Minecraft 设计思想
Volume I · History and BusinessDissecting the Business Model

The Grey Economy

Free mods sell the paid game, and the biggest server networks take millions a year without paying Mojang a cent. This chapter explains why Minecraft's community creation is at once gaming's most open ecosystem and the one platforms find hardest to take a cut from.

This chapter's thesis: Minecraft's community creation is at once the most open user-generated-content (UGC) ecosystem in gaming history and the one platforms have found hardest to take a cut from. This is not a contradiction but two faces of the same fact: the authorship sits exactly where the official side holds no technical chokepoint — the Marketplace sells goods; mods sell the authorship.

1The problem

Start with two real professions, both attached to the same game.

The first is the mod author. His work may have been downloaded tens of millions of times and sits installed on millions of machines worldwide; his entire income, in the typical case, is a donation link on the download page plus a monthly points payout from the download site — most months, not enough to rent one desk in an office. The second is the server operator. His network may keep tens of thousands of daily players, selling memberships, cosmetics, and season passes, with annual turnover in the millions of dollars — and of that, Mojang's share is exactly zero.

Put the two professions side by side and a strange structure appears: the people closest to the official content — the ones making it — barely earn; the people furthest from it — the ones charging admission — earn a lot; and the official side can reach neither. Only with Bedrock Edition's Marketplace did Mojang take its first cut of player creation in a store of its own — on a different product line, in a different content format.

This chapter answers: why is Minecraft's community creation both the most open ecosystem and the hardest one to tax? Why did those two adjectives grow on the same branch? This is more than community gossip; it is a platform question: when you hand the author's seat to your users, which legitimate places are left for you to collect money from? Minecraft spent a decade-plus demonstrating both sides of that question.

Scope first. The three layers of economy discussed here — mods, servers, and the grey layer skirting the license — all belong to the world dominated by Java Edition; the Bedrock Marketplace economy and the Chinese regional-edition contracts are separate institutions, covered in "Java and Bedrock" and "Revenue Structure" and borrowed here only for contrast.

2Free mods sell the paid game

Begin with a counterintuitive claim: a substantial share of Minecraft's sales were, indirectly, sold by free mods.

The reasoning is not complicated. Take a player in 2012 hesitating over ten-odd euros — what did they actually see online? Streams of modpacks — curated bundles of dozens of industrial and magic mods — demonstrations of redstone computers, the legend that "you can even turn this game into Pokémon." None of it came from Mojang's marketing budget; all of it came from freely distributed community work. The more extravagantly community content answered "how much can this game become," the smoother the conversion to a paid copy. The mod ecosystem was this game's de facto marketing department, working free — and its long-term insurance against the first killer of long-selling games: running out of things to do.

The causal chain has a testable converse. If free mods did not matter to sales, the rational official move would have been to fold mods into a paid system early. By the time of writing, history has answered: absorption never happened (the mechanism behind why is this chapter's last section). And the hardest supporting evidence for "free creation raises the game's value" comes from the official side itself: in the Bedrock store, players have downloaded community creators' content more than a billion times, paying over 350 million US dollars in total — willingness to pay for "more content," verified by the official store's own ledger.1 The willingness exists; Java's structure is what keeps the official side from collecting it.

The structure has pedigree. In the game's earliest years, free access was itself the growth engine: the Classic phase hung on a web page, free for anyone; the paid Survival Test was where money changed hands — a free entrance first, conversion discussed once addiction set in. The mod ecosystem stretched that funnel across the game's whole life: owners of the paid copy face no ceiling on content consumption; onlookers watch the free mod videos from outside the funnel mouth and may jump in at any moment. Other games made "free trial" into a demo file; Minecraft made it a creator class that has kept producing for over a decade.

3Mod labor: unpaid, donated, tipped

The Java mod ecosystem's default price is zero. That is not modesty; it is structure.

First, the scale of the labor. A mid-sized or larger mod is an engineering project of tens to hundreds of thousands of lines of code, and ten years of continuous maintenance is normal; every major Java release sends the entire ecosystem through migration again ("Mods as the Second Author" describes this endless rework). For this labor the community has long recognized only three pay structures: nothing at all is the majority; donations are the minority, from a pinned donation link to a Patreon subscription; and platform payouts are the third — download sites pass part of their advertising revenue to authors. The third deserves its own look, because it is the exact inverse of a "platform cut": the platform does not charge the author; it splits ad money with them.

The largest operation is CurseForge. After passing from Twitch to Overwolf, it has run an author points system: each month's reward pool is converted into points by downloads and popularity, points are redeemable for money, and contracted projects have separate arrangements with a higher share.2 The upstart Modrinth also opened ad-share monetization to creators. These channels are real, and their ceiling must be stated honestly: a full-time living goes to a tiny head, while the overwhelming majority of authors hover at the level of "server bills reimbursed." Measured by the ratio of passion-powered labor to pay, this may be the most lopsided head-versus-tail split in the game industry.

Why not simply charge? Not because authors would rather not — because technically it cannot be collected. Mods are freely distributed files; once released, no mechanism prevents copying, forwarding, or re-uploading. Sites carrying each other's catalogs is community normal, and the moral plea "please download from the original post" does not stop mirrors. In a market with no technical fence, pricing power falls to zero. The only assets a head author truly keeps are two: attribution and community reputation — which is why this ecosystem's reputation mechanisms (does it update diligently, does the author answer quickly, is anything sneaky buried in the code) are far more developed than its commercial ones.

Only with this labor system in view does "second authorship" show its real weight: not an honorary title but a partnership with no salary, paid in devotion, carrying half the product's content. The mods chapter recorded the official stance of a decade-plus — "don't ban, don't pay, don't take over"; this chapter adds its economic meaning: free labor is this product's largest externalized cost. Mojang gets the world's largest content ecosystem without hiring a content team, because the ecosystem feeds itself — and sells a few hundred million copies of the game on the side.

No wages, however, does not mean no property system. The ecosystem replaced copyright enforcement with a culture of authorization: a modpack author who wants to bundle someone else's mod is, by etiquette, expected to ask first — and the download sites turned the etiquette into a feature, with permission status displayed on each mod's page, in writing. The release page of a large modpack commonly carries a whole section of "credits and permissions," like a contract appendix. The system is enforced not by courts but by community memory: whoever carried off someone else's work without permission is remembered on the forums for years. Enforcing by reputation costs almost nothing and works startlingly well — because what it settles in is prestige, and prestige happens to be this market's only hard currency.

4Server monetization and pay-to-win

The server economy is far more straightforward than mods: it has a technical fence. To play on your server, a player must connect to your machine — that alone lets operators charge, by selling "privileges": faster starting gear, extra land-claim slots, exclusive titles, the ability to fly. The usual names are Rank or VIP — paid tiers granting perks free players lack.

The business went through a period of excess. In the mid-2010s, selling balance-affecting advantages — pay-to-win, meaning money that makes you stronger — was close to the industry default: big servers paid for hardware and staff with it, small ones recouped their electricity bills. The turn came in 2014: Mojang updated the End User License Agreement (EULA) and drew a line for server monetization — selling cosmetics and conveniences that do not affect balance is fine; selling "paying makes you stronger than players who don't" is not.3 The full story of that storm, and each side's case, is told in "The Server Is the New Game"; here we record only its economic consequence: the big servers survived, and did better than before.

The reasons are worth unpacking. The EULA did not ban charging; it reclassified what could be sold, from "advantages" to "cosmetics and conveniences." The head networks already had the game's widest audiences and the strongest art and engineering teams, and the regulation swept their grey competitors out of the arena: only big players could afford compliance. Skins, particle effects, pets, crate keys — goods that never touch balance filled the space advantages had held, and the big servers' revenue came to resemble a free-to-play MMO's cosmetics model rather than a casino's. For small servers, life genuinely got harder; for the ecosystem as a whole, it is a textbook case of trading concentration for order through regulation — who lost and who gained depends on which link of the pipeline you stand in.

Just as worth writing down is the relation between legality and royalties: server charging is fully legal (within the EULA's frame), but it pays no royalty. A server network taking in millions of dollars a year yields Mojang exactly zero — for a technical reason: anyone can download the server software, opening a server requires no approval, and there is no link in the chain the official side can technically pinch. "The Server Is the New Game" called these operators "local government"; here is the public-finance phrasing: one of the largest untaxed economies on earth, where the central government holds a constitution (the EULA) and no tax office.

The business also grew an employment layer that is easy to overlook. Behind any decent mid-sized server there is usually more than the owner: builders paid commission per blueprint, plugin developers taking custom orders, artists drawing skins and promotional art, administrators drawing monthly "wages" — in yuan, dollars, or in-server perks. Around the servers a genuine outsourcing market grew: owners who cannot build a cathedral hire a build team; servers wanting exclusive gameplay buy out a developer's plugin outright. The server economy, in other words, does not only "charge players" — it is the employer sector of the Minecraft world, and a remarkable share of young people's first technical or artistic income came from a game server. No official ledger records this, yet it is the most "real economy" part of the whole ecosystem.

5The grey layer: an ecosystem fact, not a how-to

Any honest economic chapter runs into grey zones. This section states facts and offers no operating guide.

Three things have long existed at the edge of the license in the Java ecosystem. Pirated clients: ways to log in without paying; community attitudes have always split — most players call it theft, while a substantial minority cite Notch's own early, publicly stated leniency: most people downloading pirated copies were never going to pay, and some of them become buyers later. Offline-mode servers: the server software contains an option to skip Mojang's account verification — a legitimate feature built for LAN play; the portion misused beyond that constitutes piracy's multiplayer layer. Reposting: mods, maps, and skins copied to every corner, the original author's attribution and wishes routinely lost — the grey layer of the creation side, larger in scale than the other two and the quietest in its damage.

The official strategy toward all three, over a decade-plus, sums up as "clear boundary, restrained enforcement": the EULA and the account system draw the legal line clearly, but a mass pursuit of pirating users has never happened; real enforcement concentrated at two ends — commercial fake launchers and infringing stores passing themselves off as official, and, from the 2020s, tightened online verification after account migration. That is not soft-heartedness; it is arithmetic. Minecraft's sales structure counts on emerging markets, where penetration rode precisely the "play first, pay later" culture; exterminating the grey layer would cut off the game's own future customers. The judgment is falsifiable: if the grey layer were pure loss, the year-by-year tightening of verification should have accompanied falling sales — the observed trend is the opposite.

One honest sentence more: the grey layer is not a romantic story. The same layer nourishes malicious attacks on servers, unauthorized commercial use of authors' works, and a pile of bundleware that uses "free" as bait. The price of an open ecosystem is always paid in both directions — good things spread fast; so do bad ones.

6Two kinds of UGC: the Marketplace sells goods, mods sell the authorship

Now the opening question can be answered: why did one brand grow two opposite creation economies?

There is exactly one criterion: whether the official side holds a technical chokepoint — a necessary link on the path content must take to reach players, which only the official side controls.

Bedrock Edition has one. Its extension format consists of officially defined data files; its player accounts live inside Microsoft's account system; the platforms it runs on (consoles, phone stores) mandate review and controlled distribution. Together these three facts put every necessary path from content to player in official hands — and so the store works: contracting, review, a token currency, the cut, revenue share; every standard part of a platform economy is present. Players have paid community creators hundreds of millions of dollars in total, and official and creators each get what they came for.1

Java Edition has none. Mods modify things at the bytecode level; distribution runs over websites anyone can stand up; nothing binds an account to content. For the official side to take a cut, there is no valve to turn — you cannot attach a meter to the entire internet. So Java's UGC took the other road: an unpaid authorship, settled in reputation and donations; an un-taxed server economy, bounded by the EULA but sharing no revenue. Money is not absent from this road — there is plenty of it — but every pipe it flows through bypasses the official side.

So the two economies are not two ways of doing one thing; they are two different goods. The Marketplace sells content goods: skins, maps, texture packs — one-time purchases, used as bought, the author standing to the platform as supplier to channel. Mods sell the authorship: what they offer is the position of "helping define what this game is," the author standing to the game as partner to platform. In a goods economy, the platform's greatest anxiety is inventory; in an authorship economy, the platform's greatest asset is its unsalaried partners. Mojang learned the former on Bedrock and lives on the latter on Java — one company, two identities, the line drawn by where the chokepoint sits, not by sentiment.

"Take a cut of Java mods" is not untried; every attempt has fallen in the same place. In 2012, Mojang publicly planned an official mod repository and official API; the conclusion was silence. Across the decade that followed, "an official store absorbs the mods" was proposed in the community again and again, and every time it snagged on the same three things. First, format: anything entering an official channel must fit the official format, and the value of Java mods lies precisely in exceeding the official format — push mods through a review pipeline and you admit only "the unimportant kind." Second, authors: a mod author's pull comes from free distribution and open source; accepting a cut means surrendering distribution, and the head authors would refuse first — and a store without head content is an empty shelf. Third, players: a decade-plus of the "free, open, try-then-donate" convention has zeroed out payment expectations; the mod priced in the store has a free twin on the mirror site one click away. None of the three walls is a legal wall; all are structural. Legal walls are the fastest to climb; structural walls do not need defending at all.

The chokepoint decides the economy

groundworkwhether the official side controls a necessary path from content to players → control it and the store stands → fail to, and the authorship can only settle in reputation → openness and hard-to-tax are the same fact

The judgment is falsifiable: if the official side ever built a working cut-on-mods mechanism for Java Edition, the chokepoint argument would be wrong — it would mean a payment pipe technically existed and was overlooked. In a decade-plus no such attempt has succeeded (the 2012 official mod API initiative fizzled; the mods chapter tells it), and this book takes the bet: as long as Java Edition keeps its defining trait — any code can be injected — that pipe will not appear.

7What this chapter leaves behind

  • For players: the dozens of mods on your disk and the server you live on are backed by an entire economy of unsalaried labor holding up half the house — next time you pass a mod author, even if you never click the donation link, leave a like.
  • For developers: what you can take is the chokepoint argument: before designing any UGC platform, draw the necessary paths by which content reaches users and count the links you control — that count decides how large a cut you can take and how much freedom your community keeps. What you cannot take is Java's historically formed exceptional position: flourishing without a chokepoint rests on a reputation-settlement system the community spent twenty years accumulating — not something designed; something grown.

注释

  1. Wes Fenlon, "Minecraft modders have sold a billion mods, made over $350 million," PC Gamer, 2021-04-28; see https://www.pcgamer.com/minecraft-modders-have-sold-a-billion-mods-for-more-than-dollar350m-since-microsoft-bought-it/ (Nadella announced on the 2021-04-27 earnings call: Marketplace content has passed one billion cumulative downloads and creators' cumulative share exceeds 350 million US dollars; the figures measure store turnover) ↩ ↩2

  2. Wikipedia, "CurseForge," retrieved 2026-09-18, https://en.wikipedia.org/wiki/CurseForge (passed from Twitch to Overwolf in June 2020; the author points system converts a monthly reward pool into Author Reward Points; contracted arrangements carry a 70% revenue share) ↩

  3. Mojang, "Let's talk server monetisation!", Mojang official blog, June 2014, archived on archive.org; see https://web.archive.org/web/20140617065433/https://mojang.com/2014/06/lets-talk-server-monetisation/ (the red lines for server charging: entrance fees allowed but advantages must not be sold in tiers, advertising permitted, servers must state they are not affiliated with Mojang) ↩

On this page