Episode Thirty Two: The Machine; Route Around Failure; The Most Expensive Selfie

by danhon

1.0 The Machine

Ok, so I caught a little bit of flack for my freshman late-night philosophical rambling about the Internet being the most complicated machine, more complicated than the Space Shuttle. And yes, if you think about it, anything can more or less be a machine if you loosen your definition well enough. But I promised myself I would dig myself further into the hole I started.

Whether the internet is a machine in traditional-moving-parts sense is a bit of a red herring. Yes, it’s useful to consider it that way from the building-and-design-and-engineering side of making a machine, but I’m more looking at this from an end-user’s point of view. Look, here’s another Star Trek reference.

“Computer,” in Star Trek: The Next Generation is a singular thing. Sure, its processing and transport might be distributed, but it’s one singular system in terms of its (magic, conversational) interface with an end-user.

In the same way, I reckon, that the constituent parts of the “internet” with its all-pervasiveness and the fact that it powers so much stuff now, are becoming more machine-like – as in, a singular piece of machinery that accomplishes a task. I refuse to stoop to the level of school essay and looking up the definition of the word “machine” in wikipedia, so you’re not going to provoke me that way.

It’s interesting, sure, as an intellectual exercise (and also in understanding the internet as a complex system) to envisage all the moving parts involved in an http request[1] – an exercise I facetiously remarked as not being fair due to the profusion of SSDs and asking if electromagnetic waves or photons counted as “moving parts”.

But although we didn’t top-down design the internet as a closed-system machine like Shuttle, that includes lots of moving parts, we have instead designed something bottom up that has started to feel like it fits together in a certain way. And at that point, where do you stop, what constitutes the parts that make up the internet? Certainly the servers and the infrastructure involved in the storage and computation of data. But then also the transport of that data. And the computation involved in figuring *out* the transport of that data. And then do you count the environmental and power systems? And then what of the end-user systems like endless cellphones and laptops and desktops? Or Televisions? Do the mere end-clients, only serving as display services (witness all the IP-address bearing transport signage) count as machinery? Do they count more if they send data back (which they certainly do) that alters the state of the machine? And then: are all networks then machines? Or is it just the complexity of the internet that makes it a machine?

So anyway. That’s my contention. The internet is a fabulously complicated machine, the likes of which we have never made before, not least of which because no one has the plans for it. This has spun my mind off in a variety of wonderful directions, like: a) which museum is going to make a bona fide effort at collecting the Internet (I’m looking at you, Messrs. Chan and Straup Cope) and b) I would quite like to go to an exhibit that shows me what happens when an http request is made. Preferably in a building as big as the Tate Modern turbine hall. And probably sponsored by Google.

[1] https://twitter.com/flaneur/status/441795244611231746

2.0 Route Around Failure

This thinking about machines got me on to another point about the internet. It’s a positive characteristic of the internet that geeks get to say it routes around damage – that it was designed as a defense ARPA project for a resilient network post nuclear-fallout. Nuke out a node and the network would still work, routing around the physical damage.

It now turns out that part of the damage that we’re encountering on the internet is in its infrastructure itself. That it’s been pretty easy (ish – for the surveillance agency of the world’s superpower, but still) to suborn and listen in on.

We’re even fond, as geeks, of saying things like the internet routes around censorship, which, to be honest, I haven’t really ever had that much truck with, because that’s not a concept that the internet gets to own just like that. Humans, I prefer to say, route around censorship. Sure, the lack of centralized control means once something’s on the internet it’s always on the internet, more or less, and there are lots of ways to publish.

This talk of routing makes the internet *as designed* seem resilient to failure. But that’s routing around of failure modes that were contemplated for, by the current internet. I don’t know if it means that the internet can route around surveillance. Not when it feels like sometimes that surveillance is so deep. With knowledge that the three letter agencies are doing everything from sampling peer to peer webcam sessions through to physically splicing through fiber trunk (and sometimes even undersea trunk?) how do you route around that, when the physical layer is interrupted and repurposed without your knowing? Sure, we have things like TOR that work to anonymize traffic to a certain extent, but is it really a surprise when we find out that TOR nodes themselves have been compromised? In a way it almost feels like an achievement, when you think about it from this point of view, that the internet has been so successful when it’s been (kind of) so easy to repurpose.

Put another way: the internet we have now is what happened when DARPA asked academics and scientists to imagine an network resilient against the fear of the day: a nuclear attack. Is DARPA asking the same kind of people what kind of network they would design that would be resilient to surveillance and cyber warfare? (And how much I hate that latter term). We have made – and continue to build – a beautiful thing that is riddled with holes, and in the meantime we’ve learned so much that we can’t trust *humans* with security. So what type of system would we build from that knowledge?

3.0 The Most Expensive Selfie

Well, maybe. I’m referring of course to Ellen’s Oscars selife, the one that smashed Obama’s record of previously most-favorited and most-retweeted image. Now, you might not care about such things (in which case: well done! Have a pat on the back that you don’t have to care about such things) but in the world that I live in, an extraordinary number of people were prepared to have opinions and reckons about that selfie. And now here I am, too. No, I don’t think that anyone particularly noticed (other than, again, the people whose job it is to notice) that Ellen took the photo with a Samsung Galaxy Note. Allen Adamson, the Landor ad exec who claimed that you couldn’t buy this sort of magic virality[1] immediately fell victim to what at times feels like the journalistic prank of the unfortunately placed quote that makes you look dumb in an article that claimed you could, indeed, buy that kind of magic virality, and it turns out that it cost Samsung around twenty million dollars, all in all. “So what,” you say, isn’t this the kind of thing that Apple has been doing for ages? And isn’t this exactly what happened when Will Smith waxed about his Converse in that awesome movie about the dangers of Software Update, I, Robot? Well, yes. Which is why all this hullaballoo is bizarre and decidedly inside baseball. Instead, of course, we should be talking about the nous of David Cameron and his utter inability to realize how others see him[2] and the awesomeness that is Sir Patrick Stewart who reminds us that it’s always much more fun to dick about on the internet than to take oneself too seriously.

[1] http://online.wsj.com/news/articles/SB10001424052702304585004579417533278962674
[2] https://twitter.com/David_Cameron/status/441306815733579776
[3] https://twitter.com/SirPatStew/status/441340093643894784

That’s it. It’s Friday. Have a nice weekend, maybe go for a walk, give someone a hug. I’ll see you all on Monday.

As ever, I appreciate any and all of your notes and feedback.