AI can revive old games in just 8 hours, but the era of mass revival may be short-lived

If you've been scrolling through videos lately, it's not hard to notice a phenomenon: many of the online and mobile games that got you through countless nights of your childhood and youth, only to be tragically shut down in the end—like QQ Tang, Call of Duty Online, Yao Guai Zheng Zhuan, Girl Cafe Gun, and Kai-Ri-Sei Million Arthur—have successfully won their revival match.

What revived them wasn't the game companies themselves, but some obscure small creators whose follower counts are only in the hundreds or thousands.

Reviving old games isn't a new topic, but reviving them in bulk—and online games at that—and then being able to play them online together after revival, that's pretty wild.

Sure, let's allow that masters exist among the common folk, but... aren't there a few too many common-folk masters?

Xiaofa's skepticism comes from the fact that reviving old games sounds like putting new wine in old bottles—just throw something together and call it done. In reality, this work is absolute drudgery, pure labor-of-love grunt work.

Some guys who've tried reviving old games themselves probably know: do it badly and all your effort is wasted, with time lost measured in years; do it well and you still can't make money, and you're quite likely to receive a lawyer's letter—you might not even know what you're doing it for.

How big an undertaking is manually reviving an old game?

There happens to be a recent example: back in 2020, a bunch of people started a group chat wanting to turn an old version of Gameloft's Asphalt 8: Airborne—a memory for a whole generation—into a single-player game.

That endeavor dragged on for over 6 years. Over 6 years, friends—a sixth grader would be a college freshman by now.

It wasn't until recently, on October 3rd, that a relatively complete version was finished and released to the public.

Why was it so painstaking?

Because what they made wasn't a cracked version, but a version based on single-player mode that optimized every system.

For example, the game's currency system tied to microtransactions became a points system unrelated to spending; for race formats, a new single-car classic mode was added; the game gained a custom matchmaking feature where players enter a Rank score and the system automatically generates AI opponents near that bracket, simulating the original PVP...

If you said half this team's energy went into optimizing the game's own systems, then the other half went into figuring out how to make it work on today's devices and operating systems.

Our software and hardware are now countless times more powerful than in the past, but many old games simply can't run.

Take the Female Gunner and Male Ghost Swordsman versions of Dungeon & Fighter from ten years ago: you download and install the package through hell and high water, open it, and it instantly crashes. Many ordinary players who want to relive it can only hunt for old machines in the secondhand market.

Game immortals use game assets to hand-build single-player versions, and it takes 6 years. Only with that comparison can you feel how abnormal this era of great game revival is.

Question: guess how long those online and mobile games being revived in bulk these days took?

I'll just give the answer—at the fastest, less than half a day, just a few hours. In those few hours, you don't even need to type a single line of code; you just issue commands and act as a keyboard commander.

English text is being prepared.

The revival process is nowhere near complicated—it even gives you the feeling that "I could do that too," so comical it's a bit hard to take.

To save you time: the whole thing took 7 hours 53 minutes, used about 1.08 billion tokens, and sent only 52 messages to the AI.

Forget you—even I kind of suspect this guy fudged the numbers. 52 messages? Bro, sometimes when I'm writing articles and looking things up, the messages I send to AI casually exceed 52.

1.08 billion tokens sounds intimidating to laypeople, but for developers it's really not much.

Converted to cash, 1 billion tokens can cost thousands, tens of thousands, or even more at the expensive end, and at the cheap end it's genuinely cheap—recently there was news of a deputy bureau chief hand-building a mini-program with 1 billion tokens, and he only spent 20 RMB, cheaper than my phone bill.

The revival process for Soul Seeker isn't complicated to explain either.

Simply put, the game previously transmitted information through CDNs—servers closest to players—but after service shut down, the assets on those servers weren't deleted and were free for anyone to grab at zero cost. Is this the internet's sharing spirit? I'm tearing up.

Yes, friends, it's that simple—you really could do it yourself.

We can no longer trace who raised the curtain on the era of great revival, but online, people seem to have spontaneously started hunting for the installers of games that carry their youthful memories, shouting at their computers, "Revive, my love."

So we can now play the granddaddy of all modern gacha games, Kai-Ri-Sei Million Arthur, released in 2012 with its Chinese server shut down in 2018—even a PC version, enjoying an experience the older generation never had.

We can also play what was once countless people's regret: Girls' Frontline, which operated for eight years but was forced to shut down at one point in 2024.

In addition, Destiny Child, Merc Storia, Ar Tonelico 2, Era of Celestials, Forever 7th Capital, and a whole bunch of other games have all been dug out of some forgotten corner and grandly revived.

Whenever people discover a game they were once obsessed with has been reincarnated, one second they're still mourning and telling their ancestors in the family shrine, and the next they're sitting bolt upright from their deathbed—joy spilling off the screen.

Those with weaker hands-on skills just make wishes in the comments section.

As the era of great revival grows louder and louder, people's standards for "revival" keep getting higher.

At first, people thought it was enough as long as the game could run and they could play it normally.

As things developed, many people began building all kinds of online systems for games that originally had online features.

For example, a QQ Tang revived by a small UP with a thousand fans used 5 billion tokens, restored most of its features, and even adapted it for local, LAN, and online multiplayer, compatible with Windows 10 and 11.

This scene of teeming vitality, with all things competing to flourish—doesn't it make you sigh that technology shapes a better life?

Hey, don't start secretly celebrating just yet. The era of great revival is impressive, but it's also legally risky.

Along with the revival of countless games, some people have wondered: if past games can be revived, doesn't that also mean currently operating games could be made into local versions?

This bold idea was quickly put on hold by the creator community—after all, nobody wants to end up with a pair of silver bracelets.

But players prefer to keep to their own little circle and just enjoy simple happiness; resellers don't care about any of that.

On various platforms, people selling revived games for profit have long been everywhere.

You burn billions of tokens, stay up several nights, and make a game to play with fellow fans—and then it gets taken and listed for two yuan eighty-eight somewhere else.

How many copies sell is hard to say; the main issue is that it's disgusting—disgusting to you, and disgusting to the game companies.

So-called era of great revival, to put it bluntly, is the era of big private servers and the era of big piracy; in essence, it still uses game companies' existing resources to recreate game content.

If it's just for fun within a small circle, companies might turn a blind eye and pretend nothing happened.

After all, in the past two years, the player movement to stop killing games has been in full swing, and wanting games to "live forever" is a unanimous demand among all players.

Domestically, the atmosphere may be relatively harmonious, with not many players going head-to-head with companies over this.

Companies also rarely make single-player companion servers, and feeling guilty, they naturally go easy on this kind of gray-area great revival movement—of course, they also figure it doesn't matter much, so what if they let all the people be creators and handcraft it themselves.

This line of thinking is normal, because the games being revived were themselves shut down precisely because they no longer fit market trends.

Usually there might be some discussion, but once actually made, it's only a three-minute fad within a small circle.

English text is being prepared.

Last year, 17173 also ran a poll letting players vote for shut-down old games; after a certain number of votes was reached, they would represent players in communicating with operators to push for revival.

Currently the game with the most votes is called "Mecha Whirlwind," with only 13,804 votes, and progress is stuck at finding the copyright holder—revival is far off.

Game companies are actually quite helpless, because reviving old games isn't something you can decide on a whim; even without considering whether costs can be recouped, many factors come into play.

But as the saying goes, I may not play it, but you can't not have it.

The best case is the official side making a single-player companion version; since there isn't one, delegating the rights to the player community is still somewhat reasonable.

But bro, openly selling "revived versions"—that's not just being cocky, that's provoking the companies.

When secondhand dealers get a lawyer's letter is hard to say; some creators who want to revive games may give up because of this; game companies will also gradually tighten those open resources originally forgotten in the corners of the internet; and putting shackles on AI is almost a done deal.

In March, we interviewed an expert who, relying purely on AI and without writing a single line of code, recreated "The New Legend of Sword and Fairy," "Moon Shadow Legend," and "Legend of Sword and Fairy 2" on web and mobile. At the time, we were still marveling that when developing indie games, AI would sooner or later take over the coding work.

No one expected that this sentiment would come true so fast that people hadn't even reacted yet.

On October 16, the reverse-engineering decompilation community decomp.dev, which had operated for two years, announced its closure. The announcement clearly mentioned that human + AI can significantly speed up the decompilation process.

This is the sorrow of reality: when we realize we are in the golden age of the great revival era, we can already predict that this so-called era will ultimately end hastily like a fleeting flower.

Starfield, which ruined Bethesda's reputation, took eight years and still couldn't achieve seamless landing, while a MOD developer, with the help of AI tools, solved the problem alone in half a year.

Some go left, some go right, some keep pressing forward, but no one will go against the times, turning back all the way.

news.17173.com/content/12222025/090009338.shtml

This article comes from the WeChat public account "Chaping X.PIN," author: Zhang Dadong, editor: Mangshan Laotietou, published by 36Kr with authorization.