Spent the morning reading a well argued rant by Maciej Ceglowski titled The Social Graph is Neither where he argues that the current way human relationships are modeled on sites like Facebook is fundamentally flawed. He makes two basic arguments. The first
It’s Not a Graph There's another fundamental problem in that a graph is a static thing, with no concept of time. Real life relationships are a shared history, but in the social graph they're just a single connection. My friend from ten years ago has the same relationship to me as the friend I dined with yesterday. You're left with forcing people (or their software) to maintain lists like 'Recent Contacts' because there is no place in the model to fit this information. "No problem," says Poindexter. "We'll add a time series of state transitions and exponentially decaying edge weights, model group dynamics as directional flows, and pass a context object in with each query..." and around we go. This obsession with modeling has led us into a social version of the Uncanny Valley, that weird phenomenon from computer graphics where the more faithfully you try to represent something human, the creepier it becomes. As the model becomes more expressive, we really start to notice the places where it fails. Personally, I think finding an adequate data model for the totality of interpersonal connections is an AI-hard problem. But even if you disagree, it's clear that a plain old graph is not going to cut it.
It’s Not a Graph
There's another fundamental problem in that a graph is a static thing, with no concept of time. Real life relationships are a shared history, but in the social graph they're just a single connection. My friend from ten years ago has the same relationship to me as the friend I dined with yesterday. You're left with forcing people (or their software) to maintain lists like 'Recent Contacts' because there is no place in the model to fit this information.
"No problem," says Poindexter. "We'll add a time series of state transitions and exponentially decaying edge weights, model group dynamics as directional flows, and pass a context object in with each query..." and around we go.
This obsession with modeling has led us into a social version of the Uncanny Valley, that weird phenomenon from computer graphics where the more faithfully you try to represent something human, the creepier it becomes. As the model becomes more expressive, we really start to notice the places where it fails.
Personally, I think finding an adequate data model for the totality of interpersonal connections is an AI-hard problem. But even if you disagree, it's clear that a plain old graph is not going to cut it.
Here I think Maciej is looking at the problem from the wrong end. The question isn’t whether we can perfectly model the real world in software but instead whether we can use software to improve the quality of our lives by solving real problems. I think of this as the Xanadu vs. World Wide Web problem. You can point to a dozen problems that exist on the web as designed today from broken links and the frailty DNS to the problems caused by anonymity such as spam and phishing. This hasn’t stopped the web from becoming the center of an $8 trillion economy because it solves a lot of human problems even though it is imperfect.
Thus the question isn’t whether a product solves a problem of how to differentiate between my high school friend I haven’t seen in 10 years who I still think of as a brother from the coworkers I communicate with regularly but have no real interest in their lives outside of work. The questions are actually (i) can a product add value to people’s lives without needing to add the complexity of modeling that abstraction and (ii) does the benefit of the “improvement” of solving that problem outweigh the costs it introduces to the system?. I think the answer to the first question when it comes to the social graph is clearly “Yes”. As for the second, we really don’t know the answer to that one but from seeing the various imperfect attempts like Circles in Google+ I would bet the answer will be “No” for a long time. Which brings me to Maciej’s second point.
It’s Not Social The problem FOAF ran headlong into was that declaring relationships explicitly is a social act. Documenting my huge crush on Matt in an XML snippet might faithfully reflect the state of the world, but it also broadcasts a strong signal about me to others, and above all to Matt. The essence of a crush is that it's furtive, so by declaring it in this open (but weirdly passive) way I've turned it into something different and now, dammit, I have to go back and edit my FOAF file again. This is a ridiculous example (though it comes up with strange regularity in the docs), but we run into its milder manifestations all the time. Your best friend from high school surfaces and sends a friend request. Do you just click accept, or do you send a little message? Or do you ignore him because you don't want to deal with the awkward situation? Declaring connections is about as much fun as trying to whittle people from a guest list, with the added stress that social networking is too new for us to have shared social conventions around it. … Social graph proponents seem uninterested in the signaling problem. Leaving aside the technical issues of how to implemented, how does cutting ties actually work socially? Is there any way to be discreet, for example, or have connections naturally degrade over time? In real life, all relationships fade naturally if you don't maintain them, but right now social networks preserve ties in amber until we explicitly break them. Is my sister going to resent me if I finally defriend her annoying husband? Can I unfollow my ex now, or is that going to make her think I'm still hung up on her? There's no way to take a time-out from our social life and describe it to a computer without social consequences. At the very least, the fact that I have an exquisitely maintained and categorized contact list telegraphs the fact that I'm the kind of schlub who would spend hours gardening a contact list, instead of going out and being an awesome guy. The social graph wants to turn us back into third graders, laboriously spelling out just who is our fifth-best-friend. But there's a reason we stopped doing that kind of thing in third grade!
It’s Not Social
The problem FOAF ran headlong into was that declaring relationships explicitly is a social act. Documenting my huge crush on Matt in an XML snippet might faithfully reflect the state of the world, but it also broadcasts a strong signal about me to others, and above all to Matt. The essence of a crush is that it's furtive, so by declaring it in this open (but weirdly passive) way I've turned it into something different and now, dammit, I have to go back and edit my FOAF file again.
This is a ridiculous example (though it comes up with strange regularity in the docs), but we run into its milder manifestations all the time. Your best friend from high school surfaces and sends a friend request. Do you just click accept, or do you send a little message? Or do you ignore him because you don't want to deal with the awkward situation? Declaring connections is about as much fun as trying to whittle people from a guest list, with the added stress that social networking is too new for us to have shared social conventions around it. … Social graph proponents seem uninterested in the signaling problem. Leaving aside the technical issues of how to implemented, how does cutting ties actually work socially? Is there any way to be discreet, for example, or have connections naturally degrade over time? In real life, all relationships fade naturally if you don't maintain them, but right now social networks preserve ties in amber until we explicitly break them. Is my sister going to resent me if I finally defriend her annoying husband? Can I unfollow my ex now, or is that going to make her think I'm still hung up on her?
There's no way to take a time-out from our social life and describe it to a computer without social consequences. At the very least, the fact that I have an exquisitely maintained and categorized contact list telegraphs the fact that I'm the kind of schlub who would spend hours gardening a contact list, instead of going out and being an awesome guy. The social graph wants to turn us back into third graders, laboriously spelling out just who is our fifth-best-friend. But there's a reason we stopped doing that kind of thing in third grade!
I agree with the core premise that relying on people to explicitly declare and maintain their relationships in minute detail is a flawed enterprise. Maciej is 100% right that there are explicit social signals with real consequences when we decide to declare even in the most mundane of public ways that we are connected to others. My favorite example is the amount of scrutiny applied to Rep. Anthony Weiner's Twitter following list and how many point to the his explicit declaration of interest in a set of female followers as a catalyst which helped further his fall from grace. The thing I like about that example is that the nuance of simply declaring what kind of relationship you have with people on Twitter or categorizing them would have not helped in that situation.
The reality is that the publicness and interconnectedness of the World Wide Web is causing us to create new social norms which we are still figuring out. The same way people who leave rural areas for cities realize that some social norms they’d grown up with all their lives would now have to change so is humanity figuring out as we go along how we need to adjust our behavior and in some cases broaden the list of behaviors we accept as we become more interconnected online. In the past few years, we’ve seen Zuckerburg's law of information sharing applied in the real world and it’s amazing to think of the sorts of things people regularly share and expect to be shared with each compared to just a few years. It’s hard to believe that it was just five years ago when Time magazine was writing about how Facebook’s 8 million users would abandon the site because of the “intrusive” news feed, now Facebook has 800 million users thanks to that intrusive feature.
Now Playing: Lil Wayne - It's Good (featuring Drake & Jadakiss)