The Friending Problem

I just unfriended about 20 people on facebook. It's not that I wouldn't like to be friends with those people, it's that I'd like my online social network to bear some resemblance to reality. But I feel guilty about unfriending them. It's as though I explicitly told them all that I really didn't want to be friends; which isn't true at all.

This is a tough feature to get right on any social network. On the one hand, if you keep friend lists secret, then how do you implement the 'networking' features on your flashy social networking site? On the other hand, letting users see their friend lists makes it a big deal to unfriend anybody. So everyone ends up with a fairly large percentage of their friend lists consisting of people they don't really know or haven't yet formed a real connection with.

Someone should figure out a way for unfriending to not be a negative thing. Because it's really not. The negativity is just a function of the way the network is presented, so that unfriending someone is interpreted negatively by the person being unfriended. For starters, the term "unfriending" is negative in itself.

But also, the whole concept of the handshaking process involved in friending in the first place, creates an implicit rejection if one person later chooses to unfriend, because the other person is still in mid-handshake. Every time we add a new friend, we engage in a permanent handshake that just goes on and on and on, because the first person to release hands will offend the other. The only people who typically ever unfriend are people who stop dating each other.

One improvement would be to not create tight social networking links between people in the first place. For example, my gmail contact list automatically includes everyone I've emailed. No one would mistake their own appearance on my gmail contact list as proof of our friendship (though we might very well be friends). But it's possible that we could IM together, just because we've emailed each other. And I could configure my chat client to either show, or not show, or decide on its own whether to show, that person in my chat list. And no one would know whether they appear or not, so they wouldn't be offended by my selection.

If they also have a gmail account, I might be able to view their Google profile, and check out their Google Reader shared items, or any blogs they write, and they might do the same with me; all without ever undergoing this awful online handshaking process that's so hard to unshake.

There are other ways Google allows social networking and closeness without insisting on a 'friending' experience. But there are also ways in which a kind of implicit 'friending' does take place under Google. If I follow someone on Buzz, for example, the other person knows I'm following them. And if I stop following them, the other person knows I've stopped.

That's sort of a 'lesser friending' situation, because it doesn't require a handshake, it's simply a configuration option. I can follow, unfollow, and follow again on Buzz, without inconveniencing anybody or asking them to confirm that I'm allowed to do that. They can block me if they want, which I guess you could consider a form of unfriending; but it's a form of unfriending that doesn't require a previous friending, and is used more to avoid spammers and scammers, than to indicate any particular relationship between two users.

But the other user does know when I've followed and unfollowed them on Buzz. So it's a lesser friending, in the sense that there's a lesser implicit rejection than if we had first agreed to be 'friends', and then I revoked that agreement. If I simply stop following them, it could be that I'm just trying to cut down on the information torrent, or that I've chosen to subscribe to their public Google Reader feed instead, or their blog. There are plenty of uncomplicated reasons to not follow someone on Buzz, that don't imply dislike.

The okcupid dating site has an interesting 'favoriting' feature. A dating site has a harder time letting people friend and unfriend each other, because a lot of times one's friends don't really want to know the ins and outs of each other's sexual and dating preferences. Trying to provide that kind of feature on okcupid would be a little weird.

So, that site has a number of other methods of creating a social network. One is to have a lot of life-related multiple-choice questions, and then create match scores between you and other people. When you view your match list, the highest-ranked matches form a kind of social network of people who are more or less similar to you in certain ways, even though you may not be acquainted with each other.

Another technique is the one I just mentioned, the ability to 'favorite' someone. Doing this in okcupid just makes that person appear in your navigation bar whenever they're online. This way when you're both online together, they appear to you as being online, and you can IM them if you want, using the okcupid chat app.

Users don't necessarily know that they've been favorited. The only visible change for the one being favorited, could be that certain other users are just more likely to chat with them. And if they are later unfavorited, they might never know it, and so they'd never have a chance to be offended by it.

When you favorite someone on okcupid, you can also choose to let the automated system tell them they've been favorited by someone. So by choosing that option, you're essentially waving at them and saying "hi, I'm interested in you, check out my profile, favorite me too if you want." But if either of you later unfavorite the other, neither of you would be informed of that. So this is another 'lesser friending' that avoids the friending problem.

For me, one of the most valuable features of a social network is one that isn't even implemented on any of them anymore. And it seems like a feature that might potentially justify the explicit friending handshake. Friendster had it in the old days, and then removed it, presumably because they didn't understand its tremendous value. And no other site has yet implemented it.

The feature involved the chain of friendships that are created when two people explicitly friend each other. If person A friends person B, and person B friends person C, then A and C are two hops away from each other. In other words, A and C are actually in the same circle of friends, even if they don't know each other - the circle of friends defined by person B.

Let's extend this outwards a bit. If person C friends person D, then A's friends and C's friends, are actually friends with each other. This is because B and C are the link between A and D, and B and C are friends. If we're considering the value of these connections in building trust between people who are otherwise strangers (A and D), then clearly A and D have a fairly solid reason to trust each other in this situation. People who are friends with each of them, are also friends with each other. It's very unlikely that if A were a safe person to know, that D would be a psycho ax murderer. So A and D can feel comfortable contacting each other for no particular reason and saying, "hi! I like your profile. Want to hang out?"

Let's go one more hop out, and say that person D friend person E. What is E's relationship to A now? Well, A is friends with B, and E is friends with D. Meanwhile, B and D are both friends with C.

So A's friends and E's friends (B and D), have friends in common (C). So, people in A's circle of friends, and people in E's circle of friends, are actually in the same circle of friends as each other.

That's a more tenuous, but still quite tenable, reason for A and E to trust each other. Their circles of friends virtually intersect. And when you take all the people who are likewise four hops away from A, you will tend to end up with a very large pool of people, maybe in the hundreds of thousands or millions. That large pool gives people some cover, so that if A makes a bit of a fool of themselves with E, news of this is less likely to trickle back to A's friends. So on the one hand you have a bit of trust with people who are four hops away, and on the other hand, you have a bit of freedom to take some risks and get to know someone.

It was a great feature. I have no idea why it went away, or why no other site implemented it. I met a lot of people who were four hops away from me, with no more initial trust than what that feature gave us.

But it's a good counter-example to my point about the friending problem. It's harder to implement a feature like that one (call it trust-in-chain), if you don't have an explicit friending between two people. Person A and person E can both feel confident that B, C, and D are all vouching for the connections they've made in the network. Their confidence in those connections is the only reason A and E can raise the trust between them in this example. Take away the explicit friending, and A and E have much less of a basis for trust.

To make that clearer, let's go back to Google Buzz. If A follows B, and B follows C, and C follows D, and D follows E, that's a very poor reason for A and E to trust each other. Maybe A could trust E somewhat, since that's the direction of the following, but since E isn't following D, and D isn't following C, and C isn't following B, and B isn't following A, and since there could be very good reasons why those things are true, there's really no reason for  E to feel any particular trust towards A in this situation.

I'm not sure if friendster's trust-in-the-chain feature could be implemented using 'lesser friending'; and if not, it definitely is an argument in favor of full-on friending on these social networking sites. But the friending problem still diminishes the value of the trust-in-the-chain feature, even if explicit friending is necessary for it. If A has friended B, but doesn't really have much connection with B, and is only leaving the connection in place because it would be awkward to unfriend B, then that makes the connection between A and E more tenuous. So the trust-in-the-chain feature would also benefit from solving the friending problem.

I'm not sure what, if any, solution there might be to the friending problem. But it seems like a real problem with social networks, and something that ultimately undermines the whole point of the network. I think it's a problem that, if a solution could be found, would dramatically increase the value of the social Internet.