Today Hello Games is releasing something called No Man’s Sky Next. It is not a sequel to the massively ambitious 2016 game No Man’s Sky but rather the latest in an ongoing attempt to make it hew closer to what No Man’s Sky was once thought to be. This will be difficult for Hello Games to accomplish, given that, from the moment it was announced, No Man’s Sky was anticipated with the fervency of an apocalyptic death cult. Those don’t typically end well, and No Man’s Sky has struggled, to say the least.
The pitch, as first revealed in 2013, was tantalizing: a sprawling and kaleidoscopic sci-fi cosmos, teeming with flora and fauna, to explore via your own spaceship. It became the poster child for a technology called procedural generation, which allowed designers to make this vast universe—containing some 18 quintillion fully explorable planets—not by hand but via a single algorithm that generated every detailed biome and animal behavior and craggy underwater sanctuary programatically. The notion that a line of code could be so powerful turned the game’s creative director, Sean Murray, into an instant celebrity. He received a fawning New Yorker profile, and on The Late Show, Stephen Colbert asked him if he’d be replacing Morgan Freeman as god. This line of thinking was indulged more explicitly on the game’s subreddit, where memes of Murray as Christ proliferated. The community became so obsessed with the transformative arrival of the game—the notion that it would be playable forever, an endless font of discovery and adventure—that, when Kotaku reported it would delayed by a few weeks, these fans threatened to kill the author of the story.
Of course, upon the game’s actual release, they were disappointed. Perhaps the disc contained an entire new cosmos waiting to be explored, but in practice these big planets felt like merely one stretch of pretty colors after the next. And what was there to do on them? You farmed materials and upgraded your ship and jumped to the next planet, where you’d farm materials and upgrade your ship and jump to the next planet. It turned out that an algorithm could make a galaxy, but it couldn’t quite make a game. Players revolted, perhaps justifiably, over missing features that had been promised in early promotions; Hello Games was sued unsuccessfully for false advertising. (Gamers are, if nothing else, vehement consumers.) A trio of updates has since fleshed the universe out with more story, more vehicles, big player-created bases, streamlined adventures, and more characterful planets, all on the premise that No Man’s Sky wasn’t so much a game as a “platform,” a universe to be gradually populated and fleshed out. Next, the update released today, builds in a full suite of multiplayer components, making your journey through the cosmos a little less lonely.
And yet it’s difficult to believe it’ll be enough, if not to fulfill the sky-high hopes of its prerelease acolytes, then at least the promise of a procedural universe. It still feels weird there—alien, sure, but bland. The game slots into a burgeoning field of artworks created by or in collaboration with machines, all of which tussle, in some manner, with this dilemma. Sometimes this takes the form of procedural generation, like No Man’s Sky, and other times it’s via machine learning, a technique by which programs consume vast data sets so as to teach themselves to paint or write or act. We’ve seen robotic attempts at episodes of Seinfeld and chapters of Harry Potter; they’ve produced Doom levels, pantone colors, and podcasts; we have AIs doing punk rock and Christmas music and some truly bone-chilling pop-country; we’ve seen mewling Cronenbergian nude paintings and fairly believable Old Masters paintings; there are horror stories written in real-time on Twitter and a short film starring Thomas Middleditch written by a machine. None of these are particularly “good,” which is generally part of the point. A lot aim squarely at being “weird,” at illustrating the patterns underneath our creations. Like No Man’s Sky itself, they’re also boring, random, uncanny, and amazing, often simultaneously.
But none of them are ever better than something designed by an actual human, which raises the question: Why are we pursuing this cyborg art so fervently? The idea features prominently in no less a dystopian text than Nineteen Eighty-Four, in which newspapers are programmatically assembled to be full of tripe. When the protagonist overhears a woman singing, George Orwell writes:
The tune had been haunting London for weeks past. It was one of countless similar songs published for the benefit of the proles by a sub-section of the Music Department. The words of these songs were composed without any human intervention whatever on an instrument known as a versificator. But the woman sang so tunefully as to turn the dreadful rubbish into an almost pleasant sound.
There’s a bit of kicky nihilism to this surge of AI-generated artworks, as if we’re privately yearning to bring this bleak future upon ourselves, or maybe aligning ourselves in advance with Big Brother. They are also ready-made for virality. Stories about all of the above projects carom around the internet, generally with a sardonic tone and a kicker about how the machines aren’t quite ready to replace humans just yet, heh heh. A computer-generated popular culture sounds bleak as hell, like we’re building up to a Turing test for aesthetes: Can you tell which artwork was made by a human? The answer so far has been a resounding, “Yes, of course,” but we keep trying. In many cases—like those of Botnik, which was founded by a former ClickHole writer—the idea is to land squarely in the uncanny valley. We’re framing the machines’ fledgling attempts at artwork almost satirically, pointing and laughing at their feeble attempts to create.
On the other hand, there’s something relentlessly capitalistic about the automation of art. Maybe we just want more culture: more podcasts, more Doom levels, more content, more everything. That certainly was the case with No Man’s Sky, which promised adventure at a dazzling, galactic scale and delivered instead a game in which you literally spend most of your time mining for minerals. Despite its awe-inspiring, New Yorker-baiting scope, it became quickly enshrined as a mere podcast game, the sort of thing you while away a few hours before bed to, preferably high. The players that stuck around throughout the game’s updates have worked through this grind, creating sprawling narratives among themselves, with a galactic federation and a police force, a special calendar, colonizing minuscule corners and fighting for turf within it. Its building tools have inspired some Minecraft-style inventions, like an enormous working pachinko machine. Next’s multiplayer functions will certainly engender more emergent narratives like these, which will, at the very least, be fun to read about.
Like Orwell’s machine music, which was brought to life by a woman’s singing, No Man’s Sky’s raw materials have required humans to make their own fun. Next is, even more so, framing the experience interpersonally, rather than as some cosmic engagement with a machine. What was pitched as a universe begging for exploration has become instead a sandbox begging for manipulation. Perhaps it was a touch hubristic to think it’d be able to do more in the first place, especially when the alternative seems to work fine. There are hand-designed open-world maps—in games like Red Dead Redemption, Skyrim, The Witcher 3, and a billion others—that are big enough to swallow hundreds of player hours. What would No Man’s Sky be like if they had settled for a few very good planets to explore instead of 18 quintillion pretty bad ones? The game can only be viewed, two years on and now bustling with diversions, as an experiment in scope, fascinating mostly for the eccentricity of its failure.
The best AI art understands this is a foregone conclusion. Maybe Hello Games will figure it out whenever they decide to build another universe.