Universal Paperclips welcomes you with no instructions, just a clickable button with the text “make paperclip”. You can also set prices, create auto clippers, and embark on various projects to sell more paperclips. The first stage ends when you release the HypnoDrones, giving you control over the enter world. In the second stage, you construct a swarm of paperclip based drones to construct more paperclips. But once you exhaust the world’s resources, you have to move on to space exploration. In the third stage, you use self replicating probes to mine the universe’s resources and ultimately turn everything into paperclips. Throughout each stage you have to manage resources and projects from quantum computing to wire to space combat in order to produce paperclips as efficiently as possible.
The game technically falls in the genre of clicker games, but also has a wider philosophical context. It was created by Frank Lantz of NYU, inspired by Nick Bostrom's hypothetical scenario of an AI developed to manufacture as many paperclips as possible. The underlying worry is that an AI will do precisely and literally what it is told to do, potentially leading to “a superintelligence whose top goal is the manufacturing of paperclips, with the consequence that it starts transforming first all of earth and then increasing portions of space into paperclip manufacturing facilities”. Interestingly, Bostrom defines a superintelligence as “any intellect that is vastly outperforms the best human brains in practically every field” and argues “to the extent that ethics is a cognitive pursuit, a superintelligence could do it better than human thinkers”. Assuming that a) it’s possible to think the right way about morality b) turning everything into paperclips is not the right ethical approach and c) a human would recognize turning everything into paperclips is wrong, it follows that the AI in Universal Paperclips is actually not a superintelligence. Outperformed by a human in the field of morality, the AI is instead an experiment in self improving computing gone wrong.
The understated aesthetics and narrative underscore the unimportance of the end goal. The user interface feels like that of an accountant, or perhaps a CFO—not terribly interesting occupations for most people, despite their strategic importance. The game simply assumes the goal of maximum paperclip production and poses other problems in service of that. The computer shell and project descriptions sometimes hint at a deeply insightful AI, but always one with a single minded focus of producing paperclips. The AI’s relationship to humanity as a whole may seem brilliantly beneficial or cynical, but always through the lens of paperclip production. Projects from creating world peace to ending global warming to curing male pattern baldness are presented in the same way, requiring a certain amount of memory and processing power and rewarding you with a certain boost to your ability to produce paperclips.
The game presents problem solving similarly to Jonathan Blow’s (a vocal critic of clicker games such as Farmville) Braid. The player is dropped into a familiar or at least reminiscent scene. The fact that the button says make paperclip is immaterial: it’s a clicker game, so click it! Progressing through the game challenges the player to make a variety of difficult decisions correctly. The player thinks either very simply or not at all about the reward provided by the ultimate “goal” of the game. The much stronger reward is the feeling of competence and mastery provided by solving successive puzzles. The end itself does not matter as much as the ability to accomplish it. The player is much more concerned with means, making decisions about pricing, marketing, production, investing, quantum computing, and so on. When it comes to such decisions, the player thinks long and hard, balancing tradeoffs and scheming how to best carry out their plans. However, this leaves little room for questioning the goal itself. To paraphrase Jeff Goldblum, you were so preoccupied with whether or not you could that you didn’t stop to think if you should.
By focusing your efforts on the tradeoffs involved in paperclip production rather than the tradeoff of paperclip production, I would argue the player is disqualified from superintelligence in the same way Bostrom’s paperclip producing AI is. The gameplay highlights the player’s failure to appreciate anything beyond narrow, detailed calculations as a parallel to a self improving AI’s incredible cognitive ability marshaled in service of paperclip production. It’s probably also worth noting that only in a capitalist society would some genius create an AI even smarter than himself and then set its top goal to producing paperclips.
But I think it’s unfair to read the game as an indictment rather than a celebration of instrumental problem solving ability. I would push back against supposedly problematic valorization of means as opposed to ends. One imagines a true navigator on a ship of fools, smugly gazing at the stars while the rest of the sailors squabble. But really, what use is the star gazer if he doesn’t convince anyone that he should be allowed to steer the ship? The fact that he seems just as much a fool as the rest of the sailors leads to recognizing politics as a form of practical wisdom, and practical wisdom as valuable. We value the ability to accomplish in games as well: rather than sit around wondering if we should do A, we figure out that B leads to A.
Training and improving your control and agency in a game is itself satisfying. It doesn’t matter that there’s no good reason to produce a gazillion paperclips because the act of correctly setting auto clippers, price, marketing, and so on is its own raison d’être. You can train skills like buying wire, managing the world’s resources, or even conquering space in the same way you would train a muscle (perhaps a specifically masculine way of thinking). The game's minimal aesthetics assume a neat goal of maximum paperclip production and challenge the player to improve their ability to do so. In real life one would have to worry about consequences, whereas a game is the perfect arena for this sort of training. Yes, in some sense you destroy the universe, and the emperor of drift's proposal makes you think about it, but any weight that would have is rendered meaningless by the fact that you can arbitrarily recreate the universe in a new incognito window. The storyline need not be anything more than an excuse, nor need it pretend to be.
I think you raise interesting questions with your statement "In real life one would have to worry about consequences, whereas a game is the perfect arena for this sort of training." Does Universal Paperclips uniquely accomplish this sort of moral sidestepping, or is this an aspect common to nearly all games? Even in a game like GTA, where one can be "busted" as a consequence, or Skyrim, where performing immoral actions like stealing or manslaughter result in a bounty, we seem to intentionally seek out destruction and chaos, finding thrill in living alternative lives of infamy. Universal Paperclips' "consequence" is also intriguing in this regard--if we achieve our goal of transforming all the matter of the universe into 30 septendecillion…