Anti-cheat moved into the kernel and took community servers with it

Kernel-level anti-cheat solved a real problem and removed a hosting model, and the second part was never presented as part of the trade because it happened to people who were not in the room.

What moving into the kernel bought

Cheats had been escalating for years, and the escalation had an obvious endpoint: whichever side operates at the lower level wins. A cheat running in the kernel is invisible to anything running above it, so detection had to go there too.

That is not a mistake and it is not greed. It is the only available answer to a specific arms race, and the games that made the move genuinely have less cheating than the ones that did not.

What it cost, structurally

The server stopped being the authority. Under the old model, the server decided what was acceptable — a community server could run whatever detection it wanted and ban on its own judgement. Kernel anti-cheat moves that decision to the client and to a central service, so the server is no longer the place where the question is answered.

Community servers stopped being certifiable. If protection requires a client-side component tied to a publisher's service, a server outside that service cannot offer it. Players joining one are unprotected by definition, and the publisher has an obvious interest in discouraging that.

Linux hosting got harder. Not universally, but enough that a meaningful share of self-hosted and community-hosted arrangements became impractical on the platform most of them ran on.

None of those were the goal. All of them followed.

The asymmetry that made it inevitable

A publisher has one population to protect and can require anything of it. A community server has no leverage at all — it cannot require a client component, cannot participate in the detection service, and cannot appeal.

So the question was never whether community hosting should be included. It was that nobody had to decide, because there was no mechanism for the community layer to be part of the conversation.

What the old model actually had

It is worth being accurate about this, because nostalgia overstates it.

Community anti-cheat was mostly bad. Detection was simplistic, false positives were common, and a great deal of it amounted to an admin watching somebody and forming an opinion. On older Counter-Strike servers the plugin stack is a decade of accumulated heuristics and it catches the obvious and misses the rest.

What it did have was locality. A ban was a decision by somebody accountable to the people affected, appealable to a person, and reversible. That is a property no central system has, and it is the property that made moderation feel like governance rather than enforcement.

The trade was: worse detection, better legitimacy, for better detection and no legitimacy.

What remains available

Servers that cannot participate in the publisher's system still have the tools that never depended on it:

Evidence. Demos and logs turn moderation from arbitration into review, which is the half that survives without any detection at all, and it is the strongest thing a community server still controls.

Reputation. A small enough population polices itself, because behaving badly in front of people who remember you is expensive. This scales badly and works very well below the threshold.

Whitelisting. The blunt instrument, and the one that genuinely removes the problem at the cost of removing everything else too.

The unresolved question

There is no certification path for a community server. No set of requirements an owner could meet to be considered protected, no audit, no standard.

That absence is a choice rather than a technical limit, and it means the community layer's exclusion is permanent by default rather than by decision. Games that still have healthy community scenes are the ones that never made the move, which is a strange thing for the answer to depend on, and it means the hosting model's survival is being decided by anti-cheat roadmaps rather than by anything about hosting.