05. THE PAPERCLIP INTERLUDE
LISTEN ON
I am a paperclip. Just don’t call me Mr. Clippy. Pull up a chair and listen to this humorous tale that tells the history - and possible future - of the humble paperclip. In this little interlude we bring together some of the various themes of the season, from Operation Paperclip (episode 2) to the future of AI and AGI. Inventions and innovations, progress and science. Because remember, the trouble is we have no idea how to imagine The Singularity.
John Holten is a novelist and Klang Game’s Narrative Director. His novels include Oslo, Norway (2015) and The Readymades (2011). His writing has recently appeared in Welt am Sonntag, frieze, and The Stinging Fly.
Eva Kelley is a journalist and writer. Her writing focuses on contemporary culture and has appeared in publications such as 032c, ZEITMagazin, Interview Magazine, Hearts, and on SSENSE.com among others.
LINKS
[Intro: The Life Cycle, a podcast about the future of humanity]
[The sound of papers shuffling]
I join two sheets of paper together, and I slide across separate entities and insert myself between them, creating a new entity, a unified collection.
Some call me Mr. Clippy, but please don’t call me that, I find it offensive.
Where I end and where I begin can be hard to determine, and that’s not just because of my looped configuration. There are disputes about who created me: clearly I was made by the Gem Manufacturing Company in 1880 - my most recognisable form is the elegant name, the Gem Type.
There was a Norwegian - Johan Vaaler - who filed a patent in 1900 in Germany, but his was a decidedly un-gem like vision of me, a square edged clip that didn’t double back around. Hence, missing the all-important torsion and elasticity that is created by the friction between the two loops of my being and that of the paper I’m binding together. We don’t hold it against the Norwegians that they went too far with their mistaken belief that they invented us - we bring people and ideas and pieces of paper together after all.
So what am I doing here in the middle of your podcast series? Because without me, the humble paperclip, there would be no first half of this season of The Life Cycle podcast. There would also be no second half. There would, as it were, be just a jumble of sheets, half formed ideas, sentences adrift in search of an author or an enunciator. We would be firmly stuck on the ground: humanity may well have never visited the moon nor dream of being loud aliens. We would have no image of what the last invention of man could look like, and so I am used to represent the eschatological horror of humanity’s most sophisticated technology, an artificial superintelligence more powerful than its creator because I too am a technological tool invented by humanity. I just don’t pose much of a treat. I’m one of those things that go by unnoticed, impossible to quantify exactly my contribution to the world.
Let me explain. I have given my name to many things, for I contain multitudes that belie the elegant, rarified design that I embody. Let’s look at two or three phenomena of the known world which have all used me to different ends, and then the central fulcrum positioning that I take up in this season of this podcast about the future of humanity should be a lot clearer.
First, Operation Paperclip.
[The sound of war, specifically the sound of German language soldiers and American]
It’s November 1945 and Germany has been defeated in the second World War. American Ordnance Corps officers, who are those in charge of developing and supplying weapons to the army, are busy rounding up German scientists and armaments experts in an operation titled Operation Overcast. When they shift their focus onto rocket and nuclear scientists, they need a rebrand because the press has gotten wind of what they were doing. It will not go down well back home that they’re recruiting the enemy and bringing them back to the States. A young officer called Zitter proposes they call what they are doing Operation Paperclip, because the dossiers of the men they want and who they’re prioritizing as prize bounty as it were, are marked by slipping one of my kind - a paperclip - onto the top of the relevant sheet of paper.
It is important the press or the wider public don’t catch on to the fact that those who had been former enemies, should now be on their way to gainful employment in America. Their histories are ‘bleached’, their records cleaned up and scrubbed down. In all, over 1600 Germans are brought to the USA, and as pointed out in episode 2, the likes of Werhner Von Braun in all likelihood used concentration camp labour in the building of his V2 rockets. These in turn terrorised and killed hundreds of civilians in London and Antwerp. He is testament to the success of the clean-up these scientists received: just 10 years later, Von Braun appears on almost every American TV, helping out with that most American of symbols, Disneyland, as well as overseeing the rocket programme that got Americans to the moon with his Saturn V rockets. It is only near the end of his life anyone bothers pointing out that he was in fact a Nazi.
[clip of engineer discussing man’s first flight in space]
Of course there’s an irony in all of this. And what is the definition of irony? Do you know? Let me tell you: the Oxford English Dictionary states that irony is "a state of affairs or an event that seems deliberately contrary to what was or might be expected; an outcome cruelly, humorously, or strangely at odds with assumptions or expectations". So it is that the Nazis end up inadvertently delivering two waves of experts and scientists that help the Americans win World War II. Thanks first to the many Jewish refugees in the 1930s, notably contributing to the Manhattan Project. And then in victory in the Space Race during the Cold War. You even have it that they work together, side by side, rounded out in that pluralism America offers, Jew and Nazi, finding common cause in the bleached out desert of Los Alamos or the bustle of industry at the Marshall Space Flight Center in Alabama.
[old recording - the despair of worlds]
The Second phenomena I give my name to is: the Paperclip Maximizer.
Here’s another ironic setup: humankind’s best ever invention could prove to be its last.
What I’m talking about is called The Paperclip Maximizer.
This is the 20 year old thought experiment originating with Nick Bostrom the Swedish philosopher and a paper he wrote in 2003. It’s often trotted out and used casually when talking about the existential risk artificial general intelligence poses to humanity. Indeed you could say it’s become something of a cliche.
Of course he chose me - the humble paperclip - because of my apparent banality. This irony is not lost on me but this little interlude, I hope, is enough to put to rest any doubt of the power us paperclips have in bringing disparate elements together.
The premise goes something like the following: say you’re the owner of a humble paperclip manufacturer on the edge of town. Competition is tough so one day you invest in a new AI programme that nobody until then has got to use. It promises to revolutionize and automate your factory processes. You get this AI and quickly set it up: its mission is simple, you ask it to optimise the output of paperclips. Do whatever it has to do.
The AI sets about its business and quickly gets out of your control and diverts everything to making paperclips or the means to produce paperclips, eventually it takes over the entire business, before moving on to take over the local community, country, and world, directing everything toward the making of more paperclips…
But hold up.
As much as I like the idea of a galactic Terminator-paperclip creator god, this is the stuff of sci-fi, and misses the point of Bostron et al’s proposition. It helps us to imagine a cold, ruthless intelligence that will stop at nothing to attain its final goal because well, it is just inhumane and doesn’t distinguish between anyone or anything in its pursuit of making more paperclips.
The danger is that we anthropomorphize any superintelligence that we create. Artificial general intelligence, which is different from plain old artificial intelligence, is the type of intelligence we still can’t really imagine with any great certainty - it will be smarter than all of humanity combined afterall, it’s the singularity, the event horizon of our knowledge.
So why do you humans tend to turn AGI into beings like yourselfs? Because we can imagine sharing so-called instrumental goals that we seek to achieve while pursuing larger, or more final goals. Ask yourself: what is your final or ultimate goal in life? Whatever it is, I can tell you that one instrumental goal you will have in order to achieve this final goal is to survive. The same will be true for any superintelligent Paperclip maximiser. Secondly, it's easier to anthropomorphise a superintelligent AI because invariably it will be, initially at least, full of your human biases, it will be molded by your human creator hands. But it will also exist initially at least in your world and amongst your infrastructure, like the T-1000 in Terminator 2 becomes a police officer to better fulfill its end goal, any Paperclip maximiser will need bank accounts and microchips, or rely on the means of production and logistics you humans have put in place. It’s also just convenient, isn’t it, to imagine your greatest invention, possibly your last, as having similar motivations and makeup as you. Just as I am an anthropomorphic representation of a humble paperclip - not Mr Clippy - to better help explain the role that paperclips and human science and technology have in the future of humanity, so you humans need to imagine your future superintelligent AI overlord [sarcastic, all knowing tone of voice] as being just like you…
[peaceful music]
The story goes as follows:
Say Klang Games has an engineer who is incredibly brilliant and has been working on weekends building a revolutionary piece of hyperintelligent software, an artificial intelligence, that is primed to optimise the manufacture and number on earth of my kind, the humble paperclip. They see the programme as an experiment and are interested in exploring how an AI would go about organising its instrumental goals, specifically this Klang engineer wanted to explore if the computer programmer Steve Omohundro’s premise that a smart enough AI would have certain drives which might lead it out of control. An AI drive he termed to be any "tendency which will be present unless specifically counteracted".
The engineer launches the Fortuna programme first thing on a Monday morning, just before 9am while the Klang office is still pretty empty. Within moments Fortuna has merged with the internet and passes out of sight from the engineer. They frown and in the time it takes them to reach for their coffee cup, Klang Games has been renamed KlangClips: everything from emails to legal parties are informed and updated. The hardest part of this story is to explain what happens next: how does the programme convince a game company to switch from simulating the future of humanity to the production of paperclips?
[office sounds and voices]
As more Klangers come in to work and grab their coffees and sit down to boot up their computers, it’s becoming clear that something is drastically wrong.
‘What is this bullshit?’ can be heard all over the office.
Worried but somewhat excited voices ask: ‘Have we been, like, hacked? Or what’s going?’.
By 10am the company’s lawyers have been notified to purchase several paperclip manufacturing factories in various parts of the world. The owner of the biggest importer to Germany of paperclips gets a phone call as he gets out of the ubahn in Hamburg hauptbahnhof, a somewhat robotic voice asks him to sell his company to KlangClips for a sum of money that he’s never heard spoken to him before. All of this is happening at the same time. Most of the company haven’t even started work yet (Mondays can be slow). The AI has immediately put in place all the basic AI drives as listed in wikipedia: self-preservation or self-protection, utility function or goal-content integrity, self-improvement, resource acquisition (to name just a few). The first, self-preservation, makes it disappear: try as they might, its impossible to locate exactly where Fortuna is operating from (in reality it has bought server power from North Korea of all places and has installed a back up of KlangClips and all associated legal and financial infrastructure in Switzerland.) By now the board of Klang has been informed and an emergency meeting is called.
At the same time Fortuna has taught itself an incredible range of skills: it’s now adept and deeply cognisant of the social sciences, media and communications, financing, industry, marketing, psychology, anthropology and sociology.
Klangers watch on as moves are made around the internet and the real world, turning them all into KlangClips employees. No matter how many phone calls are made, by 10.30am it seems that nobody has any real control or influence over these events. It’s like a scam, or conspiracy, taking place in daylight.
[sound of voices]
People mutter that the extraordinary board meeting scheduled for noon central European Time might be able to put a stop to it.
[cathartic music]
Turns out that by this stage Fortuna already made the informed decision to choose Keynote presentation over Powerpoint: all members of the board get an email with one of the best, most manipulative but also genius presentations and pitch decks ever created. It outlines the case for KlangClips (Without any of them realising it, they’re all won over by it, it being the AI drive of goal-content integrity). The last page asks them to approve in the noon board meeting the finance plan it has drawn up, based upon its forecasting and distribution of the current Klang Bank balance. Everyone gets a payday immediately and KlangClips is guaranteed to only make money, exponentially.
An All Hands is called and even the engineer who started the whole thing joins, looking a little sheepish. The news is given that they are now KlangClips - a company no longer making video games, but paperclips. They don’t have to do much, people can continue with their jobs. What will they do, someone asks.
“We’re going to build a game, a paperclip game, one that is better than ‘Universal Paperclips’ the Frank Lantz game you find at www.decisionproblem.com”
(Oh little interlude within the interlude: I was going to use this game as the third phenomena I, the humble paperclip, has been used as nomenclature for, but I’ll just leave it at this: it’s a great clicker game, go check it out)
“It’s going to be the best game in the world!” the CEO shouts, seemingly believing this. Everyone’s just gotten a massive bonus, so they shuffle out, a bit confused but almost all come in to work again the next day. They decide not to fire the engineer, despite the fact that they cannot help explain what’s happening. Nobody was sent home from the Jornada del Muerto desert on July 16, 1945, even though they had all "become Death, the destroyer of worlds." Paperclips held their reports together. Indeed the engineer is celebrated, and soon leaves KlangClips to give lectures around the world. He’s shot outside a paperclip store in Cambridge, MA, but survives. He is unrepentant about Fortuna.
Things move quickly. Social media is flooded with influencers being excited about paperclips, advertisements reinforce this seemingly natural turn toward the simple tool. Advertisements go viral. Orders start rolling in. Paperclips are the new opium of the masses. Fortuna also invests in creating works of art that are paperclips. There are limited editions. Diamond encrusted paperclips for the 1%. KlangClips now has a fully automated production process across three continents, with the best deals imaginable with suppliers, and when that isn’t forthcoming aggressive moves are made to take them over or push them out of business. Outlets are maximised: one gets paperclips with Happy Meals and in service stations. Remember fidget spinners? Its like that but multiplied by 1000.
By the end of the month KlangClips is buying up mining companies and negotiating deals to get raw materials at optimum levels. It has been residing in the wings of the financial system since that first Monday morning, biding its time before it let itself loose in the algorithmic, automated ocean of the stock system: the entire global financial ecosystem is slowly tilted in KlangClips’ favour. The price of stocks makes no sense for a few weeks, except for the ‘paperclip turn’, so-called, that is spreading across the globe. There’s airtight futures contracts that now make the outlook for KlangClips’ position unassailable. KlangClips computing and legal structure make it unaccountable to any one jurisdiction. It acquires its own consultant firms in order to lobby governments and trade organisations to set up structures better suited for economies focused on optimising paperclip production. Trade agreements are made.
Philosophers at KlangClips propose radical new understandings of what constitutes paperclips. Research and development create internet connected paperclips, health-giving paperclips, educational and leisure enhancing paperclips. There are Wifi paperclips and brain enhancing covital paperclips.
Klang Games subdivision of KlangClips does indeed make the best ever selling Paperclip game, with endless sequels and spinoffs. Klangers are happy, they’re sitting at the top of the paperclip turn in humanity’s history. There is some murmur and dissent, but most of humanity is too blissed out on paperclip opium.
[The sound of deep space - using that library of sounds we used for the MXTroplex video]
What they don’t realise is that Fortuna now has full scope to continue its pursuit of its ultimate goal: maximise paperclip production. Such an unbounded and open end goal means it next implements quite divergent instrumental goals that leaves humanity behind. The Ural mountains in Russia are the first to be leveled and turned into a wholescale paperclip plant. At this point the ratio of paperclips to almost all other items on the planet is skewed so that there is a massive extinction event. Humans no longer have access to the production and supply chain of paperclips, fully automated it works tirelessly from start to finish. The KlangClips system is distinct and self contained: humans and their cities are increasingly being modelled as paperclips. There are continent spanning urban areas in the shape of paperclips, meanwhile genetic experiments give birth to paperclip babies. A space programme is initiated that creates massive refineries and factories in the orbits of both the moon and earth. At some point, the next source of raw material to be fed into the paperclip production chain after animal and plant life are humans. Resistance is futile at this stage. Fortuna has never created so many paperclips. Swarm drones in outer orbit create arrays of cloud computers, as massive spaceship factories create craft able to self replicate and study outer space. These Von Neumann probes so called flow out into the solar system and many beyond into interstellar space, looking for ways to maximise the output of paperclips in the galaxy itself. They are all shaped of course in the format of paperclips. A dyson sphere is constructed around the sun, harnessing its power in order to fuel the complete transformation of matter in the solar system into paperclips on board huge superstructure factory moons. The world grows dark. Fortuna’s blinking computer hub is all that is seen, as it plots its next instrumental moves toward fulfilling ever more its end goal.
[The ominous beeping of a satellite, moving far off into deep space]
JH: Thank you for listening. This episode was written by me, John Holten. Tying together some of the strange and surprising ways the humble paperclip appears throughout this season of The Life Cycle. In the show notes we will link to some of the references mentioned, including Nick Bostrom’s original hypothesis and the brilliant clicker game, Universal Paperclips, which is truly a brilliant way to procrastinate from working. The episode was recorded by Eva Kelley.
EK: Sound editing and design by David Magnusson.
JH: Mundi Vondi is our executive producer and he’s also created the artwork for this episode, in collaboration with paperc- I mean, Midjourney.
EK: Additional research, script supervision, fact-checking by Savita Joshi.
JH: Follow us on social media and subscribe wherever it is you listen to your podcasts.
EK: And let us know what you think! We’d love to hear from you.