MASHABLE - Tinder: The Tinder algorithm used to be based on the Elo rating system, which was originally designed to rank chess players. The "Elo score" gauges how other profiles interacted with yours. It logged your swipes. Elo is no longer used. Now, the most important thing a user can do is...use the app. The more data Tinder has on you, the more Tinder knows your preferences and the algorithm is based on how your profile is Liked or Noped.
Hinge: The dating app "designed to be deleted" doesn't have swiping. Hinge uses the Gale-Shapley algorithm. This Nobel-prize winning algorithm was created to find optimal pairs in "trades" that money can't buy - like organ donations. Say there are 10 women and 10 men. How do they get paired up? Well, tell one group (either the men or women) to pick their first choice, and if they get rejected they move on to their second choice. Continue until none of the people left want to get matched anymore. Tell the app when you've met a match in person and Hinge understands who you're interested in.
OkCupid: OkCupid has 4k questions. There's 60 sexual orientation and gender options. OkCupid calculates a match percentage. If another user has similar search preferences and responses to questions as you, and is looking for the same things relationship-wise, you'll have a high match percentage.
Bumble: Bumble is similar to Tinder in that it uses a swipe model. Bumble declined to comment about its search algorithm.
Grindr: Grindr only uses algorithms for security purposes, like detecting spam accounts. There's no recommendation algorithm to speak of on Grindr today.
Mark Brooks: Frankly, I'm surprised at the lack of substance here. Our job is to bring clarity to a complex problem. Who to date? Who to marry? Who to raise a family with? (...and perhaps guide people further on how not to screw it up). This is not an unsolvable problem, but it is highly nuanced. Culture, religion, values, and lifestyle preferences compound the problem. Also, what people say they want and what they are willing to accept can be quite different. i.e. stated preferences vs revealed preferences. Observing people's behaviors and getting feedback after each date is key. Also, we really need to know when we have been successful. When we have clear visibility on this, at scale, then AI becomes useful for helping us solve for long-term compatibility.