Episode Fifteen: I Miss Dopplr, InfiniFriends
(Also known as the My Friends Are Significantly Smarter Than Me episode)
1. I Miss Dopplr
Yesterday's episode included some scribblings about the Quantified Self, Felton's Reporter and Nike's Year In Nike Fuel. I was reading Fathom's post about the thinking behind the Year In Nike Fuel[1] and was again struck, and a bit saddened, by the tone and manner of presentation of what ultimately is intensely personal data about your daily movements. The post itself is the interesting thing: talking about gym rats, mountaineers, working dads, city slickers and how they all have distinct patterns. Or how the Fuel activity of a newborn baby looks.
But the output - so cold! Pure data! Red and blues, and, ultimately, a dashboard that looks like it's trying to emulate a Boeing Dreamliner glass cockpit. All of that emotionally loaded, meaningful, personal data gets reduced to "vigorous exercise" or "moderate exercise" or "inactive". The half marathons get turned into "highest weekday peak". Workouts get reduced to "3.0" per week, and weekly exercise gets turned into "4h 4m", which is "moderate".
Bullshit, I say.
I mean, look at this, from that Fathom post:
Those are different shapes. But preoccupation with all of this data and showing the actual numbers and, man, the obsession with the Quantified part of the Quantified Self and the fixation that that must mean numbers. You know what those charts look like? They look like they were designed to be read by computers. They look like things that computer vision algorithms could have a field day with. But the data was gathered by sensors in the first place! Those charts are ostensibly for humans and yet the aesthetic is almost as if they’re trying to look like a fictional-future in-HUD display for the Terminator. Those charts are what a science fiction movie would show when getting us to understand what a robot saw when it looked at a person.
I was instantly reminded of how much character there was in the Dopplr annual report. That dream team of Biddulph/Jones/Insam produced not only a product that was ahead of its time but had a phenomenally meaningful annual report. Their example for Barack Obama[2]:
is brimming with context that helps make the quantified part of data understandable and relatable on a human level. (As an aside, you would see more work like this in terms of design making data understandable through work like HowBigReally[3] (which is still amazing for helping me realise that either the moon is tiny or Australia is massive) and Schooloscope with its Chernoff faces.
We’re not wired to understand bar charts. We’re wired to understand other things. That’s not to say that there isn’t work that can be done to make bar charts and pie charts and graphs and sparklines and so on more understandable. But they’re not the only way of representing data, and holy shit, did you know we have computers now? That make things like saying “Your personal velocity for 2008 was 38.10 km/h, which is about the same as a six-lined race runner lizard” significantly easier! And Christ, they were doing this back in 2008. That was five years ago, merely a year after Flickr came out. It is honestly disappointing to me that we are collecting more data than ever, that means more to people than ever before, and still obfuscating it.
It makes me upset about what we’ve lost, and how far behind we are.
Try fucking harder, people. It’s not just a case of throwing Processing at something. It’s almost as if the tech industry has a reputation for being severely lacking in empathy with actual people, for fuck’s sake.
[1] http://fathom.info/latest/6796[2] http://www.flickr.com/photos/dopplr/3198685033/sizes/l/[3] http://howbigreally.com
2. Infinifriends
Tom Armitage, who is a stupendously smart person, has made a Thing which he’s finally made public and that he forgot he showed me over the Thanksgiving break, which for everyone else in the world, was around the end of November.
His Thing was Infinifriends[1], which is essentially the fucking genius realisation that a) all the Friends scripts have been transcribed onto the internet, b) you can get amusing uncanny-valley programmatic animation with algorithm induced emotion from things like Plotagon[2] or xtranormal (which doesn’t even exist anymore), which are essentially fancy text-to-speech engines with probably Unity-powered 3D animation on the back, and c) Markov chains, because, hey, you throw Markov chains at everything these days. Markov chains are the new Bayesian filters, didn’t you know?
Anyway. You end up with infinite episodes of Friends. And they are wonderful, just perfect, to my absurdist mind. They make no sense, and that’s why they’re funny. And they play off the Friends universe, so you have this super interesting juxtaposition of terrible, nonsensical writing and bad computer generated acting colliding in your head with the mannerisms of the real actors, on the real sets, backed by a writer’s room.
At some point, these are going to be spat out to YouTube and then people are going to freak out, because algorithmically generated drama is going to be ‘good enough’ in the same way that people thought people would never watch movies on a 3.5in diagonal phone screen and hey, it turns out that people will.
And because this is how the world works now, expect generalised infini-engines for pretty much any syndicated TV show.
OR, OR: and I’m looking at you, Mr. Ellis, now that Marvel has a fucking universe API, algorithmically generated Marvel Universe comics.
Ha. Where is your algorithm god now.
[1] http://infovore.org/archives/2014/02/10/infinifriends/[2] https://plotagon.com
Anyway. That’s it for today. I’m still angry about Dopplr.
Dan