Online dating is a dumpster fire. You can assume your date is 20 pounds heavier than any photo reveals, accept their witty texts won’t translate to any real-life sense of humor, and go into the first date with an open mind — and still be desperately disappointed.

That’s why millions of Americans are spending billions of dollars on dating sites like Eharmony, or OKCupid — services that claim they can use “scientifically-proven matching systems” to weed out all those sub-par suitors and leave you with only the cream of the crop.

However, these companies with “scientific algorithms” haven’t let any actual scientists see their miraculous algorithms. Odds are, they’re too fearful that researchers would find their science is a bunch of good-for-nothing gobbledygook, which is precisely what psychologist Dr. Samantha Joel of the University of Utah discovered after years of researching attraction.

“Studies show that people have all these preferences, requirements and deal-breakers for what they want in a potential partner, but as soon as they’re presented with someone, all those things go out the window,” Dr. Joel explains from the stage of a recent TED Talk.

She proved this with her own study, in which she presented people with dates that had at least 2 of their personal deal breakers — things they had explicitly said they did not want in a partner. The vast majority agreed to go on the date anyway.

This is because even if a potential date has qualities we thought we’d never touch with a 10-foot pole (for example, they still to Limp Bizkit), if that person has other redeeming qualities (for example, they have a great ass), we’ll usually give it a shot. Since we’re so terrible at predicting what we’ll like in our partners, it’s no surprise that a computer might be terrible, too.

Yet countless sites still insist that their innovative calculations can use personality traits and romantic preferences to find your potential soul mates.

So Dr. Joel and her team created their own algorithms. The goal: figure out which partners will find attraction based on their personalities and romantic preferences.

They took 350 straight college students at Northwestern University, had them fill out questionnaires quizzing them on over 100 traits and preferences theorized to be essential in relationships — things like self-esteem, sense of humor, intelligence, ambition, values, patience and creativity — and put everyone on 4-minute speed dates. Then, participants rated how attracted they felt to each person they met.

Surprise, surprise, the scientific algorithm couldn’t make any damn sense of the results. No matter what combination of the 100 traits and preferences the computer used, the machine was completely useless in predicting which pairs of people would be attracted to one another.

"It predicted 0 percent of the matches,” Dr. Joel tells NPR. Some of the models we ran got a negative percentage, which means you're better off just guessing."

Online dating sites that pride themselves on their “scientifically-proven matching systems” have yet to acknowledge the real science that pokes massive holes in their business model. It’s highly doubtful they ever will, either, because these companies don’t really care about making matches. They care about making money.

Ultimately, Dr. Joel’s study highlights just how little scientists currently understand about attraction. As of now, it seems impossible to anticipate whether or not we’ll want to hump someone before meeting them face-to-face.

In the frustrating free-for-all of online dating, it’s alluring to think that technology could do all the hard work for us. But all those nights we waste on dates that fail to live up to the promise of their profile seem to be a necessary evil.