By subtracting choosiness and attractiveness from daters’ scores of romantic interest, the researchers had a more accurate measure of compatibility. “Some people are more attractive than others and we can predict who tends to get the most matches,” says Joel. “That is not the goal of these matching websites. They are not saying they will filter your pool so you only have attractive people to choose from.”

Joel found that her algorithm could predict actor desire and partner desire, but not compatibility. Not even a little bit. It could only predict negative percentages of variance – which is like being accurate less than 0% of the time. This might sound like a bit of a head scratcher, but, Joel says that her algorithm would have been better off using mean results for every dater rather than offering a tailored response. “It was completely useless,” says Joel. “It really should have done better.”

“My take is that when two people actually meet they form a shared dynamic that is more than the sum of its parts and cannot be predicted a priori,” says Joel. “Their individual preferences do not make up the substance of what they find attractive. My rating of whether I found you funny after meeting you will predict whether I like you, but my desire for a funny person and your measure of whether you are funny do not because we might not agree on a sense of humour.”