[jk8]

"You can choose your friends and not your family".

1. I don't think an algorithm could be developed to find friends. On the other hand, I think people make friends on game sites where, they strike out on their own to make friends. Even meet up sites, where you go to play golf with a bunch of strangers or go play chess.

2. I don't think this kind of site will get traction and people will not pay for it.

eHarmony exists because it makes finding your potential mates easier, it is for people who have given up on bars and friend's recommendation. Yes, there are other ways to meet people but the frequency of finding a good mate will be very less, I think.

I think people just become friends who get along with each other and don't look for friendship actively, like we look for dates on a dating site.

The concept of a site finding a friend has always bothered me, like facebook is trying to find a friend for me or saying "share more info to find more friends"...

The best thing you can do is launch a portal for like minded people to mingle.

Everything cannot be solved by technology.

[zackattack]

>I think people just become friends who get along with each other and don't look for friendship actively

I think this is right. So the question is can you sift out attributes that identify compatible partners? Or is it more biochemical?

[jk8]

Yes, there may be some common attributes and I don't think it is biochemical when it comes to male friends.

I feel that someone can look at the problem differently. To make a site based on a common interest and get people to pay for it. The users of the site will find a way to make friends.

[zackattack]

What makes you qualified to dismiss the possibility of biochemical effects?

[zackattack]

for example, tribal identification via scent, status chemicals