s2e16: Battlesuits; Simulations and Puppyslugs; Scratchpad
0.0 Sitrep
Monday, August 10, 2015. I didn't write on Wednesday or Thursday last week. Wednesday because I was taking a Proper Holiday which meant not even writing a newsletter after having been away from home for a few days for Foocamp and work, Thursday because it go to the evening and I was too tired and Friday because Jesus Christ it turns out that words are really hard to write at the end of a long day and all you want to do is collapse. But now it's Monday and thanks to Nick Sweeney prodding me (thanks, Nick), I am guiltily writing to you at 8:53pm. First, a bit that I started on Friday...
1.0 Battlesuits
Vague potential spoiler warning for William Gibson's The Peripheral: the internet, corporations, legal, startups.
I'm reading - but haven't finished - William Gibson's The Peripheral[0]. I'm probably about... well, I was guessing about 3/4 of the way through and then I checked and technology means that we're so productive now I can know for sure that I *am* 3/4 of the way through, and there are a few things that have taken up space in my brain and are, I dunno, background-processing until my subconscious will spit out something interesting or relevant. Namely:
- lawyers (yes, but only once you have sufficient capital) are a battlesuit, a way of getting things done, a way of moving things around so that your environment can do what you need. I don't need a fulcrum, I need a law firm (or rather, an international association of limited liability partnerships that are willing to do my bidding) and, I suppose, control of the financial markets. But: lawyers are tools that reality engineers use.
- Cities are a battlesuit[1], a force-multiplier for lots of people. Millions of people. Miles and miles of extruded infrastructure of all different kinds, for moving things around, for moving information around, for extending your sphere of influence past just your own body. Systems for bringing things to you. All of this, increasingly, looking like it kind of *grows*, because our best attempts to *design* it so far, as opposed to let it sort of spontaneously self organize tend to have been just more instances that dictionary editors can use to illustrate the word "hubris" with. Cities, it turns out, are more like grown things that obey laws that we don't quite understand yet - not without trying[2] - and are the product of stupendously complicated interactions between already stupendously complicated (and, most of the time, largely-not-rational-albeit-still-rather-predictable) homo sapiens.
- startups are battlesuits? Startups are baby corporations, sometimes infused with a bit of acceleration capital, sometimes encouraged to go hockeystick and punch through some kind of phase change. Get the right kind of startup and the people who're in it (and the infrastructure that *they're* able to access) lets them do... big things? But I suppose it's not really the startup that's a battlesuit that a founder wears. It's more that legal envelope that a collection of people - the founders and early employees of a startup - shroud them in. The rest of it is just infrastructure that most other people can use nowadays. Startups have a bit of a force-multiplier, but not very much. They're like, I don't know, I light-weight battlesuit. Not like the big hulking things that cities are. Startups are, I suppose, in a way that cities aren't, like lightweight things that groups of people can slip in to. Something more comfortable, perhaps, while we get ready to disrupt?
- the internet *makes* battlesuits. The internet is a medium for battlesuits. The internet connects you to millions of people who have battlesuits or want you to have their battlesuits. The internet wants everyone to have a battlesuit. At least, the best bits of it. The battlesuit is a thing that an individual can shroud itself in, a collection of tools that surrounds you at all times that has something for everything, a kind of worn swiss-army knife that can't be shrugged off, something that's still part of something - embedded in your phone right now, your wrist later, who knows where else, after that. Who knows what Uber wants, but one of the possible futures is one where Uber just wins at logistics. And then what, anyone in the world can - well, some anyones, I suppose - move anything they want, from anywhere to anywhere else? For a reasonable cost and reasonably quickly? The only other people who're good at that are either international shipping (which isn't that accessible to regular people) or FedEx or, well, the US military. How eaten by software has FedEx been, anyway? It's not like they're all looking at drones, anyway. So here, here's the thing, here's the thought: the internet as a suit. It's already a wearable. You're in it. You're bathed in it.
[0] The Peripheral - Amazon
[1] The City Is A Battlesuit For Surviving The Future - Matt Jones
[2] Geoffrey West Finds the Physical Laws Embedded in Human Cities | DiscoverMagazine.com
2.0 Simulations and Puppyslugs
Boris Anthony has written the second part (and because he's a computer scientist, the second part is numbered 1, because all good computer scientists start counting from zero) of his Puppyslugs 'R Us series[0] and in it he finally explains what he's going on about by giving us a brief overview of what all this deep dreaming business was about. See, all this deep learning business revolves around two different kinds of neural net - and here I'm sure Greg Borenstein will write in reply and tell me exactly how they work, but just come along with me for now. There's recurrent neural nets and there's convolutional neural nets, but all you really need to know is that they're neural networks and they have hidden bits where they think more about the things we want them to think about. Instead of just one layer of thinking neurons, there's multiple layers of thinking neurons but we only really get to see what goes into the first layer and what comes out of the last layer. That's because all the ones in the middle are, well, hidden. And that's why we call it deep learning - because of the number of layers of neural nets.
ANYWAY. These models - especially the ones that are good at recognizing things in images - are modeled, more or less, after the way that our own vision system works in that there's a bit that recognizes colours, a bit that recognizes edges, and squares and so on. And because of what we've trained these particular Googly deep learning neural nets on, they've become very good at recognising a) puppies and b) slugs. Which is why we call the things that they see puppyslugs because they're trying, very hard, because we told them to, to see bits of puppies and bits of slugs.
At this point I'm going to skip over all the stuff in the middle and just tell you that: a) if you think we're just going to use these deep learning nets to recognize kittens, I've got a big targeted ad campaign to tell you, and b) of course we're probably going to start using them to simulate people at some point. There's a somewhat complicated paper/presentation from Geoffrey Hinton, who's incredibly smart and knows what he's talking about, about something called thought vectors[1] which I think is moderately readable, but if you want to get the main gist of it, it's this: you don't necessarily need to know what it is that you're *thinking* more that the if you get a description of what's inside your head, you can try and document or describe it as a vector. And it turns out that computers - at least, the ones we have so far - aren't that bad at doing things to vectors. And that if you can turn a sentence into a thought vector (which Hinton shows) what you're doing is you're taking a sentence in a language and turning into the underlying *thought* that the sentence expresses. And once you've done *that*, you're a hop-skip-and-a-jump (here I really *do* expect Borenstein to jump in) to predicting what thought vector follows any given thought vector. Because we can do that to sentences.
If all that was a bit hard going, here's a short story I wrote based on the Google Barges[2] (remember those!) and the day they opened which is a nice little introduction to, well... just read it.
[0] Puppyslugs ‘R Us: Part 1 - Boris Anthony
[1] hinton-aether.pdf - Google Drive
[2] The Day The Barges Opened — Medium
3.0 Scratchpad
A series of idle notes in bullet form:
- peak anthropocene thanks to the Rolling Stone declaring that we've reached the climate change point of no return[0]
- the industrial revolution never had any marketing (citation needed, I suppose) thanks to lunch with @tobybarnes[1]
- a whole thing that's still ringing in my head about designers with critical thinking skills *as well as* praxis skills, ie: ones who're thinking about the pixels that they're pushing and not just pushing the pixels that they're told to push, or ones who're not just concerned about making the pixels that look nice, but think in larger, systems ways. The point being, of course, that this is again the problematic and artificial divide between "doers/makers" and "thinkers" and that sometimes good things happen when you've got a combination of the two, instead of just hoping for a unicorn who can do both. Anyway. Here's me raising a glass for the people who think. Because thinking is important. And if you want to, as they say, "do the hard work to make things simple" that hard work by my definition at any rate necessarily includes thinking about things.
- I am seriously considering spending a couple days doing a value chain Wardley mapping exercise of the US Healthcare System because I am crazy.
- are we going to have to go through the whole "but are they art" thing about virtual reality in the same way that we've had to do with videogames or are we going to be able to skip the whole boring godawful mess?
- There's this whole bit in some post-scarcity economies in fiction where people do things because a) they like doing them, or b) they're just pitching in, but not because in any way c) they have to because if they don't they won't have any money and then they'll die. So in, I think, some of the Ken McLeod's Fall Revolution series of books, characters bus tables and serve meals because it's just something you do to help out. No-one's making you do it, and you don't starve to death if you don't. Similarly in the Culture series, people just do art because they feel like doing art. There's a slight argument, I suppose, that some people live in a comfortable enough post-scarcity and secure environment where they've got the ability to do things for fun and for free that other people still have to do for money and because they don't have any other choice. So whilst you have Uber drivers who are Uber drivers because it's the best way to make money for them (and they need money to survive, and Uber is their *only* source of income) you also have people for whom Uber is an additional source of income and then on top of that, potentially and just a reckon on my part, people who drive "for fun" and for whom Uber income is literally shoe money or fun money. Another aside here: Uber would be a lot more accurate in calling what it provides to people as "sources of income" rather than "jobs" because the colloquial and common understanding of "job" is something different, I think, than "something that pays me money". But again, that might just be my privilege showing. The point here though is that there's a bunch of people - first identified by Clay Shirky in his book Cognitive Surplus[2] who could just dick around on the internet and contribute to an encylopaedia open and free to all about everything from, well, important (sorry, "notable" subjects) to, er, very long-winded and well-researched articles about all the characters in Star Wars[3]. Anyway. Things that privileged people are now able to do, for fun, ish, include: writing books, making music, driving cars, renting out homes, and, I guess, if I wanted to do this properly, a much longer and better researched list about some sort of, I don't know, productivity surplus-slash-market-arbitrage where on the one hand you've got things like airbnb (sorry, I keep coming back to them as an example) for individuals to fully participate in a capitalist market economy and make sure that the utilization of resources is priced effectively, and on the other hand, you've got the ability for, er, recreational pastimes and the amateurisation of production (ha, Etsy, I guess) and access to markets with less friction than hitherto seen before start to, uh, I don't know, sublimate this and transcend that. Sorry. I didn't quite stick the landing with that one. But hey, main thought: amateurisation and democratisation of production by people who have the security and privilege to do so.
[0] The Point of No Return: Climate Change Nightmares Are Already Here | Rolling Stone
[1] Toby Barnes (@tobybarnes) | Twitter
[2] Cognitive Surplus: How Technology Makes Consumers into Collaborators - Clay Shirky, Amazon
[3] List of Star Wars characters - Wikipedia, the free encyclopedia
--
Phew. I don't think I really wrote about what I wanted to write about, but I'm tired, it's 9:48pm and I have to switch Twitter off because that tweet[0] is going bonkers still and is playing havoc with my notifications. So, about an hour, and about two thousand three hundred words, so a *very slow speed* of only about forty words a minute. Blarg.
[0] Dan Hon is typing on Twitter: "Google release notes"
Anyway. Send notes. Say hi. Say you're excited about me finally being able to publish and talk about my Code for America work that's attempting to help governments make better technology procurement decisions by mainly not procuring them, or procuring "solutions" in the first place.
Best,
Dan