than immediately judging a photo is another option. An app called Jigsaw makes users’ profile photos unavailable until after contact has been established and matches begin communicating online.

Introducing data and technology into a process as ephemeral and intuitive as finding love and relationships is an ongoing experiment. On Valentine’s Day it’s important to remember that there’s no equation that will explain it perfectly—but we can keep trying to find one.