Why Dating Sites Lie About Algorithms, As Told By a Dating Site CEO

Mr. Furmansky.

This is a guest post from Alex Furmansky, founder of Sparkology, a luxury dating site for young professionals. A Penn grad, Alex traded in his finance and tech career to follow his passion for innovation and belief in chivalry.

Last week was riddled with reports about a new study being published in the journal Psychological Science in the Public Interest. With each day, editors managed to create more provocative, click-inspiring titles like:

Friends, investors, and even my own mother forwarded me the articles—each hoping to elicit a shock-filled response. However, there is one tiny problem: I LOVE the study.

Behind the Headlines

At the heart of the study is the authors’ determination that mathematical formulas cannot correctly identify pairs of singles who are likely to have successful, long-lasting romantic relationships.

Such claims have been the crux of ad campaigns by eHarmony.com, Chemistry.com, PerfectMatch.com, and GenePartner.com. These sites require users to take extensive “personality tests” that are supposedly crunched by a supercomputer harboring secret formulas for everlasting love. With results in hand, the user is paired with their ideal match.

The study has simply confirmed my claims that this premise is completely bogus.

Every marketing professor should tip their hat to eHarmony. The company that managed to brainwash an entire nation into believing in magic. Singles, however, should listen to science rather than ad men.

Just for fun, I’d like to share an excerpt from Sparkology’s business plan, which we used to secure our round of seed funding last April:

Many current online dating sites have found an ingenious way of defending their price points or differentiating themselves from competitors: The Hidden Algorithm, The Secret Matchmaking Analytics, or the Dr. [Insert Foreign-sounding Name]’s Guaranteed Personality Test. From a marketing perspective, the concept is brilliant – a claim that can neither be proven nor disproven. Yet in reality, these algorithms do not add any measurable probability of success. Does the fact that Person A likes fishing and Person B likes sushi mean they are meant to be? Or is this just a matter of statistically insignificant correlation rather than causality?

Dan Ariely, a professor of behavioral economics at MIT currently researching such dating algorithms concludes “the sites are claiming a lot, but show no evidence of doing anything useful in terms of matches”, further mentioning that “their algorithms are placebos”. In other words, algorithms seem effective because the participants believe the marketing hype. This is ideal for online dating sites defending their pricing, but a complete hoax for honest members hoping to find someone special.

So: Do Dating Sites Work?

Too many dating sites have either engaged in false marketing, unethical practices, outright fraud, or a combination of all three. I’m happy to see more journalists take a stand and educate the public. I see this as an opportunity for a new crop of moral, forward-thinking, and quality-driven dating sites to disrupt the current ecosystem of disingenuous giants.

Dating sites offer an incredibly efficient and effective way to meet wonderful people outside of your social circles. The key to success lies in taking the time to look beyond the fancy TV commercials filled with embracing couples, and instead understand each site’s principles.

  • How does the site moderate its pool of singles? The candidate pool is the most important, objective criteria you should consider. All formulas aside, you want to be in good company.
  • Does the site’s business model encourage hookups, dating, or marriage?
  • How easy is it to get from registration to meeting someone in person?
  • Are you proud to join the site? Or do you feel “icky” about it? Trust your gut.
  • Does the site have any policies about how it handles your data? Sharing/selling/buying profiles? Fake profiles? Fake interactions?

I urge you: Do not hold an entire industry accountable for the missteps of the bad actors.

Are algorithms good for anything?

Yes. Sparkology has a behavioral algorithm that recommends results based on users’ actions, including clicks, messages, likes/dislikes, and others. Here’s the difference: instead of claiming to identify a user’s soul mate based on a personality test, the goal of our algorithm is to identify people who the user would be most attracted to based on the user’s past behavior. It can also help illuminate people who would be more responsive, by analyzing whether the person has been responsive to persons like you in the past.

The algorithm allows us to showcase candidates worth your time without requiring extensive profiles. No algorithm can pinpoint exactly what makes two people spark… but when a spark occurs, our algorithm can use that information to predict future sparks.

In other words, we don’t survey everyone in the bar and match you by astrological symbol claiming that you and your match are meant to be, or put all the self-described Type A people together in one corner and Type B in the other. But we point out that the cute blonde stirring her Manhattan has a lot in common with the adorable brunette you clicked on, and clicked with, last week.

It’s similar to how Google or Facebook show results or wall posts that are more relevant to you. And it’s better than just showing you everybody at once with no order all, which is what you get when you walk into Max Fish on the Lower East Side.

We’re also not getting into the other issues of walking into a bar… it’s hard to tell who’s really single, hard to talk over the loud music, etc. 🙂 Why Dating Sites Lie About Algorithms, As Told By a Dating Site CEO