How OkCupid predicts who you will like
tech.okcupid.com
tech.okcupid.com
What I mean is that taking the convenient statistic and saying "seems to be 80% good" instead of looking at the end-user-experience is a huge part of the problem. Psychology studies show over and over that often more choice can lead to less satisfaction, for example.
[To point, maybe giving a user 5 bad matches and 1 good match is more likely to result in a date than would giving them 6 good matches. If you don't look at the big picture experience-wise all of your complicated math thoughts may be of 0 or negative value, ironically]
Now I get it's much easier to model online dating as though it's a marketplace of assets rather actual (emotional, universally irrational) humans. But I think the real value proposition is in the latter.
Only used it for a couple of months but the user experience was superior to the other one I tried (PoF which felt like it was abandoned).
She's pretty much a female version of me minus the programming (though her fascination with shipping and logistics is pretty much as bad as mine with computers).
I mean, sure, fight the good fight and all that, but if you're questioning whether you're even a woman I'm afraid dating me is not going to work out. It makes me wonder what sort of signal I sent that their algorithm locked onto.