Minecraft Bedrock is killing the things that helped it grow.
PocketMine-MP has ended support. NetherGames has shut down. Zeqa moved from PMMP forks into a Featured network. Other Bedrock software survives by paying very different bills. These do not look like unrelated accidents. They look like an ecosystem steadily narrowing the space that once helped it grow.
“Information wants to be free.” - Stewart Brand
I like that sentence from Stewart Brand, founder of the Whole Earth Catalog, because it has the kind of stubbornness the old internet used to have. Information does not want to be free because every piece of work should be treated as worthless. It wants to be free because when information is locked away for too long, the people outside the gate start running out of air. People who want to learn cannot learn. People who want to fix cannot fix. People who want to build something new end up standing in front of a closed door, waiting for someone with access to feel generous.
With Minecraft Bedrock, that feeling has become harder to ignore. The community around servers, plugins, modding, reverse engineering, and unofficial software helped make Bedrock feel alive. It kept players around. It let small server owners build their own communities. It gave someone like me a path from downloading Minecraft PE, finding IPs, hosting a server on a phone, editing PHP plugins, and eventually learning to code.
Then Minecraft became stable, huge, commercial, and mature enough to have Marketplace, Realms, partners, Featured Servers, and a polished official funnel. The painful part is that the software and servers that helped the Bedrock community grow are now being pushed into a technical environment where surviving is harder every year.
I do not have Mojang or Microsoft’s internal documents. I cannot claim there was a meeting where someone ordered PocketMine-MP or NetherGames to be killed. But pretending that only official press releases count as reality is also dishonest. Sometimes a pattern becomes clear only when many decisions are viewed together. Here, that pattern is clear enough. Independent Bedrock servers increasingly depend on non-public information, partner access, a few maintainers, unstable funding, and the goodwill of the partners Mojang has already selected.
That is why this is a criticism, and also a small obituary. Farewell PocketMine-MP. Farewell NetherGames. Farewell to a part of Minecraft Bedrock that was messy, open, buggy, human, and far more real than a curated server tab.
A death that did not happen overnight
On 09.07.2026, the main PocketMine-MP repository on GitHub was archived and made read-only. Only a few days earlier, version 5.44.3 had been released on 06.07.2026. From the outside, this can look sudden. But if you read the signs from the last few years, PMMP did not collapse in one night. It reached the end after being stretched for years.
The PMMP homepage is now a farewell page with the line 2012 - 2026. It says a key developer has left Minecraft behind, and that no one on the PMMP team is left willing to update the project for new Minecraft versions. The team gathered documentation on the update process so forks can try to continue, but the PMMP team itself will not provide further updates.
If you stop there, the easy conclusion is that PMMP died because it ran out of people. That is true, but incomplete. Running out of people is the final symptom. The harder question is why a project with more than a decade of history, thousands of dependent servers, a plugin ecosystem, and a large public repository reached a point where losing a few people meant losing official updates entirely.
To answer that, it helps to read the “I'm retiring” issue opened on 21.03.2024 by Dylan, better known as dktapps and PMMP’s lead maintainer for many years. He said he would step down from PocketMine-MP and move on from Minecraft. He had lost his passion for the game years earlier and mostly launched it to test PocketMine changes. More importantly, he described the future of the project. PMMP had become a foundation that thousands of Minecraft servers depended on, but maintaining it required serious commitment, responsibility, and expertise. It was also a thankless task that did not pay the bills unless a business using it happened to sponsor the work.
That sentence hurts because it is true. PMMP was not a toy project. It was infrastructure. But that infrastructure depended on a handful of people with time, reverse engineering skill, community stamina, and sometimes funding from Lifeboat, a Featured Server network that sponsored PMMP for years. A healthy ecosystem should not place its backbone on so few people and then act surprised when they get tired.
Dylan also predicted a future that now feels very close. Large servers would maintain private forks. Small servers would die off. Not because small servers are lazy, but because understanding new Bedrock versions, updating protocol, updating data, and fixing plugins has become too expensive for most small communities.
Lifeboat saved PMMP, but it could not manufacture a successor
Any fair criticism of this story needs to correct an easy mistake. Lifeboat was not a company that drained PMMP and walked away. Dktapps’ own words show almost the opposite. Lifeboat turned what had been unpaid hobby work into work he could continue doing, paid for the parts of PMMP its network needed, and still allowed the results to remain available to the public community.
On 08.05.2023, dktapps wrote that “without lifeboat y'all would still be stuck on PM3 ... and we wouldn't even be talking about PM5”. In the same conversation, he said Lifeboat did not control the open-source side of PMMP. That is an important difference between a sponsor with legitimate interests and an owner capturing a project.
A year later, dktapps said “PM4 would never have released if i hadn't been hired by lifeboat”. That sentence is enough to place the credit correctly. A Featured Server used its own money to keep an open upstream alive for years, even though it could have kept the work inside a private fork.
But this is also where praise becomes a difficult question. Why did infrastructure serving thousands of servers have to survive because one network voluntarily subsidized it? Why did knowledge of protocol updates and Bedrock Dedicated Server, or BDS, data remain so concentrated that in 2022 dktapps described the overlap between “people with skill to do protocol updates” and “people with reason to do it publicly” as tiny? People capable of doing the work were rare. People capable of doing it while paying the bill for everyone else were rarer still.
By May 2026, that reality was even clearer. Dktapps said both he and Dries, NetherGames co-owner and the developer who led its move to in-house systems, were paid by Lifeboat, yet “it's still not enough to make it worth it”. Money can make labor less invisible. It does not make protocol updates less exhausting, reopen Bedrock, restore a lost interest in Minecraft, or instantly create a maintainer with the skill and desire to carry the responsibility for years.
On 10.07.2026, after the end-of-support announcement, dktapps corrected another easy misunderstanding. Lifeboat had offered to continue sponsoring PMMP, but no one on the team with the necessary skills wanted to continue. The sponsor had not closed its wallet. There was no suitable person left for it to sponsor.
The conversation immediately after that shows that “no one wants to continue” was not an evasive answer. When a member called the cause burnout, dktapps replied that burnout was only part of it. A larger part was years of demotivation caused by the constant stream of problems coming from Mojang. He said he had considered quitting as far back as 2018. In the final years he watched Geyser servers grow and dominate while PMMP carried PHP’s technical barriers, performance work, technical debt, and its distance from vanilla. This is a severe self-assessment by the person who maintained the project longest, not an objective verdict that PHP is useless. It does explain why last-minute funding could not make work that had lost its meaning feel worth doing again.
Farewell notices rarely contain every difficult question. In this exchange, the value does not come from the questioner’s speculation. It comes from the fuller answer the question drew from dktapps. “No one wants to continue” then becomes more than a short notice. It includes burnout, accumulated demotivation, and the feeling that the work had become less worth continuing.
The Lifeboat paradox therefore makes the criticism of Mojang more precise, not weaker. A responsible Featured Server can keep a public upstream alive for years. Corporate generosity is not an ecosystem policy or a succession plan. Lifeboat pulled PMMP back from the edge more than once. It could not fix missing documentation, uneven access, a low bus factor, a narrow contributor pool, or an architecture that had to reimplement Minecraft while Java projects could build around the official server.
PocketMine-MP was a massive doorway
For me, PMMP was never just PHP server software. It was a doorway. I already wrote more technical background in What is PocketMine-MP?, and the more personal memory in Minecraft since 2015, before I knew what an IP was. If I had to compress it into one sentence, PMMP gave many people a way to make Minecraft Bedrock their own.
Not everyone had money for serious hosting. Not everyone knew Go, C++, Java, or network protocols. But PHP, plugins, config files, Poggit, DevTools, old forums, and YouTube tutorials were enough for a student to open a server, edit spawn, adjust economy, add ranks, translate messages, and slowly realize that every feature in the game had real code behind it.
PMMP was never perfect. It was not vanilla. It lacked Mojang-style world generation, complete redstone, mob AI, and a long list of gameplay behaviour. Plugins could turn a server into a soup of errors if an owner downloaded everything without thinking. But precisely because it was imperfect, it taught people to look inside software. There was no magic box. A server was events, packets, ticks, permissions, schedulers, memory, logs, crashes, and many nights where fixing one bug revealed two more waiting politely.
An ecosystem like that does not only create servers. It creates learners. It creates developers. It creates local communities. It creates plugin archives, Facebook groups, Discord servers, GitHub organizations, tiny tools, and a habit of sharing resources. If Minecraft Bedrock has a large community today, those unofficial things cannot be erased from the story. They were the dirt under the foundation. When the house gets tall, people love to forget the dirt. The foundation did not grow from a clean logo.
NetherGames is a rare public accusation
NetherGames shut down forever on 28.06.2026. Its announcement was not a lazy status update about being tired. It was a history of almost ten years. The server started on 18.01.2016, from a small setup running public plugins downloaded from PocketMine forums. It went through the Better Together update in 2017, when mandatory Xbox login damaged player counts for many servers, then merged with Dries’ GameCraftPE. Dries’ technical expertise helped NetherGames move away from public plugins and build its own in-house systems.
During the pandemic, NetherGames reached 3,314 concurrent players. This was not a tiny server failing because it never tried. It was a large server with operations, staff, products, a community, web infrastructure, custom games, and nearly a decade of experience. And still, it shut down.
The most important part of the NetherGames closure announcement is where it says Microsoft and Mojang made it increasingly difficult for non-Featured servers to exist. NetherGames says Mojang deliberately cut non-Featured servers off from technical information that had once been freely provided, and that repeated assurances about restoring that access never turned into reality. Keeping up with Minecraft updates became, in their words, a resource-intensive battle they were never going to win. They also had a meeting with Microsoft in 2024 about becoming a Featured Server, then were ghosted by their contact.
The Discord posts by Callum, NetherGames founder and co-owner, preserve both important halves of that history. The first says NetherGames grew from a mess of public plugins on the PocketMine forums. The second directly describes the field for non-Featured servers as uneven. On closure day, Callum also promised to open source most games and maps so other people could run their own version of NetherGames. A server accusing the platform of closing information answered by opening the parts of its legacy it could release. That makes the farewell sadder, but also better.
Still, Callum’s accusation should not be turned into a single-cause explanation. When a member asked why NetherGames never became Featured, a Support Representative said there were several internal issues he could not discuss, and those issues also contributed to NetherGames never being Featured. This does not refute Callum. It shows that both layers may be true. Mojang created an uneven field and let the contact path go cold, while NetherGames had private internal weaknesses the public could not see. A harsh system rarely needs to be the only cause. It only needs to make every existing weakness more expensive.
That account does not stand alone. On 23.06.2026, days before NetherGames closed, dktapps said he knew almost all of its people. He wrote, “i saw all the nethergames guys slowly get the hope drained out of them” as negotiations over partner status dragged on. This is still the testimony of someone close to the team, not an internal record. But it matches the exhaustion visible in the closure announcement. Before a server switches off, there is often a long period in which the people building it stop believing that more effort will change the outcome.
That is NetherGames’ account, not an internal Mojang memo. But it matters because it lines up disturbingly well with what PMMP’s own technical documentation says. One side is a commercial server saying it was cut off from information and could not win the update race. The other side is open source server software saying that updating new Bedrock versions requires BDS data, and some of that data now requires access through server or Marketplace partners.
These two stories do not mean NetherGames closed because PMMP ended support. NetherGames had long moved beyond public PocketMine plugins, and Dries was a separate technical pillar for that network. That actually makes NetherGames more important as evidence. Even after building in-house systems, even after becoming large and technically capable, an independent Bedrock server still hit the same wall. Bedrock is a platform where crucial information is not distributed fairly.
Zeqa is part of this story too
Zeqa has been quieter than NetherGames about the software behind it, but not quiet enough to leave no trace. The Zeqa Network GitHub organization identifies itself as the network’s official GitHub. It contains NG-PM, forked directly from NetherGamesMC/PocketMine-MP. Its own description calls it a “Modified fork of NG's fork of PM.”
The Git history makes the relationship clearer than the repository name. After diverging from NetherGames, Zeqa added a group of its own commits in October 2025. One says “customisations for use on Zeqa”. The next switches to Zeqa’s BedrockProtocol fork. Other commits add hashed block IDs and move Discord logging into Zeqa core. In the source, VersionInfo names the software NG-PocketMine-MP, while composer.json uses NetherGames BedrockData and BedrockProtocol packages.
In early 2026, Zeqa created another fork directly from PMMP. This time the core was renamed ZeqaMine-PM. It was not left untouched as a bookmark. Wraith0x10, a member of Zeqa’s Administration team, and other contributors added spears, lunge behavior, slow falling, and fixes for experience bottles and splash potions through 17.03.2026. Zedstar16’s OnlineTime repository is also a PMMP API 5 plugin whose README says it is currently used on zeqa.net.
So the historical claim no longer needs to be phrased as “Zeqa seems to use PocketMine-MP.” The public record is enough to conclude that Zeqa ran and customized backends derived from PMMP at least from late 2025 into early 2026. What it does not establish is that every Zeqa node ran the same core, or that today’s production backends remain identical to the public repositories.
Zeqa’s public Discord adds a more everyday layer of confirmation. On 09.07.2026, right after PMMP ended support, a user asked “Pocketmine wont get any updates anymore, will zeqa update pmmp themselfes?”. Earlier, on 20.06.2026, when someone asked whether Season 10 used Nukkit or PMMP, wraith answered simply “PMMP”. On 23.06.2026, in the same chat, wraith identified Jack and Zed as the two people handling protocol and version updates. These fragments do not replace the GitHub evidence and do not identify every current backend. But they show that PMMP was not only an outside guess attached to Zeqa. It was part of the operational vocabulary the Zeqa community used when asking about seasons, updates, and maintenance responsibility.
When I pinged zeqa.net:19132 on 13.07.2026, the public endpoint only returned WaterdogPE Proxy. The WaterdogPE documentation explains exactly why. The proxy is the front door players see, while gameplay, worlds, blocks, and entities live on downstream servers. Those downstreams may run PocketMine-MP, Nukkit, or something else. The current ping confirms WaterdogPE at Zeqa’s edge, but it cannot lift the curtain and name the backend.
This matters even more when seen from WaterdogPE itself. Issue #428 describes a failure when moving a player from a Java + Geyser downstream to a native Bedrock Nukkit downstream through WaterdogPE. The setup had no custom items or special plugins, and connecting directly to Nukkit worked, so the problem sat in the transfer layer between two downstream types. In WaterdogPE Discord, maintainers also warned that mixing server software should not be treated as naturally compatible because flags such as authoritative movement, block breaking, or inventory authority are sent in StartGamePacket. PR #420 later said the transfer changes were tested against GeyserMC, Nukkit, and PocketMine. WaterdogPE is therefore not a backend label. It is traffic infrastructure, and different assumptions may live behind it.
WaterdogPE’s BDS history also shows how living software changes scope. In late 2024, its maintainer said WaterdogPE did no protocol translation, was only a proxy, and did not support BDS at the time. By July 2026, BDS “kinda works,” while test builds kept addressing login, transfer, and same-machine WaterdogPE plus BDS setups. This is not proof that every case is solved. It is an example of a project moving from PMMP and Nukkit assumptions toward additional BDS support through repeated compatibility work. When a shared upstream weakens, the complexity does not disappear. It flows into proxies, bridges, and each downstream.
The public forks also lag behind the 1.26.20 version advertised at Zeqa’s edge. Several explanations are plausible. Zeqa may have a newer private branch, may use a different fork, may keep PMMP for only some game modes, or may have migrated part of the network. Secrecy here is not automatically wrongdoing. A large PvP server has legitimate reasons to keep anti-cheat, matchmaking, economy, and production operations private. But the gap between public code and running infrastructure also gives dktapps’ forecast a concrete shape. Large servers keep private forks and continue. Small communities lose the shared upstream they could follow.
The story then changed direction when InPvP acquired Zeqa. In the acquisition announcement, Mohamed Weheba, announcing the deal for InPvP, called Zeqa the largest Bedrock server outside the Featured list. He said the former owners would move into manager and developer roles and described plans for major investment and Zeqa becoming Featured. The later Season 10 announcement said Mineville and Zeqa were merging to form Mineville Zeqa, while the official store now says Zeqa is part of Mineville Zeqa, a Featured Server.
Aeris makes that concentration pattern sharper. An Aeris Discord announcement on 02.05.2026 said Aeris had become part of InPvP after the recent Zeqa acquisition. On the Zeqa side, the PizzaMC account in its announcement channel also said that Aeris was joining InPvP and the aim was to build a central Bedrock PvP hub around Zeqa. A FAQ the same day said Aeris would no longer be joinable by itself, joins would redirect to Zeqa, ranks would transfer to Zeqa, and announcements would move to Zeqa and eventually Mineville. This does not prove anything about Aeris or Zeqa backends. It does show that Zeqa did not merely survive. It became a place where another PvP brand could be folded into the InPvP and Mineville route.
Dktapps’ comments make this route less abstract. He argued that being acquired by an existing partner was realistically the only way Zeqa was going to enter the partner program. On 07.07.2026, he said “i think zeqa going fully closed source is shitty and selfish but i understand it from a business perspective”. These are dktapps’ judgments, not admissions from Zeqa. Still, they describe the incentive cleanly. Open code feeds the commons, while closed code preserves an advantage for the company paying its own survival costs.
There is no evidence Mojang forced Zeqa to sell. I am not turning an acquisition into a conspiracy. But the shape of the path is difficult to ignore. NetherGames tried to enter the Featured list, says it was ghosted, and eventually closed. Zeqa was a large non-Featured network with a core descended from PMMP and NetherGames. It was acquired by the studio behind Mineville and moved into that Featured network. Zeqa did not die like NetherGames. It crossed the gate.
This may be the most subtle part of the whole picture. Platform pressure does not produce only archives and closure announcements. It also produces concentration. A shared upstream can burn out like PMMP. An independent server can close like NetherGames. A server valuable enough to acquire can continue through a private fork, internal developers, new capital, and eventual absorption into a Featured network. Three different outcomes, all pointing toward the same fact. Remaining independent on Bedrock is becoming more expensive.
Debug symbols sound dry, but they are oxygen
To understand why this matters, we need to talk about debug symbols. It sounds dry. For independent Bedrock server software, it is closer to oxygen.
Bedrock Dedicated Server, or BDS, is Mojang’s official server. A normal BDS release is a compiled binary. It runs, but it is difficult to understand from the outside. Debug symbols map parts of the binary back to useful names such as functions, classes, structures, and identifiers. On Windows, that information is commonly stored in Program Database files with the .pdb extension, which is why these discussions often use “PDB” as shorthand. With symbols, reverse engineers can tell what they are looking at. Without symbols, it is like walking through a city after someone removed every street sign.
In 2019, dktapps wrote 1.13, debug symbols and the Titanic on the PMMP forum. He explained that debug symbols were an essential tool server and mod developers had relied on for years. When symbols were removed, reverse engineering the protocol became much harder. Automatic tools could become useless, and manual binary analysis became far more expensive. He warned that supporting a new Minecraft version could move from a few days to weeks or months.
At that time, dktapps still did not want to say Mojang was trying to kill third-party servers. He leaned toward the explanation that Mojang was trying to make life harder for hackers, while third-party servers became collateral damage. I respect that posture. It avoids lazy accusations. But in 2026, the picture is different. When an unfavorable decision lasts for years, when communities object, when large servers close, and when PMMP documentation itself says crucial data now depends on partner access, it is much harder to call the whole thing accidental.
PMMP’s documentation page Implementing new Minecraft version support in PocketMine-MP describes an update process that needs the matching Linux BDS, BedrockProtocol, BedrockData, block and item upgrade schemas, data dump tools, packet traces between BDS and a vanilla client, and BDS mods that extract the needed information. The crucial note is about Bedrock 1.21.40. Mojang no longer publicly provides BDS builds with debugging symbols. To generate the needed data, developers need access to builds provided by a server or Marketplace partner. Public protocol docs can replace only part of the missing information. PMMP’s conclusion is that third-party servers depend on the goodwill of Minecraft partners.
That phrase sounds polite, but it is heavy. It means independent server software no longer stands on public documentation. It stands on relationships. If a generous partner helps today, the ecosystem survives. If tomorrow that partner changes priorities, a maintainer leaves, NDA terms tighten, or data stops being shared, the community is back in the dark.
The protocol is more open, but the ladder is still tilted
One positive change deserves explicit credit. Mojang published the protocol specification and continues to update it. On 09.07.2026, dktapps said “we don't need to decomp for packet structures anymore”. On another day, he said the coding side had become much easier since the protocol spec was published. This is information being freed, and it did make the maintenance bill lighter.
The protocol is not all of Minecraft. A specification can describe packet fields without providing complete block-state ordering, upgrade schemas, mappings, runtime behavior, gameplay quirks, or the internal organization of a closed BDS binary. On 25.04.2026, dktapps said the mapping mod used for PM BedrockData “breaks every update”. The accurate claim is therefore not that Mojang publishes nothing. What it opened helped substantially, while some of the most expensive work remained behind closed internals.
BedrockData shows that PMMP also created a layer of shared technical memory. pmmp/BedrockData describes itself as data generated from Minecraft Bedrock for PocketMine-MP, including block properties, biome maps, item upgrade schemas, protocol info, creative inventory data, recipes, and other things a server needs to talk correctly to the client. On 11.07.2026, Geyser also had an issue titled “pmmp/BedrockData will not be updated”. Even if Geyser does not depend on all of that data directly, the issue’s existence shows that PMMP’s end made people outside PMMP think about a shared data source. Meanwhile, Mojang/bedrock-protocol-docs says the protocol is shared with server partners and may change between releases. The public protocol docs are a real improvement. The data around the protocol is still the expensive part.
Cloudburst and Nukkit show the same wound at the data layer. In Cloudburst’s #protocol channel, Kaooot, a participant in its technical discussions, said on 14.06.2024 that debugging symbols were unlikely to return, protocol docs were incomplete, and much BDS-only data was no longer easy to obtain. Earlier, when BDS Preview 1.21.10.22 lost symbols, he asked where protocol data, block palettes, and block attributes were supposed to come from if Mojang/bedrock-protocol-docs did not regularly support preview protocol. Another user said the PDB removal was really affecting PMMP. The important part is that these comments were not in the PMMP Discord. They came from a Nukkit/Cloudburst community, which means the data burden did not stop at PHP.
Alemiz, a Cloudburst maintainer, gave a fuller explanation in the same thread. Protocol updates remain possible without symbols, but become harder because the docs contain errors, miss changes, and do not provide enough packet context. Server software also needs game assets such as block palettes, and not all of them can be obtained from packets. He said open source projects were forced to rely on partner servers willing to share assets, while contributors who had tried to persuade Microsoft to retain symbols or offer an alternative received no useful response. This is a Cloudburst maintainer’s account of his community’s experience, not a Microsoft record. But it describes the same dependency PMMP’s documentation later formalized as an update process.
That history predates the complete PDB removal. In 2018, SupremeMortal, who carries the Project Lead role in Geyser’s Discord, said partners received release candidates before the public. In 2019, he relayed warnings that Mojang had removed most symbols from the 1.13 beta. In 2020, he estimated that adding support could take more than twice as long when the client lacked debug symbols. By 2022, he summarized that Bedrock had no public protocol documentation, the community had to reverse engineer it, and vanilla server debug symbols made that work possible. The 2024 shock did not appear from nowhere. It was the endpoint of a narrowing path developers outside PMMP had watched for years.
The same account appears in Geyser’s own history. On 04.01.2020, SupremeMortal wrote that “The community has tried for quite some time now but it doesn't make a difference”. Referring to the removal of debug symbols, he then added, “They didn't even care.” This is the sharp judgment of a project developer after a series of changes he observed, not a record of Mojang’s internal intent. It does preserve the community’s state of mind in early 2020, years before PDB files disappeared completely from public BDS releases.
On 10.08.2020, SupremeMortal supplied the technical half of that frustration in Geyser’s Discord. Debug symbols expose stack traces, function names, and arguments from a C++ program. Without them, “it's a mess of unknown functions”. He also recalled an attempt to match old function signatures against a new binary, but said it was ineffective. This connects two things that are often separated. Maintainer anger was not simply emotional hostility toward a company. It came from watching the technical map that made their work possible get erased piece by piece.
Discord history shows this information ladder taking shape over time. In 2019, before working for Lifeboat, dktapps worried that a partner NDA might compromise his ability to continue open-source PMMP work. In 2023, after joining Lifeboat, he said the network received some material weeks early, but he could not publish it because of NDA. In early 2024, he said directly that he had debug information by virtue of being a Lifeboat member while non-partner open-source developers did not have the same access.
That does not mean everything PMMP published was copied from confidential material. In the same 2023 discussion, dktapps said the things that reached the public could still be figured out without NDA resources, but that NDA resources made the process significantly faster. This distinction matters. Outsiders are not always locked out completely. They are made to take a longer, more expensive route that consumes more people and produces more opportunities to give up.
Partners do not live in a perfect garden either. In June 2024, dktapps said he and the partners had explained exactly why symbols were needed, only to hear Microsoft respond that “featured servers don't need symbols”. By May 2026, he said even Lifeboat was still struggling to obtain usable private symbols and repeatedly receiving material it could not use. Having a pass through the gate does not mean the people inside are always heard.
Partner access should not be presented as magic dividing the saved from the abandoned. It is a tilted ladder. A partner may receive earlier information, different builds, or a more direct contact path. Outsiders can still survive, as dktapps himself said, but it is harder. And when symbols are narrowed even for partners, every dependency beneath them becomes riskier.
A partner is not just a nice badge
Minecraft now has a very clear partner system. The Minecraft Partner Program page says official creators can sell creations on Marketplace, need a high-quality portfolio, must operate like a business, and have submissions reviewed before reaching players. The Minecraft servers page emphasizes the Official Minecraft Server List, Partner Servers inside Bedrock, Xbox profiles, community standards, and safety.
I do not deny the need for safety. Minecraft has many young players. Public servers involve chat, purchases, user-generated content, and plenty of real risk. Mojang has legitimate reasons to control part of the experience. But there is a wide grey area between protecting players and making non-partner servers struggle to breathe. The problem is that technical decisions are pushing too much important information into that grey area.
Partners have a distribution advantage. They can appear directly inside the game, or have an official route to Marketplace. Partners have a trust advantage because they carry the official label. And according to PMMP’s documentation, some server or Marketplace partners have access to BDS builds with symbols that the public does not have. Those three things together are not just marketing. They are a power structure.
If a server is shown in the official tab, gets better access to players, and may receive better technical information, outside servers are no longer competing on the same field. They must market themselves harder, update with less help, reverse engineer more, carry more risk, and then when they shut down, people can say the market chose the winner. But that is not a natural market. It is a designed one.
What bothers me is that unofficial servers and software helped Bedrock build community life long before the official ecosystem looked this polished. Before Marketplace became smooth, before Featured Servers became a powerful distribution gate, people joined servers by IP, downloaded plugins, watched tutorials, joined forums, opened small servers for friends, became staff, built lobbies, configured ranks, and created thousands of reasons to stay with the game. When the ecosystem became large enough to monetize cleanly, the part that helped it grow was pushed toward the edge.
Dragonfly survives through a narrower path
In the Bedrock Gophers conversation I read, Seb, who was directly involved in Dragonfly’s data work, once said that “dragonfly relies on pocketmines existance”. When asked after PMMP’s end-of-support announcement whether Dragonfly still relied on PMMP for data generation or protocol-related updates, the discussion turned to debug servers with symbols. Ambient, another participant in the exchange, said updates were easier with debug servers that had symbols, even if those servers were private. Seb answered that debug servers with symbols were the only way Dragonfly had been kept alive. Fortunately, Dragonfly now mainly needs BDS data for block upgrade schemas, and those have barely needed updates recently.
This should be read carefully. It does not publicly prove that Galaxite provides private BDS files to Dragonfly. It also does not prove that every Featured Server gets the same technical resources. But it does show a practical reality. A project close to a partner ecosystem, with someone who can access private builds, or a project that only needs a smaller slice of BDS data, has a better chance of surviving than an independent project that must update many layers of protocol and data.
Dragonfly is Bedrock server software written in Go, built around async design and often used as a library. The df-mc organization also has datagen, a tool for generating useful data from BDS, and mapping, a BDS mod heavily based around pmmp/mapping. This reinforces the larger fact. Whether the server is written in Go or PHP, independent Bedrock software still needs data from the BDS world.
Older history makes that dependency more concrete. On 01.10.2025, Seb said Dragonfly was waiting on PocketMine to update BedrockData. The next day, he said datagen could not be updated because BedrockData lacked the state meta map. df-mc/datagen describes itself as a tool that connects to BDS through gophertunnel to generate data needed for Dragonfly updates, and it requires canonical_block_states.nbt plus block_state_meta_map.json from BedrockData or bds-mod-mapping. In short, Dragonfly may need less data than PMMP, but it still lives in the same Bedrock data economy.
The striking detail is that Dragonfly once had a plan to remove that waiting point. In April 2024, Seb said he intended to finish tooling that generated data for both Dragonfly and PMMP, then take over BedrockData maintenance after Dylan retired because he had access to appropriate builds for finding symbols. The plan was sensible and showed real succession intent. Eighteen months later, Dragonfly was still waiting on PMMP’s BedrockData. This is not proof that Seb broke a promise. It is evidence that one person’s plan, even with access and skill, is very different from a durable succession process with time, ownership, and replacements.
Recent Go-stack PRs make the same point in another way. In Dragonfly PR #1315, the author cites PocketMine’s changelog as evidence that UDP 19133 is the IPv6 port used for Bedrock LAN discovery. In gophertunnel PR #468, the fix for modern self-signed login claims says current wire evidence lines up with PrismarineJS and PMMP, then verifies a real client joining PocketMine-MP 5.44.3 through the Oomph proxy. This is not proof that Dragonfly depends on PMMP as a runtime package. It is proof that PMMP still functions as a practical map of Bedrock behavior for other stacks.
Dragonfly may be able to keep going because it needs less data and appears to be in a better position. PMMP had to carry a large plugin API, a legacy ecosystem, a long history, many data types, and a huge pile of expectations. In other words, this is not only about which programming language is better. It is about who has data, who has people, who has access, who has funding, and who still has enough energy to continue.
Geyser makes the board look even more uneven
Another important piece is Geyser. Geyser does not try to reimplement a Bedrock server from scratch. It translates protocol so Bedrock players can join Java servers. That model has a huge advantage because gameplay and server logic sit on Minecraft Java, where the server software and modding ecosystem is much older and stronger.
An important misunderstanding needs correcting here. Geyser does not use PocketMine-MP as a dependency or backend. On 28.11.2024, Auri, who carries the Geyser Developer role in the project Discord, answered directly that “Geyser does not make use of PocketMine”. Kastle, another participant in Geyser’s community discussion, added that Geyser uses Cloudburst’s Java protocol library while PMMP has BedrockProtocol in PHP, making direct reuse unnatural. The BedrockData issue therefore does not prove a runtime dependency. It shows that people around Geyser still noticed the disappearance of a data source and an independent implementation in the wider ecosystem.
The more favorable architecture does not make Geyser exempt from the maintenance tax. On 27.11.2024, chris, who carries a Geyser role in the project Discord, said support for 1.20.80 and several other releases would be dropped when a new Bedrock version arrived because of “complexity related to maintaining different versions with different behaviours” and because new features were unavailable on old Bedrock versions. That is a reasonable scope decision. It also shows that “alive and updating well” does not mean a project can absorb an unlimited number of behavioral branches left behind by Bedrock.
PrismarineJS shows another branch of the same cross-dependency network. bedrock-protocol issue #756 for Minecraft 26.33 tracks third-party patch streams from PocketMine, gophertunnel, and CloudburstMC. PR #743 fixed self-signed login because without the audience claim it could not log into PocketMine-MP through OpenID. In minecraft-data #1027, a contributor said new block palettes could come from GeyserMC’s mappings generator, BedrockBlockPaletteArchive, or Endstone DevTools. This is not a clean dependency tree. It is a mesh of projects watching each other to understand what Bedrock is actually doing.
Extremeheat, an important bedrock-protocol maintainer and former PMMP contributor, described that practice directly. When a format changes, look at the diffs in other third-party implementations such as PocketMine to infer what changed. In 2021, he also named PocketMine, gophertunnel, and Cloudburst as three projects tracking the protocol. What disappears with PMMP is therefore more than a PHP backend. The ecosystem loses an independent implementation for comparing diffs, testing real clients, and cross-checking errors in the implementations that remain.
I am not saying Geyser is bad. It is an impressive project. But its success exposes a sad contradiction. If you want to build a strong server experience for Bedrock players, a practical route may be to run Java server software and let Bedrock players in through a translation layer. That says a lot about the clean-room Bedrock server ecosystem. Going around Bedrock through Java can be more realistic than building directly on Bedrock.
Dktapps summarized PMMP’s ending as three pressures landing together. Bedrock servers became less popular, Mojang made maintainers’ lives harder, and clean-room servers were squeezed by projects such as Geyser. That framing matters because it refuses a single-villain explanation. The market changed, the technology changed, and platform decisions pushed in the same direction.
The nearest practical evidence is Lifeboat itself. On 10.07.2026, dktapps said he had worked with Lifeboat’s new software for a while and could confirm that “working with the java server makes it vastly easier to get stuff done”. Lifeboat once paid to move PMMP forward. When that same network finds a Java route substantially easier, this is no longer a forum argument about languages. It is an operational decision.
Java is not automatically better for every job, and Lifeboat’s new software is not public enough for a complete architectural audit. The largest advantage here is not merely JVM speed or native threading. The official Java server already contains Mojang’s gameplay, sits beside much better documentation and wikis, has a larger plugin ecosystem and a wider developer pool, and can then add Bedrock support. PMMP worked in the opposite direction. It accepted the Bedrock client first and then carried the burden of recreating the world behind it.
That is unfortunate. Bedrock is the edition many people play on phones, consoles, and Windows. It should have strong independent server software, clear docs, stable APIs, and public data good enough for the community to build correctly. Instead, software like PMMP must reverse engineer. Servers like NetherGames build internal systems and still say the update race became unwinnable. Other projects need partner proximity or reduced dependency scope.
Why are other Bedrock projects still alive?
This question deserves a direct answer. If Allay, PowerNukkitX, Endstone, LeviLamina, Geyser, and Dragonfly can still update, then PMMP’s ending may look like a problem belonging only to PMMP. But software that accepts Bedrock players is not necessarily doing the same job, and those projects do not pay the same maintenance bill.
Endstone is the clearest example. It has full vanilla compatibility not because volunteers reimplemented redstone, mob AI, world generation, and all of Minecraft in Python or C++. Endstone runs alongside the official BDS. Mojang’s code still provides vanilla behavior while Endstone injects a plugin API around it. It therefore inherits features PMMP had to reimplement or leave missing.
The bill does not disappear. It changes from reimplementing Minecraft into attaching to a changing closed binary. The Endstone port to BDS 1.26.20 contained 57 commits. The public work shows symbols being dumped from a Windows PDB, byte patterns scanned on Linux, and fixes across virtual tables, hooks, signatures, and application binary interface (ABI) changes. A changelog saying Endstone no longer needs debug symbols at runtime does not mean symbols disappeared from the porting process. The repository shows PDB data and binary patterns still involved before a build reaches users.
Endstone also preserves a rare timeline of what replaced public PDB access. After a meeting with Mojang on 15.07.2024, Vincent, the project’s primary maintainer on GitHub, said the PDB would not return publicly, but might be provided privately to selected community projects. The next month, he explained that Levi would receive temporary PDB access and transform it into one-way symbol lookup data that could not be converted back into a PDB. This was better than complete darkness, but it did not restore the old model. Knowledge that had existed in a public artifact became a derivative service operated by a small number of parties.
By BDS 1.21.60, even the workaround broke. Endstone’s February 2025 notice said more aggressive compiler optimization made bindiff ineffective, forced the team to reconstruct symbols manually, and inlined functions in ways that made old hooks impossible. In 2026, Endstone added a symbol dumper that can consume PDB data and fall back to signature scanning, while LeviLamina uses versioned bedrock-runtime-data. The ecosystem has adapted impressively. The adaptation also reveals that shared infrastructure is being kept alive by specialized pipelines very few people fully understand.
Endstone says that lineage out loud in its own Discord. Vincent wrote “without Nukkit, there probably would not be Endstone today”, placing Nukkit in a line inspired by PocketMine. That matters because it is not praise from an outsider. It is a BDS-native project acknowledging that its family tree runs through PMMP. Endstone v0.11.6 added support for BDS 1.26.33 very soon after PMMP’s public ending, while Endweave was described as a layer that lets newer clients join older servers while Endstone catches up. Even the BDS-native route has to treat update lag as an operational fact.
LeviLamina, formerly LiteLoaderBDS, belongs to the same architectural family. It is also a BDS mod loader, and its build configuration packages a SymbolProvider and BedrockData by version. This turns reverse engineering knowledge into repeatable tooling. It is not a free route. Instead of implementing world generation and mob AI, maintainers track BDS internals, ABI changes, symbols, and hooks.
BDSX is the counterexample that makes this distinction clearer. Its repository describes the project through the formula BDS + Node.js. It inherited the full vanilla server, then let plugins intercept packets, hook functions, and call into low-level BDS internals. Those are nearly the same broad advantages that Endstone and LeviLamina have. Yet BDSX stopped receiving updates in August 2024, and its owner archived the repository on 14.11.2024.
On 23.09.2024, SacriPudding, a member with the Plugin Dev role in the BDSX Discord, explained that the project could not move beyond 1.21.20 because Microsoft had removed the debug-symbol file used to locate functions inside the binary. Without it, “we can't as easily hook functions inside of the game”, and “BDSX as is right now is on hiatus for the foreseeable future.” SacriPudding does not carry a maintainer role, so this is a technical explanation from a plugin developer in the community, not an official statement from the repository owner.
The explanation nevertheless matches the public artifacts. The BDSX organization has bdsx-dev, described as a PDB Parser for BDSX, while bdsx-core contains pdbcachegen. Less than two months after that Discord message, the main repository was archived. This sequence does not prove every mod loader became impossible without a PDB. SacriPudding herself noted that other efforts to keep mod loaders alive existed. It proves something narrower and more troubling. A previously maintainable architecture can enter indefinite hiatus when its function map disappears and the project lacks enough people or specialized infrastructure to reconstruct it.
Endstone and LeviLamina surviving therefore does not disprove the BDSX story. It makes that story heavier. Their survival shows that alternatives exist through signature scanning, derived symbol lookups, versioned runtime data, and a large amount of expert labor. BDSX shows what happens when a project in the same architectural family cannot build or sustain those replacements in time.
One LeviMC comment states the frustration very plainly. Pops wrote that PocketMine, Endstone, Nukkit, and Levi all had limitations, all suffered in some way from “not being immediately updated”, and ended with “It sucks to not belong to the club of the big boys”. This is community feeling, not technical documentation. But it captures the lived version of the maintenance tax. Most users do not care about ABI or symbols. They see that their server waits longer than the group with a gate inside the system.
Allay and PowerNukkitX sit closer to PMMP because they reimplement much more of the server. They still do not begin from nothing. Allay’s README thanks Endstone because Endstone DevTools are essential for dumping key BDS data, thanks the protocol library Allay uses, df-mc, gophertunnel, and pmmp for useful information. Allay’s Discord also says Allay does not handle Bedrock network protocol updates itself and uses CloudburstMC’s library. AllayMC/BedrockData describes itself as data generated from BDS and says Allay has used Endstone DevTools since 1.21.1.03 to generate the data it needs.
PowerNukkitX also had a clearer relationship with PMMP data than a README line can show. In December 2023, someone in PNX Discord said the 1.20.50 protocol update might be delayed because the pmmp/BedrockData file used by PNX had not been updated. A GitHub issue bridged into Discord just before that proposed using AllayMC/BedrockData to reduce version-update difficulty because the data PNX used would become more controllable. That is why PNX and Allay should not be treated as simple counterexamples where “other projects survive, therefore PMMP was just weak.” They survive through inheritance, shared libraries, data dumps, BDS tooling, and different scope choices.
This dependency did not remain static. In late 2023, a PNX contributor said he would update his branch as soon as PMMP BedrockData published the new data. By July 2025, he described how PNX generated its own palette while other Bedrock data came from Cloudburst Nukkit. A month later, most states were updated automatically by script. PNX did not become self-sufficient. It moved from direct dependence on one upstream toward coordinating several sources and automating the parts it could own.
Allay shows succession happening through specific people too. On 13.07.2026, with the main maintainers less active, Zernix, an Allay contributor, said he had never performed a version update alone and had to trace how Endstone DevTools exported BDS data. A few hours later, he had the latest Endstone data working and continued the resource, item, block, and entity update. This is a hopeful result. Knowledge can be transferred. It also shows that bus factor does not disappear because a repository still receives commits. One day, “the project is alive” may depend on whether one contributor has enough patience to learn the process before the previous people leave completely.
Altay is the most direct response to PMMP’s ending. Altay’s GitHub says plainly that it is a fork of PocketMine-MP. In Discord, xrookiefight, the person announcing Altay’s technical roadmap, said that Altay would keep API 5 stable with protocol updates until June 2027 while developing API 6. A later 5.43.3 release targeted Bedrock 1.26.30. But the same general channel also had the line “The protocol is ready but the bedrock data isn't ready yet”. Altay is hope, not magic. It may keep one PMMP branch alive longer, but it still pays the protocol-and-data bill that exhausted the old upstream.
Geyser delegates gameplay to a Java server. Dragonfly narrows its scope and is often used as a library for custom servers. Endstone, LeviLamina, and BDSX delegate vanilla to BDS, but BDSX stopped where the other two built replacement pipelines. Allay and PowerNukkitX reimplement more while sharing protocol, data, and tooling. Cloudburst still receives commits but has no public releases. One word, “alive,” hides at least five different questions. Does a project track protocol, ship usable releases, run in production, provide vanilla behavior, and have a succession plan when its main maintainer leaves?
The active projects do not disprove Bedrock’s pressure. They show how different architectures avoid, transfer, or pay the maintenance tax. Some push gameplay into BDS or Java. Some reduce scope. Some inherit years of shared code and libraries. Some have a company or large network paying developers. Some simply still have a few unusually persistent people.
PMMP stopped when no one on the team remained willing to perform version updates. That does not prove PHP is impossible, and it does not mean every independent Bedrock server must die on the same day. It proves that a shared upstream survives only while people with rare expertise and enough time keep paying the entire bill. When the last person puts it down, private forks such as Zeqa’s may continue for their owners, but the public path thousands of small servers once shared has closed.
PMMP had problems of its own
If I blamed everything on Mojang, this article would be easier to write, but less fair. PMMP had its own internal problems too.
PHP is not the most natural choice for high-performance game server software. PMMP did many clever things to live with PHP, but as time went on, Dylan also agreed with the idea that much effort was spent working around PHP’s limits instead of developing PocketMine itself. Major API changes broke old plugins. Being non-vanilla confused many server owners. Old community plugins were often low quality. Some owners used leaked, copied, or badly shared plugins, then blamed PMMP when the server broke.
But PMMP’s own farewell did not say PHP killed the project. The published immediate cause was simpler and more frustrating. No one on the team remained willing to update for new Minecraft versions. PHP made that problem harder by shrinking the pool of people who understand the language, game servers, protocol, binary data, and performance at the required depth. It was not an automatic death sentence, as PMMP’s fourteen-year life already proved.
PMMP also stopped being a small playground where every idea could be merged. A project used by many servers has to become more conservative. That makes it harder for newcomers to contribute. Reviews slow down. Maintainers are limited. The door into core narrows. Update knowledge concentrates in very few people. All of these are real problems, and PMMP carries responsibility for not building a wider succession path.
But those problems do not erase the platform pressure from Bedrock. A project with technical debt can still survive when the platform is stable, documentation is good, and the needed data is public. A tired maintainer base can still transfer knowledge when the update process does not require an extremely specialized reverse engineering layer. PMMP had weak points, while Bedrock turned each one into interest accumulating on the bill.
In other words, PMMP had pre-existing conditions. Mojang did not create all of them. But the Bedrock technical environment made them much harder to survive and made a successor much harder to find.
Why I call this a way of killing
Killing software does not always mean sending a legal order to shut it down. There are slower methods. Do not publish enough documentation. Remove debug symbols from public builds. Let protocol change constantly. Provide necessary data only to some partners. Avoid a stable official plugin API that could replace the need for reverse engineering. Build distribution channels where partner servers reach players far more easily. Leave non-partner servers harder to update, harder to compete, and harder to staff.
Each individual decision can be explained by a legitimate reason. Anti-cheat. Child safety. Security. Avoiding leaks. Reducing support burden. Standardizing the experience. But together, they create a maintenance tax. Every Bedrock project pays it, while each architecture has different ways to avoid, transfer, or absorb it.
A project does not need a secret partner rescue to survive. Some delegate vanilla to BDS. Some delegate gameplay to Java. Some narrow their purpose to custom games. Some share protocol libraries and data dumpers. Some have companies paying developers, large networks keeping private forks, or maintainers who still have the energy. The problem is that small projects and community upstreams usually have the fewest ways to transfer the bill while sharing their results with the most people.
This is the strongest thing I want to criticize. If Mojang truly cannot publish BDS builds with symbols for security reasons, they can still publish the generated data. They can publish official schemas. They can provide detailed protocol changelogs. They can create a fair developer channel for independent server software. They can separate interoperability information from genuinely sensitive internals. They can say, no, we will not publish a debug binary, but here is the data so the community does not need to work blind.
The fact that these alternatives are not provided fully, while partners still appear to have access paths the public does not, makes the story difficult to read as pure security. It looks like a platform keeping power to itself. A platform can welcome outside creators while it is growing, then treat them as disposable once control matters more.
What if Mojang publishes the PDBs tomorrow?
If Mojang restores public PDB access, publishes the Bedrock data independent projects need, and writes documentation good enough to update without searching through binaries, I will call that good news. This article does not require Mojang to keep information closed forever in order to remain valid. It records what the community paid in time, maintainers, and abandoned projects while access was restricted or distributed unevenly.
A good policy arriving late would not restart PMMP, reopen NetherGames, or erase the years maintainers spent building workarounds. It could prevent the next name from appearing in an obituary. Useful criticism should not need the problem to survive forever so the critic can remain right. It should be glad when the cause is fixed.
If that change happens, the responsible response is to update this article, record the date of the policy change, link the new documentation, and separate the historical record from the new reality. The story would not become “the community exaggerated.” It would become “a door that had narrowed was finally reopened.” That improvement should be credited without erasing the cost paid before it.
Information wants to be free in Minecraft Bedrock
Stewart Brand’s line should not be read naively. Information is not free just because it wants to be. Companies hold it. Laws hold it. NDAs hold it. Commercial incentives hold it. Security fears hold it. But the line reminds us that information creates value when it flows. In software, this is especially obvious. Open documentation creates developers. Clear protocol creates interoperability. Public tooling creates forks. Public schemas create servers. Open knowledge bases create the next generation.
PocketMine-MP is a living example of that. It survived through open code, open plugins, forums, Poggit, docs, examples, and the endless loop of one person helping another fix a broken server. NetherGames began with public plugins and built its own fork. Zeqa then took that NetherGames fork, customized it for production, and later continued with a core named ZeqaMine-PM. Dragonfly has open repositories, datagen tools, mapping work, and a community around it. This chain of inheritance did not come from secrecy. It came from people being able to look, learn, modify, break, and rebuild.
When information is locked away, the best people may still find a path. But beginners lose the door. Small servers lose the ability to rescue themselves. Plugin developers lose motivation. Maintainers spend time guessing things that should have been documented. Eventually, fewer people learn to code through Minecraft. That is not a small loss. To me, that is the loss of an educational path.
In Vietnam, this was never only software
Reading the Vietnamese Facebook posts and comment threads made this ending feel less abstract. I would not use those memories as a strict timeline, because many of them are being reconstructed after years and some details may be fuzzy. But emotionally, they are consistent. PMMP was where Vietnamese students opened ports, ran servers on phones or cheap VPS plans, translated plugins, edited configs, wrote their first plugins, and eventually found their way into Information Technology.
The comments also add a local layer that a global technical article can easily miss. The decline in Vietnam did not come from a single blow. It unfolded through the 1.2 update making older devices struggle while Xbox Live became another barrier for young players, through weaker YouTube discovery, seasonal servers, DDoS, leaked plugins, repeated Skyblock formulas, and then the PM3 to PM4 transition that left many plugins and small owners behind. People remembered Genisys Pro, ElyWing, Steadfast2, Allay, PM3, PM4, servers pushed to run on weak hardware, anti-Toolbox work, Pickaxe Level, FarmingIsland, dupes, crashes, coin bugs, data loss, and maps rebuilt after mistakes. It sounds messy because it was. It also sounds alive because that mess was how people learned.
The mentions of Miheisu, Bravocraft, and a few private projects show that the flame is not completely out. It is just small. Miheisu is described as a server choosing stability and simplicity, passing through earlier names such as MineSAO and PuckVN before returning in 2024. Some people still fix old plugins, keep private projects, share resource packs, or hope a new server will launch. But the feeling is no longer a busy public square. It is a small circle around the last warm coals, telling old stories while knowing the old atmosphere will not return unchanged.
So I am not treating those posts as proof of what Mojang did. They are not technical documentation or official sources. What they do show is how people lived through it. PMMP’s ending is not only a repository being archived. It is a sigh from people who learned code by repairing servers, grew up with awkward plugins, made mistakes, broke things, apologized, rebuilt, and eventually realized that part of their youth had become an archive.
Thank you, PocketMine-MP
I want to give PMMP its own farewell here, because criticizing Mojang should not turn PMMP into an abstract symbol and erase the real people behind it.
Thank you to Shoghi for creating PocketMine when Minecraft Pocket Edition was still a very different world. Thank you to Dylan for carrying PMMP for so long and leaving behind code, documentation, technical decisions, debates, and warnings that were not always pleasant but were often necessary. Thank you to the maintainers, reviewers, moderators, donors, plugin developers, bug fixers, documentation writers, and people who answered the same beginner questions again and again. Thank you to Lifeboat for funding development in the later years while allowing the work to remain public.
Thank you, PMMP, for showing many people that servers are not magic. Thank you for making PHP a doorway into Minecraft Bedrock. Thank you for Poggit. Thank you for DevTools. Thank you for plugin.yml, onEnable, event listeners, permissions, config files, tasks, forms, virions, red console text, and all the errors that made people angry enough to learn something.
Thank you for helping small servers exist. Some lived for a few months. Some opened during a school break and then disappeared. Some were full of bugs. Some sold ranks a little too confidently. Some had only a few friends online. But for the person running them, each one was a real world. A world with a spawn, rules, players, hope, and the feeling of making something that lived outside one machine.
The end of official PMMP support does not mean every fork dies immediately. People will try to continue. Large servers will keep private forks. Archives will remain. Plugins will still run on older versions. But the era of PMMP as a shared upstream for the independent Bedrock community has closed. I think that loss is worth mourning.
Goodbye, NetherGames
NetherGames also deserves to be seen as more than one server shutting down. It is proof that a server starting from public PocketMine plugins could grow, build its own systems, reach thousands of concurrent players, and become part of many people’s memories.
The closure announcement has a very human shape. Callum started the server at age 11, using hosting paid for by his parents as a birthday gift. The lobby got griefed because OP was given to the wrong person. Backups were not considered at the right moment. Then the server grew, changed names, merged, matured, went through the pandemic, built staff teams, created a web portal, developed games, and gathered players. That is community history, not a KPI chart.
The painful part is that a server like that still concluded it was in a battle it could not win. Not because it lacked love for Minecraft. Not because it failed to try. But because technical access and distribution favored selected servers. When a server says it tried to become Featured and was then ghosted, the story is no longer simply that whoever does better survives. It becomes that whoever gets the door opened has a better chance of living.
NetherGames said it would open source as many games as possible so people could host matches and keep playing what they loved. If that happens, it is a beautiful way to leave. Do not close everything in a box and bury it. Leave code, leave memory, leave a path for someone else. That is also the ending most faithful to “Information wants to be free”.
What remains should stay open
After deaths like this, the easiest reaction is exhaustion. The more useful reaction is preservation.
Preserve old plugins. Preserve docs. Preserve forums through archive services. Preserve tutorials. Preserve NetherGames code if they release it. Preserve the PMMP update process. Preserve schemas. Preserve datagen tools. Preserve explanations in Vietnamese and English so the next person does not have to start from dead links and scattered messages. Not because everything will come back the way it was, but because technical memory is infrastructure too.
If PMMP forks want to continue, they will need more realism than nostalgia. They need smaller scopes. They need documented update processes. They need to separate public data from anything that cannot be public. They need to avoid depending on one person. They need to say clearly which versions are supported, which things are not, and they should not sell a vanilla illusion they cannot maintain. They need a calm relationship with the truth that Bedrock is no longer an easy platform to work around.
But the larger responsibility still belongs to Mojang and Microsoft. They can do better. They can avoid publishing sensitive binaries while still publishing the data communities need. They can support interoperability while still protecting players. They can create a developer program for independent server software that does not require being selected as a commercial partner. They can admit that the outside ecosystem was not a parasite. It was part of why Minecraft lasted.
No one needs to sign a death certificate
An ecosystem can kill its outside edge without ever ordering a shutdown. Close a little more information each year. Tilt distribution a little more. Increase update costs a little more. Move more access behind partners. Let maintainers burn out. Let small servers fall. Then say it happened naturally.
But people who lived on the edge can see that it was not completely natural. PMMP did not die only because of PHP. NetherGames did not close only because a server stopped being trendy. Zeqa survived not because the pressure was absent, but because it had a private fork, internal developers, an acquisition, and a route into a Featured network. All three fit the same pattern. Bedrock is increasingly governed as a closed commercial platform, while the projects that helped it grow are forced to shut down, cut their scope, or secure partner access.
Minecraft Bedrock will continue. Marketplace will keep selling. Partner servers will stay busy. Realms will still have users. Official things will not vanish because PMMP ended support or NetherGames shut down. But a rougher, more creative, more community-shaped part of Bedrock has lost a lot.
And to me, the most frustrating part is that this loss could have been avoided. Mojang did not need to open source the whole game. They did not need to publish debug binaries if there are real risks. They only needed to treat independent server communities as something worth supporting with decent documentation, data, and APIs, instead of leaving them to survive through unofficial relationships and partner goodwill.
Information wants to be free. Minecraft Bedrock grew because so much information was passed around, opened, reversed, rewritten, uploaded to forums, GitHub, YouTube, Poggit, and tiny servers running on weak hardware. When that information gets locked down, software is not the only thing that dies. A path into programming becomes narrower.
So farewell, PocketMine-MP. Farewell, NetherGames. Thank you for making Minecraft Bedrock bigger than its official menu.
Sources worth reading
PocketMine-MP’s farewell page announces that the project will no longer update for new Minecraft versions and thanks Lifeboat and the community.
The PocketMine-MP repository shows the archive state on 09.07.2026, the end of support announcement, and release 5.44.3.
Dktapps’ “I'm retiring” issue explains why PMMP’s future was already uncertain in 2024.
PMMP’s protocol update documentation describes the BDS data process and the removal of debugging symbols from public BDS builds beginning with 1.21.40.
pmmp/BedrockData, Mojang/bedrock-protocol-docs, and Geyser mappings-generator issue #98 show the boundary between the protocol being more open and Bedrock data still needing separate maintenance.
Cloudburst Discord discussions about debugging symbols being unlikely to return, Alemiz’s technical account of the limits of protocol documentation and game assets, and SupremeMortal’s history of reverse engineering with vanilla server symbols show that the data problem had been visible for years from the Nukkit and Cloudburst side.
Geyser Discord discussions preserve the community’s frustration over symbol removal in 2020, the role symbols played in naming functions, confirmation that Geyser does not use PocketMine, and the reason Geyser drops some old Bedrock versions. Together they show an independent branch that still has to control scope to pay its compatibility costs.
The 2019 PMMP forum post about debug symbols shows that this warning has been around for years.
NetherGames’ closure announcement, the original Discord sections about its PocketMine history and the Featured Server playing field, plus a staff account of internal issues that could not be made public, form a multi-causal record rather than a single neat explanation.
Minecraft servers and the Minecraft Partner Program show the official Bedrock server, partner, and Marketplace systems.
Dragonfly, df-mc datagen, and df-mc mapping are related pieces showing how independent Bedrock software still needs data from BDS.
Dragonfly PR #1315 and gophertunnel PR #468 show PMMP documentation, changelogs, and behavior still being used as technical comparison points by other stacks.
Bedrock Gophers Discord discussions about Dragonfly relying on PocketMine’s existence, debug servers with symbols, waiting on PocketMine to update BedrockData, and datagen being blocked by a missing state meta map should be read as operational traces, not formal statements for every Go-stack project.
The official Zeqa Network GitHub, its NG-PM fork, its ZeqaMine-PM fork, and the OnlineTime plugin are public traces confirming the relationship between Zeqa, NetherGames, and PocketMine-MP.
The Zeqa Discord links about PMMP for Season 10, who handled protocol and version updates, and whether Zeqa would update PMMP itself add community evidence. They should be read as operational traces, not formal statements about the whole production backend.
The WaterdogPE documentation explains why a public endpoint reveals only the proxy, while live status for zeqa.net records that proxy layer in use.
WaterdogPE issue #428, PR #420, and Discord discussions about self-signed auth payloads and auth types between PMMP and BDS show how a proxy has to bridge different downstream assumptions.
The InPvP acquisition announcement, the Season 10 announcement, and the Zeqa store show the network moving from outside the Featured list into Mineville Zeqa.
Aeris Discord announcements about Aeris becoming part of InPvP after the Zeqa acquisition, redirecting Aeris joins to Zeqa and transferring ranks, and moving Aeris Discord activity to Zeqa add evidence for Bedrock PvP brand concentration.
Endstone, its BDS 1.26.20 port, Vincent’s account of private PDB access and Levi’s one-way lookup data, plus bedrock-runtime-data, show how BDS mod loaders pay at the symbol, hook, ABI, data-pipeline, and update-lag layer.
The archived BDSX repository, its PDB parser, pdbcachegen in bdsx-core, and the BDSX Discord explanation of the hiatus after symbols disappeared form a consistent evidence chain. The Discord message should be treated as a plugin developer’s account, while the archive status and the project’s use of PDB data are anchored by public repositories.
Altay and its Discord announcements about keeping API 5 until June 2027, release 5.43.3, and the BedrockData gap show that a PMMP fork may continue, but it does not escape the data bill.
PrismarineJS bedrock-protocol #756, PR #743, and minecraft-data #1027 show Prismarine comparing PocketMine, gophertunnel, CloudburstMC, Geyser mappings, and Endstone DevTools while tracking Bedrock.
Allay, AllayMC/BedrockData, Allay Discord discussions about learning the update process from Endstone data, plus PowerNukkitX discussions about pmmp/BedrockData delaying updates and later generating palettes and other data through Cloudburst Nukkit, show how reimplemented servers still depend on protocol libraries, data dumps, and maintainers to different degrees.
Dktapps’ end-of-support announcement in the PMMP Discord is a direct source for the farewell. The Discord threads about Lifeboat’s sponsorship, partner feedback on symbols, burnout and accumulated demotivation, and Lifeboat’s new Java software add first-hand technical context. These links may require a Discord login and should be read as dktapps’ testimony, not announcements on behalf of Mojang, Lifeboat, NetherGames, or Zeqa.
You made it to the end If you need the short version
The article argues that Minecraft Bedrock is making independent servers and unofficial software harder to sustain by closing off technical information, shifting important access toward partners, and raising the cost of keeping up with updates.
PocketMine-MP ended official support and NetherGames shut down, but both point toward a wider problem. Independent Bedrock servers must keep up with protocol and BDS-data changes while operating in a distribution system increasingly tilted toward selected partners.
In this article, PocketMine-MP is the open-source server project and PMMP is its common abbreviation. PocketMine MP is a common spelling without the hyphen, while PocketMine can be a short name or refer to historical versions before PMMP. The PocketMine-MP history article explains the distinction in more detail.
No. According to dktapps, Lifeboat sponsored PMMP for years, made the PM4 release possible, and helped carry the project into PM5. It even offered to continue funding PMMP. Support ended because no one on the team with the required skills still wanted to update it for new Minecraft versions.
No. The article does not claim there was an internal order. It draws an inference from public events, technical documentation, and community experience to describe a pattern of pressure that slowly weakens independent software and servers.
The Vietnamese community section shows that PMMP was not only a software repository. It was an entry point into Information Technology for students, small server owners, plugin translators, and self-taught developers.
Public evidence shows that Zeqa ran and customized backends derived from PocketMine-MP at least from late 2025 into early 2026. Zeqa's official GitHub has an NG-PM fork from NetherGames and another PMMP fork renamed ZeqaMine-PM. However, the public address now sits behind WaterdogPE, so a server ping cannot identify every current backend.
- NhanAZ - 10.07.2026