Facebook is a news feed. Amazon is a storefront. Google is a search bar. These multibillion-dollar companies work hard to look simple. But of course, they’re not. They are built on global server networks, endless algorithms, constant data tracking, and, of course, ingenious design–aspects of their business that we, as end users, rarely see.
So can we ever get a peek into the machine? Probably not, for all sorts of technical reasons–and since neither Facebook, nor Google, nor Amazon have any reason to share the details. Unwelcome Gaze, by artist Vladimir Mitrović, is a visualization of this deliberate, corporate obfuscation. Think of it as a series of portraits of Facebook, Amazon, and Google, built from network deconstruction, technical inference, and artistic liberty.
“The main goal of this project was to produce a ‘sensory’ visualization, remaining accurate but not aiming for ultimate precision–which is what traditional data visualization is all about,” says Mitrović. “I wanted to present these networks using different styles because these companies’ public perceptions differ considerably, and yet their public-facing infrastructures are very similar, since they’re all using the same underlying infrastructure of the internet (just like the rest of us).”
Mitrović started the project by running traceroute scans on Amazon, Facebook, and Google. A traceroute essentially follows information sent from your computer through the various layers of routers and other pipelines of the internet, back to the companies themselves. “For example, if you run a traceroute scan against google.com, you will first get the address of your own computer, then your home or office router, then your network provider’s first router, and so on. The last machine in that trace will be the actual machine serving google.com,” says Mitrović. He also wrote his own script, using the public information of each company, which “slowly but meticulously” runs what you might think of as a deeper traceroute, scanning to spot every possible machine connected to these networks–basically charting the corporate servers that likely power all those Facebook Likes, Google’s PageRank algorithm, and Amazon’s web services we all know so well.
“It took a very long time,” says Mitrović.
After all this data collection, Mitrović was left with a pile of traces, which as he points out, are really just multi-segmented lines, bouncing around from his home computer to servers, much like planes do from one’s own city to airports across the globe. He assembled these lines into charts with more software of his own making. Then he re-ran the entire visualization, cutting out bits of information, another three or four times on each graphic, stacking the information like an incomplete, skewed photocopy, to represent the abstraction of our real ability to see into these machines or make sense of their methods.
In Facebook’s first scan, for instance, the most literal portrait of corporate infrastructure is conveyed in dark blue lines. But on top of that, Mitrović layered pale blue lines–these are the destinations of your information only, “blurring, if you will, the routes leading to them,” says Mitrović. “The third and fourth [layers] are even more abstract, but I’ll leave it up to the viewers to interpret.”
Ultimately, the graphs that Mitrović created look like expressionist displays of unconstrained power, lightning bolts crossed with scribbles, a move-fast-break-things world, the soul of Silicon Valley excess exposed–as much as it can be peering from the outside, at least.
“The overwhelming sentiment while developing this project was my utter shock with the size of the internet address space of each of these companies,” says Mitrović. “In some cases I had to ‘prune’ the graph and present only subsets of available data, otherwise the end result would be just a huge unintelligible mess of lines. There was a fine balance between showing too little and overcomplicating, and I think in the end it came out well.”
If you’re a fan of Mitrović’s work, prints are available on his site. They start at $21 apiece.
Swiftype Custom Site Search