s07e17: Snow Crash’s Graveyard Daemons and

This is a long chapter to recap!

0.0 Sitrep

Lunchtime in Washington, D.C. on Tuesday October 29, 2019. I’d intended to send this yesterday, but work, life and jet lag got the better of me so it’s coming out today instead.

I’m here to work with New America, Code for America’s partner for next year’s Summit and then set up our first committee meeting, reviewing proposals. Walking around the city I was pretty much overwhelmed by East Coast Capital City Sensawunda: compared to Portland, so many languages before 9am, so many people who, bluntly, aren’t white and significantly less segregated. I mean, it’s still a bunch of white dudes in suits zipping down the bike lanes on dockless scooters, but it’s just so refreshingly different. (And, I acknowledge, only refreshingly different inside a very restricted western ecosystem. It’s not like I’m raving about how different Nairobi is, for example).

I stopped myself from saying that I missed living in such a city, though. I like and miss being someplace like this, so different from Portland. But at this point I’m reasonably sure that living is just hard and I know how to live in Portland.

1.0 Some things that caught my attention, and…

Sorry, I have two long things today and frankly I feel like I could’ve done a lot more typing, but I suppose I should learn how to stop. I can’t write all the things that caught my attention.

1.1 Snow Crashing - Chapter 13

Last time, in chapter 12, Hiro and Y.T. learned about the Rat Thing at Mr. Lee’s Greater Hong Kong, and we saw our first glimpse of Y.T.’s home life and her mother.

Chapter 13 starts with us back in The Black Sun. Hiro’s killed the Nipponese businessman and our attention is brought to the fact that he’s an empty polygonal shell: “no flesh, blood, or organs are visible through the new crossections that Hiro’s sword made”, as if the other avatar was “a complex inflatable doll”.

You can kind of see how it makes sense to not model the inside of an object unless there’s a reason for it. The Black Sun is mainly a place for socializing, so why bother expending designer and programmer time on modeling interiors that aren’t going to be seen?

I also just noticed that in this book, we talk about programmers, not developers. The concept of programming is central to the plot and the concepts Stephenson explores, so even if Snow Crash were written now, I don’t think it’d make sense to use describe people like Hiro as developers.

Stephenson also uses this to remind us that avatars are a metaphor. There is nobody, nothing there; the ghost has left the shell. “It reminds all The Black Sun’s patrons that they are living in a fantasy world. People hate to be reminded of this.”

Well! We did learn back in chapter 7 that part of the reason why The Black Sun has been so successful is because Juanita thought faces were important. (That part was also followed by Stephenson pointing out that Juanita was the only one who thought faces were important, a minority in an “all-male society of bitheads that made up the power structure of Black Sun Systems”, her views dismissed as trivial and superficial and this being a symptom of the “virulent type [of sexism] espoused by male techies who sincerely believed that they are too smart to be sexists” and I am not quite saying that Stephenson is sexist himself here. What I will say is that over the course of his career and the books of his I have read, his books haven’t felt completely *un*sexist, and more that he appears, in my reading, to be quite comfortable with what I see to be his own non-male character tropes.

Back to the issue of people not liking being reminded they’re in a fantasy world. I get where Stephenson is coming from this; he wants us to understand that The Metaverse is an escape from the hellish, polluted, corporatized physical world. I don’t think though, that we ever see anywhere else in the book evidence that there are people who have completely retreated from his capital-R Reality, save for one person in particular. But Stephenson does say that the patrons don’t like to be reminded that they are *living* in a fantasy world.

I suppose there are people now who also feel like they are living in a fantasy world, but I don’t feel like the problem is necessarily that serious? Yet? There’s the issue of being immersed in some media or activity and being pulled out of it. That irritation has applied to such things way before the advent of virtual reality. At least in Stephenson’s telling of the Metaverse though and this strange future world where one can apparently make a living by selling “intel”, I guess it might be possible to spend quite a lot of time in the Metaverse. But again, this is just a simple thing, like... somebody being beaten in a swordfight, an occurence that is either common or notable enough for there to be a leaderboard. (Although that leaderboard may exist only because of Hiro).

The dead not-body is dealt with by Graveyard Daemons. The daemons were a direct result of Hiro having invented sword-fighting algorithms, and it’s implied that Hiro’s sword-fighting algorithms were also directly responsible for the first ever avatar “death”. Turns out the original Metaverse specification described avatars as indivisible and atomic - “the creators of the Metaverse had not been morbid enough to foresee a demand for “[cutting someone up and [killing] them”.

So Hiro’s sword-fighting algorithms invited a kludge to deal with the results of a successful sword-fight, described as a kludge (I suppose Stephenson has enough taste to not have described them as a hack), otherwise the Metaverse would “over time, become littered with inert, dismembered avatars that never decayed.”

Now, look. It has been a very, very, very long time since I have played something like tinymush or tinymoo, old-time multi-user text-based dungeons. The latter, tinymoo, was something very much like the Metaverse in that users could program it themselves, create rooms, and create programs that defined objects and actions and so on. So, this is a bit like that, but I’d find it hard to believe that there wasn’t an existing garbage collection mechanism that couldn’t be extended to deal with discarded Metaverse objects. But that doesn’t serve the story or help us learn about Hiro, so here we go: Graveyard Daemons.

This is Hiro’s hack for dealing with losing a sword fight:

First: the loser gets disconnected from the Metaverse, which again, is not like dying, and is only annoying, but is “the closest simulation of death that the Metaverse can offer”. Nowadays you’d have something like permadeath where the real-world user would lose all the attributes accrued to the avatar, so if it were possible to have possessions in the Metaverse (e.g. the space that Hiro has, and the hypercard etc), you’d potentially lose those, too. This is pretty much just like being /kicked from IRC. I don’t have many newer metaphors or explanations, I’m afraid.

Second: there’s a cooldown. You have a few minutes before you can log back in. This is not for a game design or moderation reason, which is why these things exist now. It is because the dismembered avatar “cannot exist in two places at once” which is dumb, and just a Metaverse rule.

Third: the avatar is disposed of by the aforementioned Graveyard Daemons, a new feature invented by Hiro. They appears as small black not-ninjas and quietly and efficiently emerge from invisible trapdoors, “climb up out of the netherworld” and “within seconds, have stashed the body parts into black bags [to] vanish into hidden tunnels beneath The Black Sun’s floor.” This is a tunnel system that only the Graveyard Daemons and Hiro can use, and I’m sure that ability is going to come in useful later. I mean, it should.

This is a lot of hacking! One would think that if the Metaverse defines an avatar as an indivisible atom that things like “walls” or “floors” would similarly be so. But remember, Hiro can mess the Metaverse’s rules (“code”), in The Black Sun, at least, because avatars can be sliced in half. So everything else can be sliced and divided, too.

Fourth: the Graveyard Daemons take the dismembered avatar to “an eternal, underground bonfire beneath the center of The Black Sun, and burn it”. When it’s all burned, the avatar vanishes, the two-places-at-once rule can again be satisfied and the owner can sign in again, “creating a new avatar to run around in” and hopefully more cautious and polite the next time around.

There’s a lot going on here in that last part. I suppose the eternal underground bonfire will be a nice reminder to the people who’ve read Fall; or, Dodge in Hell recently. And, really? A bonfire? Fires can be pretty expensive to render! All those particle effects and everything! But... that fire would only be visible if there were something to look at it, no? Unless... the entirety of the Metaverse is being simulated in realtime (again for the Fall;... fans), in which case no wonder your world is helplessly polluted, you idiots are computing a whole of bunch of inefficient crap.

This seems weird and funny to me because on the one hand we have the admonition that the death of the avatar in The Black Sun is a metaphor, but on the other hand, apparently those death metaphors are literally taken to a literal Metaverse place, are subject to an actual burning simulation begging the question of the verisimilitude of that burning: what, exactly, is burning? Are the zero-thickness polygonal avatars burning like paper? Like aircraft grade friction-stir-welded titanium? Whatever is burning, it definitely takes a few minutes, because the user can’t log in until the burning has finished. And the only person who can access the Pyre under The Black Sun is Hiro, of whom we have no reason to believe has a habit of jumping down into the tunnels to gaze, pensively, at the burning not-dead avatars.

Hiro may be a good programmer, but he’s certainly not a smart one, in my opinion. (For me, this brings back memories of the bank teller in sneakers: “You break into places and steal things to show that you can break into places... and people pay you for this?” “Yes, it’s a living” “Not a very good one...”)

Then, the user can “create a new avatar” which also implies, with the burning, that the avatar was literally destroyed. The account wasn’t deleted, but that... character was deleted? So we still don’t know if the character’s possessions were also destroyed and thrown into garbage collection. So if someone had spent a lot of time on their avatar, if someone like Sushi K had his amazing hairdo associated with that avatar... it’s gone forever? I have to create a new Nintendo Mii? That... sucks.

And then lastly, that after having their avatar destroyed and being kicked from The Metaverse and to have to wait for a cooldown, “he will be more cautious and polite the next time around”, a) Stephenson betrays his assumption that Metaverse users are male (too smart to be sexist, again?) and b) there are not enough laughing crying emojis for me to respond to this.

Hiro’s audience fades out and he loses his view of The Black Sun. Outside, without his goggles, he’s in the U-Stor-It, “holding a naked katana”. It’s much later, and there’s a bunch of people watching him out here too. And for some reason, Hiro is still holding a katana. Did he have to hold the katana when he was sword fighting in The Black Sun? This might have something to do with it being hard to control. If so, did he stop, take his goggles off, go and get it and then get back in? It doesn’t look like that happened.

Anyway, Hiro can’t see The Black Sun anymore because it was being lasered into his eyes via his goggles from his computer, and Vitaly’s ass is in the way. It is time for them to go, but not before Vitaly checks to see if Hiro won his sword fight (“Of course I won the fucking sword fight, I’m the greatest sword fighter in the world.” “And you wrote the software.”) Hm. I wonder if Hiro will turn out to be a good sword fighter in Reality.

Vitaly and his band (Vitaly Chernobyl and the Meltdowns, very in now in 2019 thanks to the HBO show) came over to Long Beach via a hijacked ex-Soviet refugee freighter and I wonder what sort of wall does not exist in this world. There doesn’t seem to be that much commentary on keeping refugees out (or that much appetite or time or effort expending in doing so) in this America. This America appears to be an uneasy truce sort of melting pot. They’re very into reinforced concrete to practice their art, and the L.A. River turns out to be a great place for that.

The crew get there using a really old VW Vanagon and I’m reminded that I don’t think there’s much discussion about the price of gas in this America, either. In any event, this is the kind of van that has been lived in, and there’s a very Stephenson narrative phrase I like here: “the ownership of the Vanagon is subject to dispute, because Vitaly owes Hiro more money than it is technically worth.” Anyway, the Vanagon is loaded up and we get a description of the U-Stor-It location being somewhere that’s a cross behind a refugee camp and low-income, culturally mixed housing. There’s kids, Mayan encampments, Buddhist shrines and white trash and a long list of drugs and their paraphernalia. And also, what looks like the delivery mechanism for a new drug, “little tubes, about thumb sized, transparent plastic with a red cap on one end”, interesting because the caps are still on them, notable to Hiro as “something he hasn’t heard of before, the McDonald’s styrofoam burger box of drug containers” and I can’t remember if we McDonald’s stopped using styrofoam packaging in our world because of CFCs and the ozone layer or because cardboard was cheaper.

The van gets loaded. Nothing much interesting happens, apart from an observation about the loading carts being “technically community property, but no one believes that”, and then they’re on their way to the venue. I mean, the L.A. River. And it’s LA traffic, so Hiro’s got some time to kill and it’s time to use his computer again.

Now, because Hiro’s in a car, he can’t connect to the network (not the internet!) by fiber (fiberoptic cable, no shorthand here), so “all his communication with the outside world has to take place via radio waves which are much slower and less reliable.”

There’s a lot here, too! Last time I expressed skepticism that fiber would be run to the U-Stor-It (although I suppose it could be pirated fiber that Hiro hacked access to in an act of altruism for his local neighborhood) given Comcast and co’s current strategy of internet not-provisioning. Second there’s the issue of exactly how much gigabit fiber would cost (again, a point in favor of Hiro pirating it), although I suppose if there’s anything Hiro’s going to pay for, it’s going to be internet access. And then finally: radio waves! Cell phones weren’t really a thing in the 1990s but pagers were (no, I do not have to explain pagers to you, you have seen Captain Marvel and that end-credits scene from that Avengers movie), and yes, they were pretty low bit-rate. But while we all don’t have fiber, there’s a fair amount of LTE coverage and 5G coverage, whatever that is, is rolling out across the U.S. You can get 300 mbit+ connectivity over LTE right now, so let’s just assume that Hiro isn’t pirating his Verizon Unlimited account and it’s his largest monthly expenditure. Or, the only other thing I can think of, is that Da5id is still looking out for his socially incompetent star programmer by accidentally not turning off his company internet benefits.

It doesn’t make sense for Hiro to go into The Black Sun because his ping and bandwidth are too low (apparently even if The Metaverse were rendered server-side and streamed to his computer a la Google Stadia, GeForce Now or... Netflix), but he can go into his office(!) because that’s generated client-side.

Hiro having an office is hilarious because Stephenson has gone all-in on throwing out command-lines and flat, 2D desktop metaphors. Hiro is all in with his Oculus or Vive or whatever, and can’t even open an Excel spreadsheet without goggling in to an immersive 3D environment.

AND! This first mention of radio waves is a prompt that social media or text communications pretty much doesn’t exist *at all* in this world. There’s no Twitter. No IRC, no Facebook, not even any text messaging. Hiro and Y.T. spend the rest of each other, I swear to god, making voice calls with each other. VOICE CALLS. Y.T. doesn’t have any problem *talking* to people on her phone. The Metaverse has done that thing that technology doesn’t really do, which is to completely supplant other methods of communication. I mean, Hiro doesn’t even really appear to get emails? There’s no Slack? There’s no DM to Y.T. (SORRY. I AM SORRY.) In the world of Snow Crash, email is finally dead outside of, I think, the Government.

Anyway, Hiro’s office is just like those space in The Matrix, which is to say: quite Nipponese. There is “silvery cloud-light filtering through ricepaper walls” so there’s some really serious ray-tracing going on here.

And here I’m going to stop, partway through Chapter 13, because we’re about to get introduced to Earth and the Librarian, and that’s a whole thing. And I’ve already written over 2,700 words.

1.2 Automated ethics testing?

[Reworked and expanded from this earlier thread]

There is a thing my brain does where I take a word like, say, ethics and then puts that word next to other unrelated words (and, to be honest, words from a completely different context) because my brain thinks it would be funny.

Digression: Someone once pointed out to me the uncomfortable and valuable feedback that I might be trying to hide what might be genuine insight (whether smart or not), or at least a prompt into thinking, behind an attempt at humor (“I see you’re trying to minimize yourself again, and I think you’re on to something again”). It’s strange - I clearly like to think I’m smart, but not in a way where other people might think I’m bragging, and certainly not in an area where I think I might be unqualified compared to professionals in the field. This is, again, all related to feeling whether I belong or not.

So, I did this:

  • ethics playbooks

  • automated testing for ethics

  • continuous ethics integration

  • ethicsops

  • ethics smoke testing

  • ethics tiger teams

  • ethics red team

  • o’reilly ethicscon

For the avoidance of doubt, I think ethics playbooks are pretty dumb in the way that many playbooks in tech are dumb: if you’re the kind of organization producing a playbook, they make it easy to think like you’re doing something (you are: you’re making a playbook), but making the playbook only really solves a small set of problems (winning hearts and minds, explaining what you’re going to do to make the work easier to do with people) and doesn’t exactly solve whatever underlying problem you have. They are statements of intent, not a description of action taken.

All of which is to say that the bullet points are provocations. Or, in an uncomfortable-for-me sense, trolling. I can think of a dozen reasons why “automated testing for ethics” would be a bad idea, but then I guess this is the point of satire which is to say something exaggerated in the service of criticism or examination. Which is, honestly, why I did it! It would just be less provocative to have to say at the beginning of all of the tweets like that “I am saying these as a provocation because when one thinks about them, one encounters issues such as x, y, and z and if you knew me, you’d already know my position that outcome x would be a dumb outcome and one you’d want to avoid from doing this”.

Anyway, the tweet spawned an interesting conversation with @ashedryden. Ashey pointed out:

… which I agree with as a correct and reasonable worry. When something like “ethics” appears to be dealt with because “automated ethics testing” is happening, then it’s all too easy for a checkbox to be treated as completed, and that the presence of the testing suite would imply that the job was done.

Any organization that thinks a technique or infrastructure like “automated ethics testing” means their obligation is disclosed is clearly not… actually giving a shit about ethics.

What I’m more interested in is some sort of idea that some coverage is better than no coverage at all with due consideration that some coverage is absolutely not all coverage. IN other words, thinking about what automated testing is good at, for example, making sure that you do not regress behind some previously established line or stance.

This kind of situation makes me feel weird about my history in software and also studying law. There are a whole bunch of people who think that law is like software, or that software is something that could be great at solving legal problems and, for another time, I would want to point out why all those people are horribly wrong. These are the kind of people who like things like Smart Contracts and think they’ll do away with lawyers. I agree that there are a bunch of lawyers who can probably be done away with! But not all of them, and not because of Smart Contracts :)

Where software and law are similar though, and when I say law I mean common-law based systems like England and the U.S., is that they’re built on edge-cases. These systems can be simplistically seen as hundreds of years of evolved code that are continually re-interpreted by a judicial system acting as a runtime, along with a language that itself is continually being iterated on for traits such as readability (e.g. the English replacement of the term “plaintiff” with “complainant”, I think).

I get confused, because it feels like software people forget that pull requests are constantly happening in law (I mean yes, put your primary legislation in a version control system, sure, that makes a lot of sense). Law is an evolving system, so there’s a direct analogy to continuous (albeit feeling slow!) development and integration. Courts are where the integration happens! Courts are where testing happens!

So in this way, any “automated ethical test suite” and coverage should be informed by how this works in society. Not to say it has to exactly work the way it works in society, but at least be informed by it, and you could also think about how the benefits of automation could be applied. But in any event, you’d want to frequently and regularly revise it (which might be easier when you’re using software infrastructure!), and you should probably explicitly accept that it’s going to be around forever and get more and more complicated.

I get that this is complicated. There are valid questions about how this might work in practice. But again, my position is that there are benefits in automation that might be able to be applied in this area. I think National Security Letter canaries are a primitive application of automation in an ethical area. And yes, there’s a whole issue on even being able to express and codify ethical values in a coded way that can be automatically tested. But… maybe we should try?

I’m also aware of a related issue, if you’re familiar with the Checklist Manifesto by Atul Gawande which made the case for the usage of checklists in other areas of life after their success in the medical profession. The problem is that there’s a paper published in Nature examining whether this push to use checklists in a medical context actually resulted in an increase in patient safety. Spoilers: the answer is - sometimes - “An easy method that promised to cut complications in surgery may nbot be so simple after all”. You may be unsurprised to see that the difference in replication is in part explained by a culture in some hospitals of ticking boxes and not actually addressing the underlying issue. An ethical checklist is a bit like this, and an automated ethical checklist or test suite might go some way toward addressing issues like people resisting implementation. But it wouldn’t do anything toward addressing something like an “inappropriate or illogical” checklist.

So again, all of this as a caveat: automated ethical testing might be interesting, but it doesn’t mean you don’t have to do the work. It might make it easier for you to validate the work after the fact.

There were a whole bunch of other items on the list: “ethicsops” which is partly deliberate snark at the trend for adding “operations” on the end of things that are done in organizations. What would “ethicsops” even be? Some way to operationalize the validation that an organization is living up to its ethical values and aims? “Ethics tiger teams” and “ethics red team” were pointed out as actually being “underrepresented tech twitter” who are doing a bunch of free labor in providing useful criticism of ethically problematic technology that is then more or less entirely ignored by the relevant people.

2.0 Some quick things

  • I was trying to look something up on oreilly.com earlier this morning and got an error message that “Due to the fires and power outages in California, oreilly.com is unavailable.” Now I want to propose an RFC for HTTP 461: Unavailable due to anthropogenic climate change. This would follow in the spirit of in my opinion the somewhat weaselly-worded HTTP 451, “Unavailable For Legal Reasons”.

  • Here’s a great piece in the Mercury News that’s titled: “Kincade Fire: Why your cell phone is silent” that to me is interesting because it refers to stats about cellphone base station sites that are offline due to the California wild fires 2019. It references a report prepared by the FCC that I’d want to look at as the underlying data (and also because such reports are cool) that I haven’t been able to find yet.

  • Digicortex is an engine for simulating and visualizing large-scale “biologically plausible neural networks” and was interesting to me when I saw it via a benchmark on Intel’s all 14-core 5GHz processor. Basically, we don’t have the ability to run a simulation of a sea slug’s neurons at anything close to realtime, if I’m reading this right.

  • Facebook doesn’t want to be left out in the health space, so have launched a preventative health tool. I think this is less interesting because of the health part, and more Facebook flexing its muscles about what it could do just because of its scale as a communications platform.

  • Here is a paper about ethical subroutines in Star Trek, titled “Evil Doctor, Ethical Android: Star Trek’s Instantiation of Consciousness in Subroutines

  • A Twitter account that I love (@70sscifiart, that posts… sci fi art from the 70s) got suspended and I used it as your regular reminder that this is what happens when someone thinks a process (e.g. algorithmic decision making) works “at scale” solves a problem without realizing that the problem isn’t solved until you can flexibly and easily handle exceptions. I mean, you think you’ve figured out how to solve for dealing with, say, “abusing API keys” in the context of reposting content from a Tumblr. What actually happens, again and again and again, is that the cost of false positives has been effectively externalized due to your automation (ie: you appear not to care about false positives) and if you’ve been caught, you’re lucky if you happen to be able to get the attention of a human at twitter who can intervene. This is pretty much just like examples of egregious immigration rulings in the U.K. that attract attention and lobbying from their local MP - the exception path is to find a human who can circumvent the system. And this has costs! It interrupts people at Twitter in doing their jobs! If this kind of thing was treated as a sort of dev-ops / content-ops / moderation-ops issue, you’d imagine someone would spot the human intervention occurrences and figure out a better system. If this was a priority, of course.


Oh my gosh that was a lot of words and on the one hand I’m sorry and on the other hand, hey, you can always unsubscribe.

In theory I’m writing another newsletter tomorrow, but I may not because this is just a really busy week.

I hope you’re well, I’m having a nice albeit busy time in D.C. and if this was relevant or interesting in any way, or you just want to say hi, send a note! hope you’re well, I’m having a nice albeit busy time in D.C. and if this was relevant or interesting in any way, or you just want to say hi, send a note!

(Oh and remember to like and subscribe and mash those potatoes?)

Best,

Dan