Hacker News

I used to work for one of the big Internet companies when we came out with the Personals product. We poured over the data constantly, mostly because the thing was so insanely cheap to make (2 devs, 1 ued, 1 pm, 3 months) and so insanely lucrative.

Basically, there were two types of sites: data/matrix driven where you could get a ton of results and filter them as you saw fit, and algorithm driven, where you would get a select number of profiles presented that you could either interact with (swipe right) or deny (swipe left).

The first kind generally suffered from a very simple problem: people don't meet and like people because of data. We couldn't figure out why people like people, because we didn't see that part of the equation, but we saw that the data driven approach pushed people into being too selective on the data, ending up with really unrealistic expectations, and correspondingly with people increasingly lying to match those expectations. We could tell both, because we could see the filters (e.g. the scary number of men in their 50s having 25 as max age for their dates) and the distribution of the data, that didn't match any normal distribution (e.g. the remarkable doubling of people with ages ending in 9).

The algorithm-driven approach suffers from doom scroll syndrome. Since you are separated from the perfect person for you by a bunch of "losers" in the way, you scroll over them quickly to get to the good stuff. Nobody gets the time they deserve, and while the algorithmic approach doesn't allow you to filter unrealistically, it gives you impatience and makes the unrealistic expectations worse.

Unrealistic expectations breed lies. If only the impossible is good enough for you, only the liar can deliver.

I left as we were discussing a blended approach: the algorithm presents a grid of potential mates it selects for you, but you can see them all (a thumb, at least) and interact with each independently. Then the company hit trouble and Personals was frozen in time.

My takeaway: for a change, capitalism and monetization are not the core problem. People's sense of entitlement is.

People did make good ones then Match Group buys them and enshittifies them so they are all similarly useless unless you pay them... I hate match group

Commercial interests, uninterested/ insincere/ dishonest users, it's not the tech, really. It's that the diamonds get lost in the rough.

How do two people decide they like each other enough to form a long term relationship?

Part of that has to do with cues that really do require physical interaction to gauge. Apps will never bridge that gap.

Another element of it is the inundation of choice. Because there are so many options, and new options every day, it's tempting to hold out waiting for Mr./Ms. Perfect. (Frankly, women tend to fall more into this camp though. Desperate men often try to match as many women as possible, and the exact opposite is true of attractive women on algorithm based apps.)

Finally, I think what people think they want, and what they actually settle for, are very different when it comes to relationships. Two people might become infatuated with one another because of regular contact (work, volunteering, hobby, etc) and end up in a successful relationship despite maybe not being an instant match. Apps will also never replicate this, and its sort of a distillation of the above two points.

There is one effect that I've seen described in an article a while ago that seems plausible:

Every dating app user base tends to separate itself into one group that's relatively popular and one that is less popular. For this post, let's assume women are the popular group but on some apps it could be the other way round or an entirely different split when you consider non-heteronormative matches. The problem is that this effect causes a feedback loop. Women get a lot of matches so they get more and more selective and give out fewer and fewer likes. This causes men to get fewer matches so they give out more and more likes so they don't miss the few potential matches they might get. Which in turn causes women to get flooded even more and so on. In the extreme case, women can assume that almost every time they swipe right on someone, this will lead to a match while men need to swipe right dozens or even hundreds of times and include people who don't really match what they are looking for until they get a match.

There is no easy way to fully solve this but making incoming likes visible might at least reduce the fear of missing a potential match if you don't swipe on everyone who's even remotely interesting. Of course this is the number one feature that dating apps try to monetize.