A Better Social Networking Algorithm
Those who have been reading my scribblings for a while know that I'm a declared skeptic of Social Networking Software. The main reason is that it seems to provide relatively little benefit for the effort to sign up and maintain the network, and that what benefits it does provide are often already being provided elsewhere.
That said, there is a better way. Most Social Networking Software today works by tracking Friend-of-a-Friend (FOAF) links. This works on the principle that "a friend of my friend is also my friend." So, if I declare all my friends, and all my friends do the same, then I've uncovered an exponentially larger number of friends-of-friends.
This approach has a number of problems. To begin with, many people declare as "friends" people who are, at best, acquaintances. This is done in an effort to appear more socially connected, and because of the ambiguous nature of friendship. Nevertheless, an acquaintance of an acquaintance is....nothing.
In addition, the underlying premise of a FOAF scheme is often false. Nearly every social circle includes at least one instance where two people who loathe each other also share several common friends. FOAF would mistakenly tag the rivals as second-order friends. even though they can't stand the sight of each other.
The Enemy Of My Enemy Is My Friend
These problems can be avoided by adopting a different scheme, one which tracks enemies-of-my-enemy (EOME). This is based on the long-established principle that "The enemy of my enemy is my friend." Just as with FOAF schemes, this can be tracked to higher orders: "the enemy of the enemy of my enemy is also my enemy" and "the enemy of the enemy of the enemy of my enemy is my friend" and so forth. Care must be taken to track the order of the relationship, to avoid mistakenly assigning an enemy as a friend, or vice-versa. Even-order (second, fourth, etc.) relationships in EOME imply friendship, whereas odd-order (first, third, etc.) relationships imply enmity.
This creates a much more robust system than FOAF for several reasons. First, where friendships can be ambiguous and very tenuous, most people hate their enemies with real passion. Thus, the link is stronger.
In addition, because of the social stigma of declaring someone an enemy, participants won't pollute the Social Networking Software by declaring bogus enemies, the way they often declare people they barely know as friends.
Finally, EOME overcomes the primary limitation of FOAF, which is that it allows for mutual dislike. Let's face it, we don't live in a Smurf-and-Care-Bears world where everyone likes each other and we all hold hands every night and sing around the campfire. In the real world, some people really dislike each other, and any realistic Social Networking Software needs to allow for true loathing.
Practical Considerations
Because of the unique properties of EOME networks, there are some issues which need to be treated with care in any practical implementation.
1. Privacy. As Richard Nixon understood well, any "enemies list" needs to be private. Revealing an enemy has the potential to be socially awkward, and in the worst case, can lead to broken marriages, lost jobs, and so forth. This is especially true when the enmity is covert and not reciprocated. For example, I would not want my wife to know that she is on my secretary's Enemy List. This would lead to all sorts of awkward questions. As a result, the EOME-based Social Networking Software can only reveal even-order relationships; and even then, should probably not reveal the name of the mutual enemy.
2. Ambiguity. Unlike with FOAF where every relationship implies friendship (which we all know is highly unrealistic), EOME can imply either enmity or friendship. Where there are multiple EOME paths connecting two people, the relationship can be ambiguous. For example, if one of my enemies is also the enemy of another of my enemies, is that person a friend or an enemy? To disambiguate these relationships, the lowest order connection is considered the correct one. However, it may be acceptable to reveal that two enemies also share a mutual enemy, which might lead to some reconciliation.
3. Directionality. Not all enmity-based relationships are reciprocated. For example, I may loathe Bob, but Bob thinks I'm an OK guy. For this reason, it is important to track the direction of enmity (I hate Bob, but Bob does not hate me), and ensure that the implied friendships are based on proper loathing of a mutual enemy. So, if Joe also hates Bob, that implies that Joe and I are friends. However, if Bob hates Joe, but Joe does not hate Bob, there is no implied friendship since Joe and I don't share a common enemy in Bob.
Other Implementations
There are at least three known attempts to implement EOME: hatester, fiendster, and enemyster. However, from the data available publicly, it is not clear that these take into account all the subtle nuances of a proper EOME implementation as I listed above. In fact, it isn't clear that any of them is actually a serious attempt at true Social Networking: it may be that they're all intended to be some sort of satire. If so, that would be disappointing, since EOME is such a promising concept in Social Networking Software that it deserves a more serious and comprehensive approach.
Indeed, in my opinion, EOME stands at least as much chance of becoming commercially successful as FOAF.