Huh, how interesting. I don't believe I'd want to use your system.
You're right that abstract categories and open ended text-fields can make for vague and deceitful answers, but... that's the person you're seeing reflected in the text! I want to read that person, I want to see how she describes herself or her interests. Or how she doesn't. Some girls are shy in their texts, and that says something too right?
It doesn't matter if you start off at a bar/party/box social or via text. I've gotten mails that completely fail to interest me and I'm sure I've sent them out as well. Just as I've been at parties where my smile freezes as I realize what an idiot I am and/or she is. It's as difficult to spot the crazies in text as it is in real life, but that's not really a problem per se. That's human nature. Making everything quantifiable strikes me as an engineer's attempt at a solution (no offense intended, some of my best friends are engineers :).
If I wanted the scientific approach to women I'd build my own robot, I actually prefer the little blurbs of handwritten text that let me analyze between the lines. I think of it as a richer way of communication than just a straight up list of facts*. But that's me, I have faith in my ability to parse meaningful metadata from text and voice. To others it might seem like white noise, so to each their own I suppose.
* With that said I do love being able to define "no smoking" as an easy filter rule. Good riddance!