How Silence Is Manufactured

This is a follow-up to “YSM Has Been Cracked, but Copyright Was Never the Point”.

After the crack went public, what you saw looked like an evenly matched standoff: the cracking group was loud and aggressive, while YSM’s supporters flooded comment sections in organized waves, and both sides had plenty of people and plenty of volume.

That picture is real, but it’s incomplete.

There were others hiding in the silence—people with the same information, the same technical ability, who chose not to show up.

I was one of them.

Last year, before any cracking tool was made public, I had independently completed a proof-of-concept for breaking YSM’s model encryption. I knew what it meant, and I knew what would happen if I released it. Before me, someone had reportedly figured out how to extract models from VRAM using tools like RenderDoc or Ninja Ripper—and then received a private message from the YSM developers demanding they stop and delete their work. I can’t verify the full details; that person never published anything. But the story circulated among independent researchers.

So I chose silence.

I was afraid of the other side—the asymmetry between cost and benefit was plain to see. On one side: the most prominent developer team in the Chinese Minecraft community, with a vast commercial network. On the other: a single independent researcher with no connections. I didn’t want to be the one who stuck their neck out, and I couldn’t be sure there wasn’t someone in my circles reporting back to them.

The fact that I was running this calculus at all is itself proof that the chilling effect was working.

The Structure of Silence

The core mechanism of a chilling effect is that it doesn’t need to target everyone.

It only needs the community to know that someone paid a price. After that, every researcher deciding whether to publish their work factors that possibility into the equation. The cost of pressuring one person buys the silence of many. I later learned that before the cracking group went public, at least three independent teams had been working on similar efforts. The outside world never saw most of them.

Crossing the threshold of silence requires a specific kind of will, or a specific kind of circumstance.

The group that ultimately chose to go loud does not represent all researchers on this issue. They represent the fraction that cared least about the consequences. The debate you see is always more extreme than reality: the voices that break through silence tend to be the most combative ones. The measured, nuanced voices never appeared at all.

Playing Dead

After the crack became public, the YSM developers circulated an internal assessment of the situation:

“These people are mostly middle- and high-schoolers with underdeveloped values… This is just lint on our sleeves—brush it off and it falls away.”

Their prediction framework went like this: there would be harassment, there would be aggressive rhetoric, “a grand battle of three hundred rounds, and then—nothing. They’ll disappear into the river of history.”

Their response strategy was to play dead. Wait it out.

What happened concurrently, outside their prediction framework: the cracking group’s videos on Bilibili were repeatedly reported and taken down; their GitHub accounts were reported until they were suspended.

The YSM developers didn’t orchestrate these attacks. They didn’t need to.

They had built a commercial ecosystem in which model creators’ income depended on YSM’s market position. This created a constituency with real financial stakes who would defend YSM without anyone giving orders—because they genuinely believed they were protecting their own livelihoods. The brilliance of the “encryption protects creators’ copyright” narrative is precisely that it frames the platform’s interests as the creators’ interests. I showed in my previous article that these are not the same thing, but that deconstruction requires context most people inside the ecosystem have never encountered—because the channels through which they get their information are the very channels this narrative dominates.

This isn’t conspiracy. It’s structure. Structure doesn’t need a mastermind; it only needs everyone to act in their own interest from the position they’re already in.

A Truncated Citation

Here is a concrete example of how this narrative is actively maintained.

On MC百科 (mcmod)—a major Chinese Minecraft mod wiki comparable to Feed The Beast Wiki—the Figura entry contained a section titled “Why Not Choose Figura.” It cited Figura’s official GitHub Discussion #351, arguing that because Figura is open-source and lacks encryption, “it is difficult to ensure that creators’ works will not be maliciously redistributed.”

That Discussion was opened by the wiki editor specifically to solicit an official statement from the Figura team—they explicitly wrote in their reply: “This discussion will be shown on mcmod, for part of introduction of this mod.”

The Figura maintainer’s full response explained that users can obfuscate and encrypt their own work, and that “obtaining the original files is not possible because they are converted to another format when uploading the model.”

The wiki entry kept “we currently do not have a way to prevent model theft effectively” but deleted the clause that immediately followed it.

What was deleted happened to be the qualifying statement that undermined the entire argument. An official statement solicited specifically for the wiki entry was truncated and repurposed as evidence against Figura.

Meanwhile, the YSM wiki entry said nothing about the fact that YSM’s retail format has never been tied to user identity. Anyone in possession of a YSM file could use it or resell it even before the crack—the concern about “preventing redistribution” applied equally to YSM at the time “Why Not Choose Figura” was written. Nobody put that in.

After the crack went public, the comment section on YSM’s mcmod page was closed and hidden. Closing a comment section requires platform administrator intervention—no group of zealous supporters can do that on their own. Someone made an active decision: this would not be a place where this was discussed.

Inside the Echo Chamber

An echo chamber isn’t built at the entrance.

Ordinary players encounter YSM almost always passively: it came pre-installed on their server, it showed up in a video recommendation, a friend told them about it. YSM was already a de facto monopoly. The monopoly preceded the bubble; the bubble’s function is to maintain the monopoly—to make people feel there’s no reason to learn about Figura.

Inside this ecosystem, your models are locked in a proprietary format you cannot leave; the explanation you hear is that encryption protects you; everyone around you uses the same tools and trades in the same marketplace. The debate that erupted after the crack was a footnote to all of this: supporters and opponents alike argued within the same set of premises. Nobody stepped back to question the premises themselves.

I think of a friend of mine—someone who personally knows several of the YSM developers. When I first encountered YSM last year and expressed my distaste for the monopoly, his position surprised me. His arguments were nearly identical to the highly homogeneous output you’d find in Bilibili comment sections, even though he couldn’t have coordinated with anyone. He was simply upstream in the information flow of the inner circle; his starting framework came from there. He sincerely believed every word he said.

This is probably the most effective form of narrative propagation: sincere people, unknowingly transmitting beliefs whose origins they have never questioned.

Why Now

I waited until OpenYSM went public—until the crack became an indelible, public fact—before picking up ysm2figura again, a project that was conceived at the same time as my original proof-of-concept.

I won’t pretend this is heroism. It’s a very pragmatic cost calculation; different priors, different solutions. I’m still the same person who chose not to stick my neck out last year. I still know what writing these words will invite.

Among those who stayed silent, many are better engineers than me, with clearer thinking. Their cost equations may simply be different from mine, or they may be waiting for a better moment.

What I want to document here is not what I did, or how brave I am, but the equation itself—the fact that it exists, that it operates, that it has kept certain voices from ever appearing. This is how narrative control works at its quietest and most effective: no suppression necessary, just make the cost of speaking high enough that most people do the math themselves and choose silence.

The YSM developers predicted this would disappear into the river of history. In their world, people who crack things only have two paths: either they’re too young, and time will correct them, or they’ll slide into the black market. There is no third possibility.

This essay is my answer to that prediction.

Early work on ysm2figura is underway. See: Yosemite Compiler.