Tuesday, November 5 2019, taxiing at Sacramento airport and about to take off back to Portland. Don’t even think of @ing me about my carbon footprint.
This is not Monday, which is when I thought I’d have this out, and I feel marginally bad about it. Monday did not work because I had a stupendously early flight and then a bunch of work to do and then was very tired. Instead I decided it would be entirely appropriate to watch a bit of For All Mankind, which is okay, especially by the time you get to episode 3.
Before I get on with the show, a note: I’m going to take next week off - there’s a bunch of family stuff I have to do, so I’m going to do the healthy thing and set the expectation that there’ll be no newsletter at all. If there ends up being one then that’ll be a nice surprise, as opposed to an obligation that I feel needs to be met.
OK, definitely on with the show now:
It’s a Snow Crash day, so today we’ll have some more of Snow Crash’s Chapter 13, and then I’m going to see if I can write in a condensed manner about a few Things.
Previously, on Snow Crashing, in numerical order!
Part 1: Chapters 1 and 2 - Episode Forty Four: Snow Crashing; danah boyd; Facebook and Oculus Rift
Part 3: Chapter 4 - Episode Forty Six - Snow Crashing 3; Video is a Content-Type; Blame Your Tools
Part 4: Chapter 5 - Episode Forty Nine: Living In An Immaterial World; Snow Crashing 4; Odds
Part 5: Chapter 5, cont. - Episode Fifty: Cities; Snow Crashing 5; More Television
Part 6: Chapter 6, cont. - Episode Sixty Four: Computer Says No, Snow Crashing
Part 7: Chapter 6 - Episode Ninety Two: Continued Disruption; Snow Crashing (7); Edge of Tomorrow
Part 9: Chapter 8 - Episode One Hundred and Forty Three: Email; Snow Crashing (10); 2014 (4)
Part 10: Chapter 8, cont. - Episode One Hundred and Eighty Seven: Snow Crashing (10); How The Web Works Now
Part 11: Chapter 9, cont. - s07e09: Snow Crashing (Part 11, Chapter 9)
Part 12: Chapter 10 - s07e11: Snow Crashing
Part 13: Chapter 11 - s07e13: No, not a mid-life crisis; Snow Crashing
Part 14: Chapter 12 - s07e15: Snow Crashing
Part 15 : Chapter 13, part 1: s07e17: Snow Crash’s Graveyard Daemons, and…
OK, so we’re at Hiro’s Metaverse workspace running locally off his “computer”, which he’s goggled into from a van moving slowly on some highway in Los Angeles and we’re about to be introduced to Earth.
I may as well just quote this bit:
There is something new [here]: A globe about the size of a grapefruit, a perfectly detailed rendition of Planet Earth, hanging in space at arm's length in front of his eyes. Hiro has heard about this but never seen it. It is a piece of CIC software called, simply, Earth. It is the user interface that CIC uses to keep track of every bit of spatial information that it owns—all the maps, weather data, architectural plans, and satellite surveillance stuff.
The easy bit here is to point out that “Earth” was the inspiration for Google Earth which is partly true but as always, not the entire story.
I found a blog post by someone who worked at Keyhole at the time and Earth has at least a few sources of inspiration. There’s the Star Trek tricorder (which I hadn’t heard of before) and there’s Snow Crash. But most persuasively to me is the combination of a hardware innovation from SGI (a clip-map texture unit) and a cultural artifact (the Powers of Ten film by Charles and Ray Eames). There’s also, in the background, the promise of the internet as a significantly more democratic platform to access information, namely high-resolution satellite data. The reasoning goes, Earth would be something that would be a brand new way for people to access something (a bird’s eye view of the entire globe) they would never in their lives have a chance of accessing otherwise.
This is borne out by Stephenson’s commentary:
Hiro has been thinking that in a few years, if he does really well in the intel biz, maybe he will make enough money to subscribe to Earth and get this thing in his office. Now it is suddenly here, free of charge. The only explanation he can come up with is that Juanita must have given it to him.
There’s so much here! First this idea that the information economy will actually be an information economy, where discrete packets of information will be bought and sold in some sort of market (which is how CIC’s Library is presented). This has clearly turned out not to be the case, and in my experience, it’s not that the raw information is valuable (e.g. Hiro and Y.T. learning about the Rat Thing), but in 2019 it’s the actionable insight and translation that’s useful. Fine, you might have some secret, proprietary/confidential information about Rat Thing, the kind of stuff that Uber might get taken to court by Google over. But that’s not necessarily useful on its own.
The second thing is that Hiro would make enough money from this information economy to be able to subscribe to Earth in the first place.
When we look at what actually happened, Keyhole was a paid-for piece of software that you had to buy and subscribe for (getting yearly updates) until Google acquired it in 2004 and gave it away for free. Even now, I don’t think it’s not entirely clear why Google gave the client away for free at the time other than a genuine desire to share a cool thing with people. With hindsight though, the data behind Earth, the satellite imagery, ended up being used in the nascent Google Maps service that debuted in 2005, itself an acquisition and relaunch after being worked into an web 2.0 AJAX-y web application - Google’s release of Keyhole happened in June 2005; Maps was released in February of that year.
So Stephenson got the subscription part right, but what he didn’t anticipate and what we don’t see in Snow Crash is the gradual encroachment of the ad-funded internet and the ad-funded “information economy”, whatever that turned out to be. For regular people, we pay for Google Earth through some sort of privatized life-logging service where Google gets to store and understand more of our location-based questions and needs.
There are other things that regular people do subscribe for - it’s not clear if the kind of work Hiro does would benefit from the sort of data financial analysts use (e.g. near realtime, high-resolution imagery of car parks to estimate retail traffic, crop productivity and so on), or if he’s the kind of person whose job can subsist on purely open source information (which, you know, is good enough to make pretty good conclusions about nuclear proliferation).
Anyway. Turns out that the BABEL / INFOCALYPSE hypercard, beyond containing Earth and Library information and a Librarian (which I could imagine as all running as sandboxed code on Hiro’s computer), also has the ability to modify Hiro’s virtual environment, or that Juanita’s explicitly been granted permissions on Hiro’s computer. At the very least, it feels like he’s running untrusted code that she’s just emailed or DCC’d to him because there’s now an office behind a sliding door - “apparently Juanita came in and made a major addition to his house as well”.
The Librarian walks in from the office and we get Standard Male Bookish Librarian (as opposed to Cat Lady Librarian, Sexy Glasses Librarian or the other librarian stereotypes). My point of reference here is Giles from Buffy.
There’s a few points that we get in our introduction to the Librarian that catch my attention:
Even though he’s just a piece of software, he has reason to be cheerful; he can move through the nearly infinite stacks of information in the Library... ... The Librarian is the only piece of CIC software that costs even more than Earth; the only thing he can’t do is think.
First, I can’t believe I missed out on Stephenson catching on to our trend of naming software/apps/services with the most generic noun possible: Earth! Microsoft Windows came out in 1985 and I can’t remember when the whole trademarking issues about just owning the word Windows happened. But having dumb generic nouns for things is certainly a thing now.
The second is that, other than the part about not being able to think, the Librarian functions as a search engine to the Library’s stacks. The idea again that you’d have to be able to pay for a search engine function, potentially separate from having raw access to the Library itself, is also super weird in 2019. The Library still feels more like a proprietary, private database like Lexis/Nexis (it is, after all, modeled on a merger of the Library of Congress and the CIA) with an add-on for a better, more natural, search interface.
In the next para, there’s throwaway references about the embodiment and presentation of the Librarian: he (not it!) is “eager without being obnoxiously chipper”; and also “clasps his hands behind his back, rocks forward slightly on the balls of his feet”. Given what we learned about avatar programming when we were in The Black Sun with Hiro (the parts about Sushi K’s hair, for example, and needing to be able to puppet an avatar to be able to sword fight in the Metaverse), I have a question about the source of this Librarian behavior. Who programmed it? Is it emergent? Why does it matter? Was there a team of CIC character animators and riggers working on NPC-like Librarian behaviors? It is not as if, say, Hiro starts having a conversation with the Librarian and then they get side-tracked into that one time the Librarian took an arrow to his knee.
The next part is our Librarian expository info-dump which is a classic Stephenson move, amusing in this instance because it’s literally an information dump from a) a large source of information; b) something that manages that information; and c) gets referenced later on a couple pages later.
Hiro starts off with a simple question “OK Google, Babel’s a city in Babylon, right?” to which the Librarian starts to respond: “It was a legendary city” before launching into what feels like a snippet/onebox answer, defining Babel as a biblical, breaking up its etymology, and then shifting into sounding a bit like a GPT2 auto-generated answer, with phrases like “it is probably also somewhat onomatopoeic”.
Eh, you know what, that gives me an idea: I’m going to give GPT2 Hiro’s prompt (“Babel’s a city in Babylon, right?”) to Talk to Transformer and paste in what we get:
Babel’s a city in Babylon, right?
The name comes from the Babylonian word babašu, which is a verb that means 'to write'.
But what if that was the first time you wrote? A good start:
> babašu > babaš
Babaš is a very simple compound verb, and the meaning isn't particularly interesting. It's an adjective that means 'writing' or 'writings'. It's a perfect match for Babel.
You know the expression, "Writing is the death of memory"? That's exactly how Babel's memory works. The characters are very different from how we expect them to be. But the words themselves are the same.
The name "Babel" can be found on various coins and monuments in Babylon, which makes it even more fitting. But even more interesting are the legends in which the city is mentioned. Many of the places we're familiar with in the Old Testament are named after the ancient city.
The most famous is Nebuchadnezzar's palace in Babylon, which was built...
Which, you know, is quite a bit like the Librarian! Good going, GPT2!
Anyway, this is the part where we learn a bunch about Babel that is important to the plot and also because Stephenson is interested in language and speech and so on. We learn, much like we might in a mid-90s educational CD-ROM, that Babel might have just been a tower “with heavenly diagrams carved into its top”, instead of one that actually reached into the heavens. And we start thinking about the story of Babel in terms of an “informational disaster” because the people “couldn’t talk to each other”.
Hiro using the word “disaster” turns out to set the Librarian off on a non-sequitur (which it’s a sucker for due to its internal structure), and that to me feels like Stephenson is also making a reference to the hypercard-like interconnectedness of all things. The Librarian cannot resist jumping from topic to topic and following those associations. It cannot not click on that little blue underlined link.
We then get a discussion about the Librarian’s provenance (“You’re a pretty decent piece of ware. Who wrote you, anyway?”) in which I slightly smirk at the word “ware”. The Librarian has the innate ability to learn and modifies its code, and that ability came from its creator. Hiro — or maybe Stephenson’s? — biases are on display here because Hiro thinks maybe he knew the person who coded the Librarian (“Maybe I know him. I know a lot of hackers”) and it turns out that the Librarian was coded by a researcher at the Library of Congress who taught himself how to code.
This is... amazing? Hiro lives in an information economy and it’s also one where a sufficiently motivated library researcher can teach themselves to code and create a piece of software that is ludicrously expensive and a natural language interface to the world’s biggest database. This person, Dr. Emmanuel Lagos, is some sort of uber-hacker, and we’re still talking about hackers, not coders or developers: but someone hacking together something like the Librarian.
I’m going to stop here - that’s about 2,000 words again(!) - and we’ll still have a few pages to go before we finish Chapter 13.
Look, I don’t know how to say this more succinctly, but everything is political you goddamn fools. This set of Things that I have Noticed is loosely organized around Nick Thompson’s Wired piece on Andrew Yang. Yang, for the non-Americans, is a Democratic Presidential candidate in some weird mirror universe.
It is never, as people have already pointed out, a good sign if you feel the need to write a piece saying that a person who people think is full of shit is not, actually, full of shit. In fact, the existence of such a piece is normally a pretty good indicator that the subject of your piece is indeed full of shit. But that’s not what set me off, oh no. What set me off was the fact that it’s 2019 and Thompson — who is the Editor in Chief of Wired — is saying stuff like:
Silicon Valley does not have a political ideology right now. There’s a long-standing strain of libertarianism and optimism, which is gradually being canceled by more powerful strains of progressivism and pessimism. But it’s still a muddled ideological melting pot, and perhaps it’s a place where Andrew Yang can do well.
I know someone is going to throw the oft-cited “you can’t get someone to understand something upon which their livelihood depends”, but COME ON:
Saying Silicon Valley doesn’t have a political ideology — which in America is a bit like saying Silicon Valley hasn’t nailed its flag to either the blue party or the red party — is so demonstrably false. Thompson acknowledges that the Valley has a “strain of libertarianism and optimism”, as if Wired doesn’t have a history of supporting Barlow’s manifesto and declaration of independence for Cyberspace. I mean, is a manifesto not a political ideology? The Valley - that amorphous, stereotypical collection of companies largely funded by venture capital and now capable of buying entire supply chains out and behaving as if they’d like to remake the financial markets in their own image - demonstrably does have a political ideology and it’s clearly one that’s quite tied to “making lots of money” and things like believing in meritocracies and technocracies. A political ideology is not whether you say you support the blue party or the red party, it is something that can be discerned, at the very least, from both the daily actions and the actions in aggregate of any particular actor.
I say again: it is 2019. Why is it so hard for someone like Thompson to acknowledge the inherent political beliefs and actions in the Valley? I’m now much more interested in trying to figure out why Thompson and Wired, a part of the establishment now more than ever, are unable to articulate this. Aside from, of course, the reason that it would not be politic to.
Look, here’s an example of a political ideology: YouTube apparently banned Sebastian Gorka from their private platform because he “won’t stop playing Imagine Dragons songs”:
Will Sommer @willsommerYouTube banned Sebastian Gorka because he won't stop playing Imagine Dragons on his channel. https://t.co/7Jiuvd0Jda
and yes I am being a bit flippant here but really, it’s quite simple and no I disagree if you say it’s more complicated than that. (Yes, it is more complicated than that and it is simple. It can be both at the same time).
YouTube through its actions is nailing its flag to the mast of intellectual property rights and ownership.
You want more? There was the whole thing the other day about “secret scores” and the surfacing of the mass of information held about users by companies that do what we laughingly called in the late 90s/early 00s “e-business” and would now call “oh my god we can’t escape this hell that we created”.
These secret scores - the 400 pages of all the delivery food you’ve ever ordered - they’re what happens when there’s the idea that we, as technologists, can just save everything instead of throwing away data, or having to choose what data to store because capitalism has helpfully made storage media super cheap. But hey, guess what! Turns out there’s a massive externalized cost that society has to pick up. And that’s never mind thinking about the climate change environmental impact of keeping all that data around.
I want to give technologists the benefit of the doubt. It is not as if we were holding pinkies up to our mouths in the early 2000s and days of web 2.0, sitting at O’Reilly ETech conferences thinking: Oh, what hell and havoc may we wreak by choosing to store everything. No. We just didn’t think it through. The question now, though, is: will we do anything different? Or will we make the same mistake?
It is a Tuesday, so I am minded to think that we will make the same mistake — ask me again tomorrow — which I’m going to attempt to segue into this last piece:
Beth Linas, PhD, MHS 👩🔬👩💻 @bethlinas👏👏👏” If we don't fix the biases we have in health care before we jump to training a computer to do it, we're just going to actually perpetuate those biases," #yup #AIForHealth #DigitalHealth https://t.co/WqUWi9WeBt
The bit that I caught on to here is the automation problem in the microcosm of healthcare. I love what Beth said there: if we don’t fix the biases we already have in healthcare before we jump to training a computer to do it, we will perpetuate those biases.
The thing that automation makes easy is to do the same thing at scale and so it makes it easier to do the worst thing, the most unsympathetic thing, the most callous thing, the most unthinking thing at scale and do it a billion times per second to hundreds of millions of people.
Scale is horrendous. Automation is horrendous. Automation lets us as a species be lazy at scale and apparently, if we are to believe evolutionary psychology and what we know about how our brain works, there’s no value on truth in the world, there’s only surviving and you can totally survive to pass on your offspring by cheating.
Downthread on that tweet above I wrote that Orwell imagined a future of a boot stamping on a human face but that is not the Orwellian future we face. It’s a billion boots, a boot in every pocket, a boot on every desk, a boot listening to what you say in your bedroom, a boot on every front door that’s being used to produce cutesy content marketing about halloween kids trick-or-treating, it’s that boot acting at scale, stamping on all of the human faces and making stamping decisions on a distributed network.
That’s what automation gets us if we aren’t intentional about it, and I worry more than ever that if there’s ever an opportunity for us to cheat and take the easy route, we will take it and it is the rare exception when we put aside our baser instincts and try really, really, really hard to do the hard thing.
A cat wandered on to the field of an [American] football game and it was adorable, so what’s the eSports equivalent? Or do we lose any capability for that kind of serendipity in that kind of venue? I’m less interested in for example a bunch of trolls/griefers storming the field and more like a completely unrelated process (that might have its own agency?) suddenly turning up.
I saw this tweet from the FWD50 conference about a deputy minister for transport saying “We need to focus on supporting experimentation without compromising on safety… we put provisions in place to ensure it’s safe, it’s a sandbox” and my brain turns it in to: “software-defined geofenced government innovation zones enforced by usage of the smartphone channel of the emergency alert system: YOU HAVE ENTERED A JUMP SCOOTER INNOVATION ZONE”, which is not to say that I think this is a good thing, more that: huh, if this were a thing, what would be horrible about it and what might be marginally good about it, and how do we think about “innovation” and “sandboxes”?
I made a dumb joke about Luke being exhorted to turn off Jira and estimate his software’s ship date manually and now you all had to read it. Turns out Star Wars is an example of how any sufficiently complex fictional universe is capable of supporting software development, procurement or project management parody.
Anyway, that was over three and a half thousand words and my flight is nowhere near landed yet, so sucks to be you, the person who just read lots of words.
As ever, send me notes: I love to read them and they’re a nice reminder that we’re all human beings.