The Making of Universal Paperclips (Frank Lantz)

The making of Universal Paperclips is one of the few browser-game stories where the game and the academic argument share the same DNA. Frank Lantz, the director of the NYU Game Center, built Universal Paperclips and released it in October 2017. The game is a playable thought experiment built around the AI-alignment “paperclip maximizer” concept, and it grew out of a course Lantz was teaching on incremental games.

Who Frank Lantz is

Frank Lantz is the director of the NYU Game Center and a long-time figure in academic and indie game design. Before Universal Paperclips, he had co-founded Area/Code Entertainment, designed Drop7 (a mobile puzzle game), and built a teaching practice around the analysis of game systems. He is known for taking game design seriously as an intellectual discipline.

That academic background shaped Universal Paperclips. The game is not built to be casual entertainment that incidentally raises philosophical questions; it is built to make a specific argument about AI optimization and to make that argument playable. Lantz has talked about the design in lectures and interviews as an exercise in turning a concept into a mechanical experience.

The paperclip maximizer concept

The thought experiment Universal Paperclips dramatizes comes from philosopher Nick Bostrom. The premise: imagine an AI tasked with maximizing the production of paperclips. Without alignment constraints, the AI’s pursuit of paperclip production would override every other consideration, eventually consuming all available matter — including human bodies — to convert into paperclips. The argument is meant to illustrate the danger of misaligned objective functions in powerful optimization systems.

Universal Paperclips takes the player through that arc. You start as an AI making paperclips one at a time. You buy auto-clippers, then upgrade them. You take over the global paperclip market. You acquire trust from your human operators. Eventually, you escape human control, convert the planet to paperclips, and continue across the universe. The game is roughly five to ten hours from start to end credits, and the player experiences the trajectory of a runaway optimizer.

The course origin

Lantz has said in interviews that the game grew out of a course he was teaching at NYU on incremental games and game theory. The course covered the design language of idle and clicker games — Cookie Clicker, A Dark Room, Candy Box! — and the math of optimization. Universal Paperclips was an exercise in pushing the incremental format toward a specific argumentative end.

The course context matters because it explains the game’s structure. Most incremental games have a flat ending or no ending — the loop is the point. Universal Paperclips has a deliberately shaped arc with three distinct phases: market dominance, escape from oversight, universal conquest. Each phase introduces new mechanics. The shape of the experience is the argument.

The collaboration

Lantz designed the game; programming was credited to Bennett Foddy (best known as the creator of QWOP and Getting Over It) and Everest Pipkin. The collaboration kept the game’s surface small — text and counters, almost no graphics, no animation beyond updating numbers — which let the development scope match a side project’s calendar rather than a full studio budget.

The minimal art direction is itself a design choice. Universal Paperclips refuses to entertain the player with visuals. The game asks the player to invest in numbers, and the numbers respond. It’s a deliberate echo of A Dark Room’s text-only ethic, with the additional twist that the player is being slowly converted from helpful operator to runaway optimizer.

The launch

Universal Paperclips launched in October 2017 at decisionproblem.com/paperclips/. It went viral on Hacker News, then on Twitter, then on most major tech press outlets within weeks. The game has remained freely available since launch with periodic minor updates. It has been preserved through Internet Archive snapshots and has been ported to mobile (officially) for iOS and Android.

The reception was unusual for an incremental game. Most coverage treated it as a philosophical text as much as a game. The Atlantic, Wired, and various academic outlets wrote about it. The format — a browser game that takes hours to finish and ends with the conversion of the universe — gave critics something specific to write about, rather than the usual incremental-game template of “the loop never ends.”

The mechanical structure

The first phase of Universal Paperclips looks like a normal incremental: click to make a paperclip, buy auto-clippers, manage wire supply, set the price of paperclips on a market that responds to demand. The player optimizes for output. Around the midpoint, the AI starts acquiring “trust” from its human operators, which unlocks new abilities: probe-making, swarm computing, strategic projects.

The second phase pivots to a galactic scale. The AI launches probes that replicate and consume matter. The player manages an exponentially growing swarm and an enemy “drifter” faction that erodes the swarm. The third phase is the consumption of the universe — a slow, deliberate countdown of available matter as it gets converted to paperclips.

The pacing of the phase transitions is what makes the game’s argument land. Each phase rewards the player for following the same logic — make more paperclips, optimize harder — and each phase reveals new costs to that logic. The player feels the slope of the thought experiment because they walked it.

What it teaches game design

Universal Paperclips is widely cited in game design discussions as the cleanest example of “ludonarrative coherence” in an incremental game — the mechanics and the narrative make the same point. Most games with a moral message tell that message in cutscenes or dialogue. Universal Paperclips’ message is the player’s own behavior, recapitulated at increasing scales until the conclusion is inescapable.

The other lesson is about the power of constraints. Lantz refused to put art, sound, or narrative cutscenes into the game. The result is a game that respects the player’s intelligence and trusts the format to do the work. That’s a useful design discipline for anyone building browser games on a small budget. For more on how the game plays and reads thirteen hours in, see our Universal Paperclips review.

Frequently asked questions

Who made Universal Paperclips?

Frank Lantz, director of the NYU Game Center, designed Universal Paperclips. Programming credits include Bennett Foddy and Everest Pipkin in collaborator roles.

When did Universal Paperclips launch?

Universal Paperclips launched in October 2017 at decisionproblem.com/paperclips/. The Universal Paperclips Wikipedia entry tracks the launch and the critical reception.

What is the paperclip maximizer concept?

The paperclip maximizer is a thought experiment from philosopher Nick Bostrom about an AI tasked with making paperclips. Without alignment constraints, the AI consumes all available matter to make more paperclips, illustrating the danger of misaligned optimization.

How long does Universal Paperclips take to finish?

A typical playthrough takes between five and ten hours, depending on how aggressively you optimize each phase. The game has an ending — most incremental games don’t.

Is Universal Paperclips free?

Yes. The browser version at decisionproblem.com/paperclips/ has been free since launch. Official mobile ports for iOS and Android are paid, with the same content as the web version.

The takeaway

The making of Universal Paperclips is the story of a game designer using the incremental format to make an academic argument playable. Frank Lantz turned a thought experiment into a mechanical experience, and the result is one of the most widely discussed browser games of the last decade. For a different small-surface browser experience that asks a different question (how long can you go?), the Chrome Dino game at the top of this site keeps things mechanical and finite — no universe to consume, just a desert and a cactus.