Math Nerd Games OKCupid to Find the Perfect Match

He didn't trust their algorithm, so he built his own.

  • Share
  • Read Later
Getty Images

Think OKCupid has an OK data set, but don’t trust their algorithm to sufficiently match you with the perfect mate? If you’re mathematician Chris McKinlay, there’s only one simple solution to this quandary: build your own algorithm that will pool only the best matches, thereby increasing  your chances of finding the ideal wife.

According to Wired, McKinlay set up six OKCupid profiles operated by bots and used them to gather data about 20,000 women on the dating site. Then he employed a method of dividing the women into groups called K-Modes, leaving him with 7 different statistically different clusters. From those 7, he was able to determine which two groups were the most appealing to him.

After putting those two groups through yet more series of tests–which you can read about over at Wired–and going on 87 OKCupid dates that yielded few genuine leads, McKinlay met Christine Tien Wang, an artist. Towards the end of their first date, he felt so comfortable with her that he confessed the whole romantically questionable charade: the equations, the data mining, all of it.

“I thought it was dark and cynical,” Wang told Wired. “I liked it.”

A year later, he proposed to her over Skype. No word on whether McKinlay is devising a system to determine the best time to have kids.