Did Hinge Change Its Recommendation Algorithm?

Yes. Hinge says it rebuilt the system that decides who you see around deep learning. The aim is to predict mutual compatibility better: whether you will like her, and whether she will like you back. Hinge also says the update contributed to a double-digit increase in matches across the app.

Hinge describes the change in two parts.

The first is the method. Its recommendations run on a deep-learning system. That is a kind of software that learns from huge amounts of real behavior instead of following rules someone wrote by hand.

The second is the goal. The system is built to predict mutual compatibility. It is not only guessing which women you would like. It is also guessing which of them would like you.

Hinge gives one result for the update. It says the change contributed to a double-digit increase in matches overall.

That last word is worth a second look. "Overall" means across everyone on Hinge. It is a figure for the whole app, and it says nothing about your own account. An average can rise while one man's week looks exactly the same as it did before.

Most men treat the feed like a menu. The app lays out women, and you pick.

A system built around mutual compatibility is doing two jobs at once. It guesses whom you will like. It also guesses whether she will like what she finds when she gets to you.

So your own page shapes who you are shown. The better the system expects women to respond to you, the more reason it has to put the women you want in front of you.

a dating coach has made the same point about dating apps for years. An app is a business, and it wants its most attractive users happy and coming back. A man whose photos and profile get a good response helps it do that. In My experience, the app notices and starts showing him more.

A better predictor decides who gets a look at you. That is where its job ends.

It will not choose your photos or write your prompt answers. It will not read her page and find something worth commenting on. And once you match, it has nothing to say about the conversation at all.

That second half is where most men lose. They get the match, then let the chat drift for a week. Or they ask her out so vaguely that she has nothing to say yes to. More matches from a smarter system only help the man who can turn a match into a plan.

Everything the system knows about your taste comes from what you do on the app. So act the way you want to be read.

Be selective with your likes. My experience across the apps is that liking nearly everyone backfires. An app reads an unselective man as a less attractive one, and it shows him to fewer attractive women. Like the women you would actually take out, and pass on the rest.

Read the whole page before you decide. Her first photo is often the one she thought least about. Her prompts and later photos tell you far more, and they give you something real to say.

Show up in small doses. My advice is that half an hour a day for ten days beats five hours crammed into one. A steady habit gives the system a steady picture of you. It also keeps your chats from going stale.

Men notice when the women in their feed change. A rebuilt recommendation system is one reason a feed can shift. But Hinge describes the update across the whole app, not what it did to any one account.

So do not guess at the machinery. Look at the numbers you control. How many of your likes come back as matches? How many of those matches turn into a real back and forth? How many of those become a date?

Whichever number is weakest is the one to work on. That stays true whatever the system is doing underneath.

Change one thing at a time. Swap your lead photo, or rewrite one prompt answer, and give it a week. My rule is to test, keep what works, and cut what does not.

Yes, Hinge changed it. Its recommendations run on a deep-learning system built to predict whether both people will be interested. Hinge says the update lifted matches across the app.

What no system can do is make a woman like what she finds on your page, or turn a match into a date. Those parts are still yours to get right.