Towards Better Online Dating

As soon as I deleted my OkCupid profile I started meeting more people in real life. At least for me, where I was putting my energy was a zero-sum choice.

> IMHO it's a far better bet to predict whether or not two people can get through dinner without trying to kill each other... and just send them on a date.

That's the thing. It's true that a similar world view is important for a long term relationship, problem is that there's well known sociological phenomenon where by people will not answer honestly about their actual world view (choosing one they consider to be more socially acceptable).

The very problem of course stems from the term "online dating". As I stated in my comment, it should really be about "online discovery of people you can go on a date with". The algorithms should not be about matching, but about discovery of people where interest may be mutual.

An example of meeting friends-of-friends is soshi.com. They recently launched at Stanford and leverage your facebook network to display eligible friends of friends, it two people express interest in one another, it matches you up.

>The fact that there's so many variables isn't the reason such an analysis would fail. The reason is that there's just not enough input to predict personality type with any statistical significance. Isn't this the whole idea of factor analysis/singular variable decomposition?

How about: Instead of a static profile, a tumblog-style First Impression that contains a persons consumption habits (movies, music, tv, etc) as well as pictures, feelings and thoughts. They can update it incrementally to keep it fresh and relevant. You can follow another persons First Impression to get a passive view of them over time. Way better than canned answers to canned questions and filling out surveys.

The service automatically groups similar people (demographics, location) into groups of ~20. 10 guys, 10 girls. They are then led through a week-long online introduction process where each day they respond to a prompt about a different aspect of their life. By the end of the week, hopefully people have gotten a feel for one another and if they are interested, you help them plan further communication or a date.

You use the ratings gathered (+ other usage/interaction data) during the week to assemble the next groups based on similarity, etc.

Group setting = no spammers, no awkward immediate 1-on-1, incentive to behave, multiple partner options, attentive group-mates that arent off 'playing the field' w/ 100 other people.

Week long = interaction over time builds trust, more time to learn about other people, people like to be guided through awkward/ambiguous social situations.

Pay-per-play instead of monthly subscriptions ($10/week) so that a dater can come and go as they please. You know everyone in the group is currently active and is available for communication (vs. long dormant profiles, etc).

Etc, etc, etc.

I built this and launched it and it operated from 2007-2010. It was called FlowMingle.com and had ~25k members. But, I gave up and shut it down. Why? Because I got tired, didnt have any money, etc. I still think this is an amazing way to do online dating, and my users agreed. Our sign-up conversion rate was ~22% (!!), and this is with 0 marketing dollars ever spent.

But you know what I learned? Some people appreciate legitimate interactions and thoughtful conversation and fruitful behaviors, and some people just want to 'play the field' and take their chances at landing the big one. The online dating space is every bit as complicated as the online daters themselves.

I would be happy to share my experiences with anyone who is interested.