The Social Graph is Neither | Hacker News Reader

The Social Graph is Neither

blog.pinboard.in

434 pointsconesus60 comments

This is a superb post. Social networks, starting especially from the era of the Facebook era, have foisted increasing amounts of doublethink on their users, and it's often difficult to make people realize this and explain the reasons for it being that way. This essay not only lays out good examples of that nagging cognitive dissonance, but then goes on to explain its necessity at both the technical and business levels.

I disagree. The entire first point is merely a taxonomic complaint. Yeah, taxonomies are hard to "get right" in the global sense. That's why we don't employ globally correct architecture-astronaut taxonomies for everything, there aren't very many accurate useful ones.

Instead life is made up of piles and piles of little taxonomies used for specific purposes. I'm happy with that as both a mental model and a practical one.

I agree that the notion of a single social graph is an incoherent one, but Maciej's point is largely a quibbly one. There are graphs of human interactions that can be mapped out. These interaction graphs are not static. Data sets which purport to represent reality require updating, as they are not the canonical source.

But so what? That's true for maps of physical locations too, or article quality on wikipedia. Facebook shows, if nothing else, that it's possible to not just track but mediate (and still track) these interaction spaces for users as an end to itself (even if they are the product not the customers).

Does it matter that Facebrick isn't a perfectly synced representation of reality? Well, does it matter that Wikipedia isn't a totally accurate encyclopedia? You'd be right to ask for what cause, and to what end.

And that's the whole point. We dig up or create the little taxonomies that bridge the gaps that we're looking to cross.

I think you're missing his bigger insight, that is, building a social graph is a fundamentally asocial act. And that extends to creating taxonomies, even if they are localized to my concept of socializing. Nobody goes to a party to create a taxonomy.

Being social is all about disorganization, spontaneity, creativity, breaking out of all the stupid taxonomies that we impose on other people and have imposed on us. Trying to organize socialization to the extent that a "social graph" can be created is a profoundly sociopathic act. To push this task on users so that they do all the shoeboxing and tagging and linking, all in the name of "fun," requires some coercive force. It is no surprise that many Facebook users say that they don't really enjoy using it, but they have to. Part of that was the site's smart choice to maintain real names, so that you are always fretting about your online appearance. It's not a "perfectly synced representation of reality," but it is presented as such, almost like your own personal Wikipedia page, and by the laws of social anxiety you are compelled to check up on it! Shoebox, tag, link...

The line about Facebook rarely producing creative or original material really hit home for me. 4chan and SA are creative because they are fundamentally disorganized, and this freedom produces real moments of brilliance (and plenty more of profound stupidity). Facebook and its ilk are like corporate jobs disguised as consumer web products. You are constantly being shuffled from box to box, task to task, and the blinders have to be tightened ever so constantly to keep you from realizing why what you are doing really never will be "fun" in the way that a raucous party is fun.

About those blinders: those are in a big way the "privacy options" that every social network touts. Maciej points out, and I think that it's a salient point, that if everybody was able to see the whole social graph, it would dissolve. That alone should suggest that the social graph is an unnatural concept. Yet there is somebody, the owner of the social network, that can indeed see the entire graph, and there is a whole level of misdirection that they must apply to make this seem completely normal.

Again, maciej said most of this better than I could, so I'm glad he did.

> Being social is all about disorganization, spontaneity, creativity, breaking out of all the stupid taxonomies that we impose on other people and have imposed on us.

That is a very euro-centric, post-modernist way of thinking.

What are caste system if not social?

What are far-flung familial relations if not social?

A definition of being "social" which only applies to societies where friendship is paramount, and friends come and go with ease is too limiting and leads you to solutions that can be awkward in other societies.

For example, in my country, we care deeply about family and ancestory--so much so that in every generation one member of the family is nominated to be the "memory keeper" (sorry, there's no translatable word) who keeps track of the geneology of everyone, and is expected at a glance to determine if a stranger is at least vaguely related to us.

Would a free form social graph be useful to us---not really. Our facebook pages are practically 90% family members, 10% people who through marriage will become family members.

> That is a very euro-centric, post-modernist way of thinking.

You're right, even the way I defined "social" is too limited. But that irony only helps my point. You cannot define social, but this is precisely what the "social graph" is attempting to do. As soon as you create a definition, people react to it and oftentimes rebel against it. And then you are right back to the drawing board.

I agree with you but that doesn't make the creation of social graphs a waste of time right? It just means the market has a lot of niches for different types of graph that each provide limited and specific value.

It's not about creating perfect Platonic models. It's about making cool things that people find useful. If you are trying to design a system that scaled to every single possible interaction from scratch then you've already failed, you can't compete with reality at that one.

Two things struck me:

- the quality of thinking displayed in this post

- what is in the world is it doing on the delicious clone I signed up for?

This is the same guy that writes about scaling hardware, no? Keep it coming.

Scaling hardware, steaks[1], Iceland[2], John Titor[3], scurvy[4], etc.

[1]:http://idlewords.com/2006/04/argentina_on_two_steaks_a_day.h... [2]:http://idlewords.com/2003/11/a_morning_in_iceland.htm [3]:http://idlewords.com/2003/09/best_practices_for_time_travele... [4]:http://idlewords.com/2010/03/scott_and_scurvy.htm

Posts from Maciej and Bret Victor on the same day, goodbye productivity...

Social networks aren't really social networks either. If you read Bowling Alone, it should be readily apparent that Facebook et al. fulfill few of the functions of an actual social network.

Exactly. "Social networks" are thingies that aid us with establishing and keeping up with actual social networks. It's a bit like calling a publisher's press a news agency.

In vipassana meditation there is the idea that if you learn to recognize thoughts at the precise moment they arise, then eventually any given thought will basically lose its power and stop arising. Facebook seems to have the same property, but for personality. Every time you post about yourself, you become a little bit more bland and boring of a person.

The way I think of Facebook is rolodex + fax machine + succubus.

Fascinating tidbit on meditation, but I disagree that posting about oneself on Facebook makes one more bland and boring. If, for example, one is bland and boring to start with, then yes, posting about yourself to Facebook won't really help. However (EDIT) -- if your life is boring, you can post about something else. I've known some individuals who lead the most pathetic lives you can imagine yet they had a decent-size following on Facebook because they regularly posted about politics (which I guess they used as a medium through which to vent about how pathetic their lives were...).

I do see some truth however to your comparison with that idea from meditation. For example, even if you are leading an interesting life, there are many ways to post about it that will make your life appear less interesting than it really is.

"Imagine the U.S. Census as conducted by direct marketers - that's the social graph."

That's the best quote in the whole post. I can't articulate it, but that resonated strongly with me.

One of the main points in this article is that any social graph found in a service such as Facebook, Twitter, or G+ can't possible model all of the nuance of real life relationships. That's obvious. However, the author takes it a step further. Suggesting that because of this these services have no value.

That's pretty big logical leap.

An imperfect approximation of the social graph still has a lot of value.

Exactly, I mean, just ask the citizens of Egypt.

"This revolution started on Facebook." - Wael Ghonim

http://www.huffingtonpost.com/2011/02/11/egypt-facebook-revo...

> Asking computer nerds to design social software is a little bit like hiring a Mormon bartender.

exactly.

This is bullshit.

"Computer nerds" have plenty of friends. I know factory workers that have very high depression levels due to not having enough social contact or friendship in their lives. Sure, when I was 9 I was beat the fuck up over and over but when you are friendly, smart and successful people like you. They want to be your friend. Let's play devils advocate for a second. Take the most popular person in high school and get him to design social software. It would suck. Bad.

> Take the most popular person in high school and get him to design social software. It would suck. Bad.

Let's assume they had the technical skills to do it, why do you think it would suck?

Let's assume they had the technical skills to do it

So basically a nerd with social skills?

I'm trying to figure out his argument. He's implying that someone with good social skill would suck at designing a social network, but seemingly for more reasons than they just don't have the technical ability. Or maybe that's all he's saying.

I'm denying the premise.

The most popular person in high school was almost certainly not a nerd. Nerds take a while to maxing out their social skills, but once they do I find they are almost as good at it as sales guys or MBA gals.

From my experience the most popular kid almost never had the time to dedicate to crafting the technical skills necessary to build a mega-successful online business, Social or otherwise. Nerds think about things far more than the typical prep or jock.

I think it was Adam Carolla who said that the guys who get laid consistently in high school don't go on to achieve anything. They don't have the drive.

He's implying that to build a social network, you would have to be technically adept enough to do it properly and nicely.

If you have the chops to do it, you're almost by definition, a technology nerd of some kind. (Yes, the 'popular' and 'nerd' sets are not mutually exclusive so I guess you have a point but you know what I'm getting at, right?)

Please. For the purposes of this conversation a "nerd" means what the OP wants it to mean, which is at least "someone who knows what a 'graph' is".

Whereas the technical skill required to build a basic electronic social network is "can you dial a phone?" Been that way for sixty years and more. For a dramatic musical rendition of the social networks designed by your parents (oh, wait, I mean your grandparents, maybe even your great-grandparents, how time flies) watch the first ten minutes of Bye Bye Birdie. Ooh, look, a video:

http://www.youtube.com/watch?v=gKhR8QtQ4do

(I was never a fan of Bye Bye Birdie, by the way, but you can't forget that number once you hear the darn thing.)

Now, these days every eight-year-old has mastered a far greater technical challenge: Mobile text messaging. That's a better social technology because you don't have to talk, or say "hello" or "goodbye", or avoid communicating in seventeen directions at once.

There are things Twitter and Facebook can do that can't be done with chains or trees built on one-to-one texting, but I'm not sure any of them are more social. (Is anything more social than the game of phoning people and asking them to relay messages?) Really, the point of these higher-tech "social networks" is that they fill a gap by letting you be barely social and yet still be present in a bunch of people's lives. I can follow your tweets without you even knowing. I can add Facebook "friends" and then basically ignore them most of the time. I can be "introduced" to a friend-of-a-friend on Facebook without any of the real stress of even talking to them, let alone meeting them in person.

I'm not sure what you're trying to address here. I don't disagree with any of what you said because it doesn't seem to address my point.

You're giving examples of people using social networks. I (and the person I'm responding to) am talking about people building the infrastructure to host social networks. Using your examples, the inventor of the phone and sms would be more appropriate.

Let's assume they just have to describe in it sufficient detail to someone with technical skills who would implement it for them.

I'm pretty sure it would still suck because they wouldn't know to translate their intuitive understanding of how to leverage social dynamics to gain popularity with how to facilitate them in software. But it's hard to be sure, and would depend on the latent design skill hidden in your "popular person".

I think most people don't know how to approach software design, even from a UI perspective. It's hard even for professionals. You would have to get lucky with a person who happens to get it intuitively.

Take the most popular person in high school and get him to design social software. It would suck. Bad.

What you've just described is Facebook, except that the "most popular person in high school" got cut out of the loop by the nerd smart enough to realize that the idea was good enough to be worth stealing.

And you're right. It does suck, in a million ways, reflecting our society at large rather well, overall. But it's also the most successful social network to date.

But in my opinion, at least, the main idea that the yacht club twins brought to the mix was actually the primary reason that it grew so easily: Facebook was, initially, at least, a social network for students at elite schools, let the rest of the University-of-CrappyPartySchool rabble diddle themselves on Myspace. Without that, I don't know that it ever would have taken off the way it did...

It's not a question of whether they have friends or not, it's whether they are nerdy enough to imagine that the idea of friendship can be captured by a simpleminded data structure. (And yeah, not all nerds would actually believe this).

It's the same thing that drives (some) nerds towards libertarianism, a fascination with formal models and rules over the messy complexity of reality, especially social reality.

I did like that sentence, but I can't help but think it's wrong. The author of this article is probably as much of a "computer nerd" as any person at Facebook/wherever, but they can identify all sorts of problems with the "social graph", so the problem isn't nerdiness. The problem, as the author also points out, is that real life is very complex, and you can get an 80% solution by dramatically simplifying it.

Besides, I suspect there are plenty of nerds who understand social relationships in theory very well (there's plenty of literature out there for those willing to read). Not to mention that, apparently, not all tech people are completely socially awkward.

I think Maciej is brilliant and so is this piece, but I do wish he didn't so often play the "you're all dysfunctional nerds" card.

Also, there's an unstated assumption that in order to write software that is useful socially, one has to be very social. This sounds like the arguments people used to have in the mid-90s about whether a program could ever replace a librarian. Well, it didn't -- but it didn't need to.

I think it's clear that what we call social networks don't model our friendships well. That doesn't mean they aren't useful in some other way. Perhaps the great shift will come when we stop trying to make social networks be our Advanced Friendship Substitute, and instead make them do something more useful.

> Also, there's an unstated assumption that in order to write software that is useful socially, one has to be very social. This sounds like the arguments people used to have in the mid-90s about whether a program could ever replace a librarian. Well, it didn't -- but it didn't need to.

Another example: programmers of Deep Blue didn't have to be good at playing chess. They just had to understand the rules and the concept of search trees.

this quote backs up your assertion quite well: "The idea of FOAF was that everyone would create little XML snippets that represented their interests."

I disagree.

It might be selection bias on my part, but as far as I've seen if someone is able to comprehend a concept like social graph, he's on his way to becoming a nerd. Even being able to handle social relationships on abstract level[1] require insights of mathematical concepts, on the level that is not taught to majority of human population.

That's why I dislike all those 'nerds suck at XXX', whether it's about maths, physics or computer nerds. All of those groups are good at maths, which is basically applied thinking, and thus is applicable not only to computers, but to real life as well. So, given a problem and no domain expert around, I'd definitely trust nerd or an engineer more than anyone else.

[1] - as in, go meta and talk about things like social graphs, how they evolve, transform, while having a clear model in ones head.

Ugh, I hate titles like these.

It's not a graph? Of COURSE it's a graph. It's just a very complex one. Just because you could make many such graphs representing the same relationships doesn't mean it's not a graph.

And it's not social because you're missing someone's crush? Uh, no. It's still social, just incomplete.

A better title for this would have been "The Social Graph is incomplete and complex."

A better title for this would have been "The Social Graph is incomplete and complex."

A more accurate title perhaps, but definitely not a better title!

Yes, it's like he wanted to complain about Facebook, came up with a really awesome title, and then had to backfill content.

Which utterly failed, wait, didn't even attempt to support the title of his post.

I wrote a rebuttal titled The Social Graph is Both http://lesspostmoreget.com/2011/11/09/it-is-a-graph-and-it-i...

When you publish content on the internet and it has a relation to your online identity you are just saying: look how awesome I am.

I uploaded a bubble sort in prolog and a b-tree in Haskell ("I'm an enlightened far-reaching hipster hacker").

I quoted Allen Ginsberg on facebook. "Look at me I'm smart and read poetry".

MySpace collapsed and Facebook's clean UI won because most people don't have that much identity.

Hmm,

I seems like the post is saying "The Social Graph" isn't a graph 'cause it is a multi-graph with different kinds of connections. That part seems obvious.

But the "it's not social part" is more saying "it's not something to publicly mess with", ie, take the question of signaling interest seriously. That's true but the thing remains a social graph.

I would put it another way. The social graph exists, touching it can be dynamite, yes, can breach some boundaries, yes, publishing a connect is further social act, yet. But technology is about breaching boundaries. Facebook doesn't allow the touchy, fine-grained quality of real world friendships - and there are advantages to this. It has created a lot of connections which wouldn't exist before it. Sure, further refinements may make things more nuanced as in the real world. But the crude, glad-handing Internet world is now also the real social world and won't be going away. There are ways that this is quite good.

I agree that the graph is an incorrect representation. Because people don't think that way. But that doesn't mean organic, activity/interest based communities are the only ones possible, or even desirable.

I want a way to be able to communicate with all of my different offline groups without having to be in their physical presence. I also want this on a broad spectrum of public-private.

I think adequately (not perfectly) modelling social context is a solvable problem. It's a matter of UI and metaphor. Here's my take on it: http://blog.byjoemoon.com/post/7072771434/a-new-metaphor-for...

I put my thoughts on this article in a blog post. Too long for a comment.

"The Social Graph is Both"

http://lesspostmoreget.com/2011/11/09/it-is-a-graph-and-it-i...

Good essay, thanks.

>They are useful graphs with deficiencies.

This describes all graphs representing real-world, as opposed to purely mathematical, objects and relationships.

Lots of good, interesting ideas in that post.

A few comments: In general, what data you capture, always depends on what you want to do with the data. Any data model is an abstraction, which discards a lot of information. This simplification is what makes the data model useful.

Of course, storing social relations as an undirected, graph with a single type of edge, as is done in Facebook, is a gross simplification of real social structure. Everyone knows that; Facebook know that; but the fact is that you can build a lot of useful stuff, with even such a simple model. Is it perfect? No - I'm sure we will do better in future - but you've got to admit its amazingly successful, for such a simple representation.

So, we aren't going to be able to define a format for the social graph that is so rich, that it will capture all possible uses with it - but I don't see why we couldn't define a format that captures enough detail to do many a great many of the things we might want to.

Next point: graphs are amazingly flexible data structures; if you have a multigraph or hypergraph representation, or a set of graphs, you can represent an awful lot of rich information. Don't knock graphs too much. You could build an awful lot of cool communication functionality, if you had really any rich social graph representation, not necessarily the 'perfect' one.

The author also writes: "In other domains, a big graph would be good for recommendations, but friendship is not transitive. There's just no way to tell if you'll get along with someone in my social circle, no matter how many friends we have in common." Of course, friendship isn't strictly transitive; otherwise the giant connected component of a social graph, would be one big clique; but friendship is highly transitive. That's why recommendation works so well in Facebook. If A is friends with B, and A is friends with C, then there is a vastly higher chance that B is friends with C, than with some other random person. Sociologists have called this 'triadic closure' and studied it for years: http://en.wikipedia.org/wiki/Triadic_closure

So, while the recommendations won't be perfect, as things like friendship aren't perfectly transitive, they'll potentially be very good. But what recommendation engine is perfect, anyway?

Finally, the author refers to the problem of finding an adequate model as 'AI-hard' - well, as I've said, any model - even the most perfect one (unless we go and make a copy of the entire state of the world) - is going to sacrifice something. So I see why coming up with a perfect model is impossible. But as to coming up with a model that is adequate for a great many of the possible things we might want to build; well, I'd say 'AI-hard' is an exaggeration.

Facebook has a heck of a lot more info on relationships than an undirected graph with a single edge type. In the front end alone, they now have a directed "follows" relationship. In their backend, they're keeping track of how much you pay attention to people, comment on their posts, click their links, etcetera, and using it, among other things, to decide what shows up in your news feed and what is silently dropped. Someone posted a link here not too long ago describing how to get the weights Facebook ascribes to your social connections, based on the above sort of metric.

Yes, of course they do; they have vast amounts of data; click streams, likes, comment text, etc.

But I guess I'm saying that the core metaphor, with which Facebook did much of its early growth, was that two people are 'friends', or are they are not; and that you see the activity of your friends, and sometimes see activity, and get recommendations, from the friends of your friends.

They built many iterations of the product, and got a very long way, by exposing a very simple social graph metaphor to their users. Even though that model drastically simplifies the world, it has proven to have a lot of utility for users.

Maybe now, because the users are consequently more educated about social networks, a more sophisticated metaphor would go further.

Thanks for mentioning triadic closure. It's a great example of how useful information can be inferred/suggested based on an incomplete or imprecise model. The sacrifice of simplifying the social graph is what makes the task of constructing this graph online approachable.

I hope Maciej is not going soft on the trademark "brevity is for the weak" style, which I've been appreciating since idlewords.com days.

I believe he spared the spelling out of the LOAF acronym, and scanning the very interesting http://loaf.cantbedone.org/about.htm did not clear that up. "List of a friend"?

rdf technology is coming on strong in late 2011; the tools are getting better fast -- we will be able to represent the social graph in more detail, and correlate it with the 'semantic graph' that shared human experience is coded in.

yes, RDF has taken a while to mature, but look at the time lag between Codd's paper and the commercial release of Oracle. 1000 flowers have been blooming for years and we know a lot about what works and what doesn't.

the trouble with foaf and other 'distributed social graph' is that muggles like Facebook the way it is... they don't care about data portability or privacy; unless a distributed system can provide a better user experience, it's got no hope.

I believe Maciej is saying that it's not just an issue with RDF or a technical problem really. He says: "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."

That depends on your definition of a "plain old graph". If you allow arbitrary numbers of edges between nodes and allow edges to carry data, everything will be ok, right? (...right?)

In math... and we're talking math if we're talking RDF & graphs, right?... "arbitrary" is perilously close to a synonym for "infinite" and "unbounded".

Well, are relationships finite-dimensional?

That's true, but "infinite" and "unbounded" often also mean "there's no exact or precise solution, but at least for certain circumstances there's a useful approximation that works".

I can calculate pi to "enough" precision, whether I'm making a hula hoop, or a piston ring, or sending a probe to Saturn - I suspect a lot of what I'd like out of a functional "social graph" probably doesn't need much more precision than "three and a bit", 3.1416+-0.0004, or 3.14159265+-0.000000005.

new entry into the market, Stardog from Clark and Parsia, extremely fast and embeddable / scalable RDF database with OWL reasoning built in (they created the Pellet OWL reasoner)

I agree - 2011 going into 2012 is the rise of the semantic web

rdf technology is coming on strong in late 2011; the tools are getting better fast

Can you expand on this? I'm very interested in RDF but have been frustrated by how little interest in Semantic Web tech there seems to be in the US. I'm interested in highly specified networks rather than slippery ones from the consumer marketplace, and in visualization/manipulation tools rather than the underlying DB technology. If you have any suggestions I'd be grateful, I feel like I've been plowing a lonely row with this stuff.

Many things can be usefully represented by graph structure, and RDF is a way to do that (a verbose and clunky way, but better than no standard at all).

Whether social relationships are one of those things is a separate question.

Well said. The ideal "social graph" is just an advertiser's wet dream, and they forget to consider whether it would be useful to the user. Or even possible.

Russian guys on the photo

""" Right now the social networking sites occupy a similar position to CompuServe, Prodigy, or AOL in the mid 90's. At that time each company was trying to figure out how to become a mass-market gateway to the Internet. Looking back now, their early attempts look ridiculous and doomed to failure, for we have seen the Web, and we have tasted of the blogroll and the lolcat and found that they were good. """

i've been saying this for ages but this guys puts it so much simpler i'm ashamed.

discuss.

Reply on news.ycombinator.com