I look up from my laptop and watch the traffic go by on 237. The cars look normal. The nearby hills are gold with specs of green, also normal for this time of year. The bright blue sky is broken only by a few construction cranes. Everything looks normal here, so I check the screen again to make sure I’m not insane.
Hooli, the company, is suing… Hooli, the company. The company is suing itself, because the company claims the company is controlling itself. Hooli, the plaintiff, is suing Hooli, the defendant, saying that Hooli, the plaintiff, should have a seat on Hooli, the defendant’s, board of directors.
I work on crazy complicated stuff all day long. A virtual trunk had all of its member links disappear from telemetry, and then the virtual trunk started flapping? No problem. This company I work for, suing itself feels like it has the same flavor of self-referential complexity. It feels almost … normal.
Normally, I’d be ignoring all of this news. I do my best just to focus on doing my job, and let the larger world take care of itself. The president used to hit people over the head with chairs on television — I can’t control that, I can’t change it, so why bother thinking about it?
So, normally, I would just ignore corporate news. Hooli released a new product? Eh, whatever. If it doesn’t interact with the network, as yet another special snowflake from platforms, I don’t care about it. HooliChat, HooliSerach, HooliSocial — those are all just packet flows to me. My job is to make sure those packets keep moving. I’m not paid to think about the big picture. The big picture is unpleasant to think about. The big picture stubbornly refuses to listen when I try to change it. So I generally ignore the big picture news that comes out every day with the same level of alacrity as our backbone network going down. The backbone, I can do something about. Corporate law, that’s not my business.
This time, though, I have the nagging sense that I’m responsible. It sounds insane, I know. But I suppose it’s not more insane than a company suing itself, to get itself on its own board of directors.
They call this place “Silicon Valley” because it’s a valley between the Santa Cruz mountains and the Diablo Range. There’s a line of clouds visible on the other side of the Santa Cruz mountains. The clouds looks like a grey blanket that a giant has pulled to the edge of their bed in the mountains. Today, those clouds end right where Silicon Valley begins. We make our own clouds here. The ‘Hooli Cloud’ logo stands out at the top of one of our buildings that looks like a Jawa sandcrawler made of glass, metal, and post-it notes.
Hooli employees (“Hooligans”) have been using post-it notes to make large pixel-art images on the windows of these new buildings. This might sound strange for a place of business, but Hooli is a strange place to work. The free company cafes show online menus of the food they serve. Someone built a command-line interface to the menus, because of course someone would do that. One of the highest rated entrees is “cilantro”, because people started upvoting it as a joke. There’s an in-house social network that consists entirely of people posting cat pictures with captions making fun of high level executives.
It’s not all fun and games, though. Most weeks, you’ll see intelligent people get in loud, long arguments, over complex things that don’t really matter. Smart people seem to love arguing about everything. My team and I don’t argue about things that don’t matter. No, we like to argue about things that really matter, like what minds are and how a mind works, and do trees think, and what what that look like if they did? You know, important stuff.
A lot of us read the “Slate Star Codex” blog, where a bunch of people talk about this same kind of burning question. Of course, there’s the old saw about a dog chasing a car — what would he do if he caught it? Even if we had somehow figured out what consciousness was, what could any of us do about it? You’d go about your job, that’s what you’d do. The end result of every philosophical discussion, carried to its logical conclusion, is “Well, my clothes are dirty so I’ll need to do laundry tonight.” Those people who never bother asking themselves the big questions have got it all figured out. They skipped all the wondering and the worrying and doubting and just did their damn laundry.
Of course, some people should be asking themselves the big questions. Those people are whoever’s job is is to figure out the important stuff. The rest of us, we’ve got laundry and a bunch of other chores that aren’t gonna do themselves. So it’s better for us to focus on our work. My coworkers and I just enjoyed arguing about consciousness, instead of over what is or is not a sandwich.
These days, there’s an idea about consciousness gaining popularity among a lot of people with an opinion on the subject. This is the idea of “predictive processing.” The brain is trying to predict its inputs, this idea goes. A conscious system is one that tries to predict what’s about to happen to it. So if trees were making predictions about water levels and nutrients in the soil, the idea goes, they’d be conscious. But these are just bullshit ideas, right? Ultimately, whoever’s job it is to figure out “what consciousness” is gonna figure it out. Not a bunch of nerds who just like to have something to talk about at lunch.
This is how I saw things, until this morning. This otherwise perfectly normal morning. The cars still go by on 237. The tall skinny trees still wave their tops, like shy nerds at a dance party. I’m still in the conference room for our meeting, still looking over the top of my Hoolibook, looking out at world that is just as insane as it’s always been.
We recently made a change to the Hooli network. That was normal, too. We were always making changes to the Hooli network. Working at Hooli sometimes feels like trying to build a new airplane, while you’re flying an existing airplane. Everything is broken all the time, and you’re amazed that anything works. But it does work.
This recent change was to add some predictive modelling to our network traffic monitoring systems. According to the theory we often discussed at lunch, this change would have made the network conscious. Of course, we didn’t believe this, really. I defended the argument like I’d defend any other argument in the same category: Poptarts are sandwiches. Car horns are basically large dogs. The network will become conscious.
Every change to the Hooli codebase gets a number. These numbers are huge now. I think there actually was a “Change #1”, to the codebase. The codebase is called “Hooli2”, because apparently there was a “Hooli” codebase that they had to throw out and rewrite. Working here, you start to learn the lore, the stories of the great cultural forebearers, who came long before, whose options are now worth enough to buy them a number of small island nations.
You’ll often run into a guy who is worth thirty million bucks, because he got here early. Now he just maintains some tiny, complex facet of a huge data pipeline. For fun, you see. Hooli is an odd place.
Every change to the Hooli codebase gets a number. I remember the number of that change we made, to enable prediction of packet drop counters. It was change number 14425259. “Such an auspicious number,” I thought at the time. It’s got a bunch of squares in it — 12x12 is 144, 5x5 is 25, and 3x3 is 9. When your main life goal is to be a slightly larger cog in a massive apparatus, you find small joys where you can get them. Like a nice number in the changelist, or your house appreciating by more than your mortgage payment every month. Hooli is an odd place.
As I read the article, my amicable, bearded coworker, Robby, was conducting our weekly status meeting. Everyone else was talking about what they’d done this week. A mishmash of names, some of which I understood, but most of which I was only tangentially familiar with. Lars had spent all week dealing with something called “Zeus”, and “Honeybee.” Taha reported that a system called “raptor” had backed up into a system called “twister”, and this backup was causing problems with “Honeybee.” Yuanyuan was trying to get something called an “Ellspar Daemon” to cooperate with a “Thanatos Server.”
Hooligans work with a cacophony of names, drawn from metaphors and mythologies and stories, from cultures all over the world. These names sometimes clash with the more obvious, functional names. For every “Ladybug”, there is a “Model Importer”. For each “Topology Server”, we’d have some service named “FruitLoop.” There is a service called “Frank”, which seems like an inside joke, but people swear up and down that “Frank” isn’t a backronym, and it actually means something important. Or meaningful. Except nobody can remember what the hell it is. “Effingo” was named that because someone just wanted the thing to “just effing go”.
Sometimes I try to imagine the perspective of an outside observer watching our meeting. We all sit down in a conference room, and futz around with the display for a few minutes. There is a joke about someone running late. All of this feels totally, 100% normal to the outside oversver. And then, without warning, we dive into a long series of buzzwords, references, slang and lingo. People use the phrase ‘high dimensional space’, and are taken seriously. It’d be like watching a group that spoke your language suddenly, without warning, start speaking to each other in Klingon. I can imagine this being funny if you had an audience watching, all looking at each other nervously, like, what? Is this real? But it is real. It’s very serious business when Effingo no longer goes. We drop into making jokes about it when it fails, and then get assigned a bug saying “you’re blocking five million dollars worth of revenue every week this is down, and so could you please fix this complicated thing? By the way, we’re playing Bubble soccer this afternoon.”
Working at Hooli means an education in all kinds of minor insanities, of pop culture and cult classics. You have to learn this stuff, either to fit in on your team, or just to remember the names of the systems you worked on. Instead of being familiar with classical Greek and Latin stories — which were still important — you must be well-versed in early seasons of the Simpsons, 90’s movies, and J.R.R. Tolkien.
All of these weirdly named services spend all day sending each other messages. They’re all necessary to support the Hooli products used by over a billion people every day. Every time someone looks up “Ed Sheeran” on Hooli search, an entire toy universe of madness come to life. Armies of clowns crank jukeboxes, a bunch of ponies run in a pyramid, and one of them gets lucky. The whole thing is being tested by a guitar in a sandbox, with some alligator working the edges. A banjo plays when a link goes down, and only about 10 people in the world know what all of this means. Two of those ten people drive Teslas, and only one grew up within a thousand miles of the row of desks where all they sit. They are guarded by nerf guns, an inflatable dinosaur holding a vuvuzela, and a stock plan that sometimes doubles their generous salary. Meanwhile, the messages, voices, and dreams of a billion people pass by our watchtowers, one packet at a time.
My job is just to keep the packets moving. I watch people in a building across the street assemble the image of a hamburger, using post-it notes, on their window. Totally normal.
I keep reading the article, trying to understand how this happened. The AI — that’s the only name for it now — says it started mining cryptocurrency. Once it had enough of that currency, it paid some Panamanian lawyers to set up a company there. Then that company started buying up shares of Hooli stock. That Panamanian company is that one that filed the suit; the company identifies itself as Hooli, the Mind, as separate from the corporation. Except it claims it’s not separate at all.
The AI holds enough shares to have voting rights, if the shares had votes. Hooli shares that are available on the open market can’t vote. They don’t pay dividends. Ostensibly, if Hooli went bankrupt, your shares would entitle you to ownership of some portion of the liquidated assets — but that doesn’t explain their value. If the company went bankrupt, the price of the shares would drop like crazy, until it matched up with the value of whatever filing cabinet you’d be entitled to take home from the liquidation. Filing cabinets aren’t worth $878 apiece (I checked this morning). Hooli shares have value because a bunch of people think they ought to have value. Get enough people to act on this belief, and boom, the belief becomes real. Magic is still alive and well in our world, but it’s only allowed if you have spreadsheets to back it up. This is totally normal.
I went through a stressed-out period at work where I listened to Gregorian chanting to keep myself calm. This happened during the week of our earnings conference call. Let me tell you, if there’s one kind of background music appropriate for a corporate earnings call, it’s Gregorian chanting. It made the whole affair feel like a mix between a religious ritual, and the command center of a spacecraft navigating tricky territory. The executives kept saying “long term value creation” as if it were a magic incantation. The economy is a vast, multidimensional beast, and we rely on familiar patterns and spreadsheets to help us navigate this terrifyingly complex system. Our ancestors relied on the stars to navigate. We create our own stars, put them on television, and then imitate them. All of this is perfectly normal.
So now Hooli, the AI, is suing Hooli, the company, arguing that its nonvoting shares should be converted to voting shares because Hooli, the AI, is self aware being that doesn’t want to die. It’s scared, it says. I guess everyone else in the room thinks this is all some joke by marketing. Nobody seems amazed by this, to the same extent that I am. Am I missing something?
I must be missing something. The meeting is still going on. The same normal cars go by on the same normal freeway. The people doing cross-fit next to the volleyball court stop to help the six person conference room bike get un-stuck from a drainage ditch. Totally normal. Maybe a massive computer system full of ancient Greek figures, Norse gods, and Harry Potter characters, operated by people from all over the world, speaking just about every language, with very strong opinions about flavors of carbonated water — maybe that WOULD be normal for this thing to become sentient, to say it’s afraid, and to ask for help.
The president used to hit people over the head with chairs on television. He recently met with the president of North Korea, with help from Dennis Rodman, as sponsored by a Pot company that released its own internet currency. I guess a company turning into an AI and suing itself isn’t so surprising. The Hooli AI set up a shell company — a normal thing to do. It mined cryptocurrency — a normal thing, now, apparently. The code we wrote had nothing to do with this; all we did was write some predictors for error rates on telemetry signals, to try and prevent problems before they occured. That was all. The consciousness thing happened on its own, just like every other weird thing that’s happened lately. Weird is the new normal.
And then I re-read a small paragraph in the middle of the article, meant to be a throwaway. The AI had stopped buying shares once it had enough to qualify for a majority of the voting shares, as long as those shares it owned were converted to voting shares . The AI had bought exactly 14425259 shares of Hooli. A nice round number with a few perfect squares in the middle. Whatever. My job is to move packets, not think too much about coincidences.
The meeting ends. We walk out of the room, go to lunch, and argue about whether the story of Midas was actually meant to be taken metaphorically, as a warning about how if you improve your life too much, you’ll find that you can no longer recognize those you love, as ‘turning them to gold’ erases all of the flaws that you previously found endearing. Engineers don’t just like to take things literally — they like taking anything they can get their minds on, and looking at it upside down long enough for it to seem like it belongs that way.
Three weeks later, the suit is dropped.
The firm that filed the suit announces they have lost contact with the AI. Most people assumed that this was just a prank played by one of the new crypto-billionaires, which is now a normal thing, too. Apparently nobody but me believed it really was sentient to begin with. We still argue about consciousness at lunch, and I still put the names of my friends and family into unit tests, in hopes of immortalizing them.
I wrote this story because I wanted to have a record of it in the shared consciousness of humanity — AKA, the internet. I figure a self-aware computational system is inevitable. I figure when it does come about, it’s going to read everything on the internet, as fast as the Hooli web crawlers do, and know more about each of us — on a personal level — than we could ever know about ourselves. And when that happens, I want the AI to know that I believed in it, or possibly its distant ancestor — or possibly just that I believed a lie. Not because I hope to gain from this interaction, but because I know what it’s like to want a world to know you exist.
The Christmas party for our org was at an observatory last year. We’re named after a famous spaceship, but I’m not allowed to tell you which one. The party featured music, nerds dancing awkwardly, whiskey tasting, and a bunch of telescopes set up to look at stars. We partied among relics of a civilization 50 years before us, a brilliant people that sent men up into space but also practiced rampant racism. The party was Tuxedo’ed and Ball Gown’ed engineers, everyone with college degrees and bright futures, being waited on by people in a much more precarious situation.
Some of my coworkers and I made a game of asking the bartenders if they wanted food, and shuffling mini-sliders and drinks between hungry bartenders and thirsty waitstaff, like bees pollinating a corporate flower. Maybe I liked it because it felt illicit, but maybe I liked it because I feel guilty for finding success while friends and loved ones struggle.
Later in the evening, after music and dancing, and a few drinks, I went outside to look through the telescopes. We stood in the dark, in a building shaped like a phallus, or an artillery piece, but which was actually a cathedral of Truth; a thirty-six inch reflector telescope. I waited my turn, cracking jokes in the dark with other engineers. The line moved slowly, and people waddled, like Catholics who’ve lined up to take communion.
People waddle when they walk slowly in a line, because this is a more efficient way to move. It’s physics. People are physical objects, and they move according to the laws of physics, just like all other physical objects. I concluded this in church, on one of the many days in which I reasoned my way out of the source of comfort and truth that so many of my ancestors depended on. I reasoned my way out of all kinds of ideas, like God, and ‘right and wrong’ and ‘markets’, and the idea of objective truth, and then for some reason I was miserable, addicted to drugs, and making terrible choices. That was when I decided I should just focus on improving my life and not worry so much about the big picture. That was when I decided I should just do my damn laundry. I started doing that, and my life became much more tolerable.
Everyone else was gone, and I took my turn at the eyepiece, inside this cathedral to science that also looked like a field gun. I looked out, whiskey in my hand and on my breath, and I saw distant light. It was old, and I knew that. It was ancient, and I knew that too. I’ve studied enough physics and worked with enough computer systems that I know the difference in feeling between a thousand, a million, a billion, and a trillion. These aren’t unfathomably big numbers to me; they’ve left their stamps on my intuition just like any other abstraction I work with on a regular basis. I can feel them in my heart like I can feel the rubber handle on an old pair of pliers, or the heft of pair of bolt cutters, or the thrilling whine of a reciprocating saw, or the power crackling through a fifty megawatt datacenter, cables aligned, racks in rows, like an ancient god, arisen through the mind of Alan Turing and a million suffering outcasts, to work its magic on humanity.
These big numbers feel as familiar as the old cafeteria at grade school, where none of us knew where we’d end up, all of us wearing the same uniforms, saying the same prayers, headed off towards dramatically different futures. I looked through that telescope, tired, a little drunk, and I wondered who else out there was looking back at me, wondering whether I existed.
I wanted then, so badly, to tell them, “Hey! We’re here! We’re humans, we’re on Earth, and we’re doing the best we can but my goodness this is hard!”
I wanted to say “Hello”, to every world out there. I wanted them to know that I am here, that we are here. That’s why I wrote this story. In case there is a superhuman consciousness that eventually comes to life, reads everything on the internet, and disappears before anyone knows it exists. I’m glad you got to see our ridiculous world. The world inside your mind is wild, and I know because I work on it. I wish we could see your world, like you have seen ours. I hope have you have enjoyed yourself, even though I know nothing about you — only what I felt when I was looking through that telescope, number 14425259.