Paul Revere, Social Graph, Speculative Writing

The readings about networks and graphs. Facebook has what it calls a social graph, which is the data it maps about all our connections. I can’t do the mathematics behind it, but it is potentially very powerful, as this post from Ditte shows. In a similar vein when the Snowden story broke recently there were arguments that if the government harvested all this information about you, and you weren’t doing anything wrong, then what was the issue. (We’ll put to one side questions about sovereignty, privacy, the assumption of privacy and so on.) Sociologist Kieran Healy, using a social graph, wrote an extraordinary (speculative – note it is framed as if written from London in 1772, calls its data set Bigge Data – as in ‘olde worlde’ – and mentions an upcoming EDWARDx – TedX – talk) blog post that used this same mathematics and theory to ‘prove’ that Paul Revere was a terrorist. For those that don’t know, Paul Revere was the person who rode through Boston (there is literally a line painted on the road, in Boston today, so you can retrace his famous ride) yelling that the “British are coming!” and alerting the American patriots to the oncoming British soldiers in the American Revolution. He essentially set up an intelligence unit. He is the American hero (patriot, solider, prosperous silversmith, Bostonian, subject of a famous poem), and as Healy shows, by using the social graph (nodes and links) you can demonstrate that Revere was a hub, and therefore a terrorist. As Healy writes:

What a nice picture! The analytical engine has arranged everyone neatly, picking out clusters of individuals and also showing both peripheral individuals and—more intriguingly—people who seem to bridge various groups in ways that might perhaps be relevant to national security. Look at that person right in the middle there. Zoom in if you wish. He seems to bridge several groups in an unusual (though perhaps not unique) way. His name is Paul Revere.

Once again, I remind you that I know nothing of Mr Revere, or his conversations, or his habits or beliefs, his writings (if he has any) or his personal life. All I know is this bit of metadata, based on membership in some organizations.

The point he is making is that just based on social links a lot of information is known, but then add one or two assumptions (as he points out, he knows nothing about these people) and it is easy for this information to shift from being information, to knowledge, to an exercise of unreasonable power.

Small World Networks, Scale Free, Kevin Bacon

My riff in response to Brian’s comment that the 80/20 stuff isn’t what really matters in the reading.

  • the internet is scale free – you can add and add to it and it doesn’t fill up (unlike a room, a book, film, and most other of our media)
  • it is made up of nodes (in social networks outside the internet these are people, in social networks inside the internet like Facebook these are generally people), which are small ‘things’ that can have connections to other similar things (friends, acquaintances, links from one web page to another)
  • preferential attachment means that some nodes are more likely to want to be connected to other nodes (in my academic hypertext essay one node got more links in and out because it turned out to the heart of the argument I was making, because it is was an essay this was why this one node was preferred, in a blog you might link to a blog that is authoritative (you value) in the field that you also write about, you might just link to a friend’s blog)
  • as a result of these three things hubs form, which have lots of connections in, and often out
  • interestingly hubs have very weak connections – you don’t know them (a strong connection)
  • and so a small world network arises

So it isn’t random, it isn’t disordered, it isn’t chaotic. A structure emerges that is understandable. But it emerges, the shape isn’t known in advance. This too, in many ways, is the opposite of what we think the world is.

A small world network means that because there are links, and hubs, it is quite simple to get from one point in the network to any other. Because there are densely connected hubs links follow a power law. A power law tells us that a few have a lot, but also that most of the material is in the tail, which is why niches now really matter.

More Richness

Denham has a post about the 80/20 rule and the inequality of the power law. Though the thing to take away is the long tail point, what lies in the tail is greater than what lies at the big end, so for online stuff something important is that while it might seem obvious that there are hubs (though it isn’t, why link to Google?) what is less obvious is the scale, and complexity, of the tail. Nga has a simple and useful account of why earlier web sites end up with more links to them than later ones (part of the explanation of preferential attachment). Lucy has a simple and elegant account for networks, centres and scale free networks.

Last Week.

Denham’s post from the unsymposium is worth a read, not just picking out the key take aways but providing some commentary on them too. The observation about film and hypertext and so much digital media making as a relational media is, I think, exceptionally important. The role of recommendations, and those systems that now elevate some people over others is what I meant about trying to work this out algorithmically. We know how to make recommendations based on things like what you buy compared to other people who buy similar things. But to do this just on comments we make is much harder – how do you tell who is more authoritative than someone else? The most common way this is being done at the moment is through peer review. I rank other people’s comments and those who consistently seem to be highly rated by others will be elevated in terms of authority in, and by, the system (this is essentially a slashdot system as they invented it). But there is a lot of time and money being spent on trying to solve this just on the stuff that’s already out there, without needing people to vote and rank.

Recommended

Molly picks up my post about recommendation systems and notes that she hates the ads on Facebook but likes Spotify. Exactly, the former is only selling ads, not recommendations of what other people like you liked. (Though imagine an ad engine that worked like that?!)

Anna D has notes from the unsymposium, including reputation networks, games and narrative. Gabrielle has three take away ideas. Hypertext and games, writing hypertext, and IBG (Internet before Google).

Hubs

Jackie has a nice post about hubs, networks and their emergence. Emergence is a specific term that means something that seems to be chaotic in fact has a structure, which emerges out of the complexity and chaos, over time. Molly as a good overview of hubs and networks and how they grow. Note, what is joined (linked) is not random, which means a link expresses an intelligible connection (e.g. this is related to that). And as a result clusters form. This happens on the web. It happens in hypertextual work that you make yourself. Molly also has a really good outline of power laws and bell curves. Kimberely notes the point that heavily linked nodes tend to be linked to more often. This promotes the power law imbalance, but also is why you link to the tail (so is for example one reason why these posts keep linking out to you). It really is a bit like not trying to be friends with the most popular kid in school but making friends with people less popular because they are worth knowing. Denahm has a really good illustrated post on this stuff, read it.

Daniel thinks about link decay. He’s right, links decay over time. The web is an amazing system which we take for granted, but it doesn’t break when pages or entire sites go away (and you get a 404 error message). Try reading a book with missing pages. A TV show with missing segments. It is an extraordinary model that can tolerate things dying and disappearing. Again, this is the opposite of previous media. Rebecca wants to know if a network has a boundary, and outer edge. The sort of network being discussed here is called a scale free network, and as the name implies, no. The web has no outer edge, there is no reason it can’t keep on growing.

Networky Networks

Cuong on the long tail, and particularly notices Anderson’s comments about scarcity. This is something that the first and second lectures picked up on, where I pointed out that I came to university because of scarcity (library, films, experts, technologies) but that this scarcity is gone, so why come now? Same argument Anderson makes about going to the video store. Tony discusses the long tail, and is surprised at the mention of Kazaa, this just shows how quickly things change here, where we often measure an internet year as a dog year, so 1 year online = 7 years in the real world. Why? This reflects the pace of change and development online. Anita discusses the 80/20 rule and scale free networks, wondering if they are natural. They are, which is part of what these people demonstrate, our bodies, for example, turn out to be scale free networks with a power law distribution in terms of the number of proteins (which are out basic blocks) and how many of these different proteins are involved in how many reactions. A small number (crazy small) are involved in a heap. Nature, including us.

More Voxies

Louisa has a bullet list of stations along the way. A semesters worth of material in 50 minutes. Patrick joins up NBN, infrastructure policy and network media. I’m with Patrick, the bigger, faster, more resilient it is built now, the better off we will be, it is the difference in defining useful as an extractive economy (dig it up, sell it) versus a knowledge economy. Olivia has another list of points from the unsymposium, very useful gloss. Millicent notices that she, like Brian, uses media ‘hypertextually’ (and so the debate that happens out in the real world is whether this is a good, or a bad, thing). Rebecca S thinks with her dad about Facebook, and dad points out that not very long ago MSN was all the rage (anyone remember MySpace?). My criticism of Facebook is that I think the network is the place for quality and niche, and I really really struggle to have that experience on Facebook. Let alone being inundated with dating ads (I’ve told them I’m married, and not looking, but they’re the ads I get??) One of my favourites is from Danielle with thoughts about games, stories, keyboards, recommendation engines and sharing the link love (link to others in the tail). This last point is incredibly important, it is what guarantees diversity and depth to the web – for all the reasons the last two week’s of readings have described. Closely followed by Lauren M who realised (very well done) that when I described hypertext as a post cinematic literacy, and that meaning is created outside of the shots, that what I was simply describing was the Kuleshov effect. Yep. Hypertext figured this out quickly, most other interactive media hasn’t. Rebecca M has another node come load of dump notes from the unsymposium… More to come…

Networky Stuff

Lucy has comments on the 80/20 rule, and its relation to the Web where 80% of links point to 15% of pages. This is why linking matters, it is how you build and nurture the long tail (and that the tail is where immense niche value lies). Lucy also discusses the second reading noting how the web isn’t static, its structure changes over time, and that hubs and connectors are important attributes of these sorts of networks, which occur in nature and online. Prani discusses the long tail, though it isn’t so much about being niches as that niches become viable in dramatically different ways courtesy of power law distributions and the long tail. Lauren tackles power laws, awkwardly but the discussion is good (must be the science). Nga on the long tail, with links to two useful clips. Danielle on the long tail, recommendation systems, and supply and demand, Tamrin with more detail on retail, long tail and the marketplace. Rebecca S has a joyfully scattered meander about long tails with various swishes along the way, it’s an excellent read. And Lauren M has notes from Watts and networks and nodes. there are lots of questions in other posts about why. Hope we get a chance to colour that in.

Small World Networks

Chantelle looks at the Barabási readings on networks, paying particular attention to the 80/20 rule (which also describes blogging activity in network media). Kevin also discusses the 80/20 distribution, power laws, preferential attachment and growth. Preferential attachment and growth are important because it is how networks work, and also before being figured out our models for networks tended to think of them as static, and random. Olivia picks up the stuff about power laws, which is the maths behind the long tail. Anna D has a nice post about getting lost and bored, going elsewhere, returning, then a moment where things went click. Louisa discusses 80/20 and wonders if 20% of employees really do generate 80% of profits. Certainly at a university there would be an affinity between 20% producing 80% of the research, and I would suspect research income. Also with the blogs I don’t bother with accurate numbers but there would certainly be no more than 30 very active bloggers in network media out of 130, and in tutorials it is clear that most questions come from a very small group… Samuel mentions power laws and links to A list tech and tech culture blogger Jason Kottke’s post on power laws. Brittany discusses the economics of the long tail, and repeats Anderson’s ‘three rules’.