Whether to shoulder the cost of censorship resistance for a mid-sized platform of only a few hundred thousand people is a question harder than any technical one. Rather than start with mechanisms, this piece of further reading follows three personal texts by the project lead, 豆泥, and turns the dilemma into terms the policy world can follow: a talk given in Buenos Aires, an essay about “the supermarket,” and a behind-the-scenes note on the neighborhood watch.

The site's introduction has already set up the proposition: for a platform that is censorship-resistant at its foundation yet must clear out spam every single day, censorship and censorship resistance are two uses of one and the same mechanism. That chapter dealt with mechanism and structure＊01-introduction（站台導論）Technically, suppressing spam and silencing dissent share the same mechanism — one score plus one threshold; in value terms, though, the two are opposed.. This page fills in another layer: when the operator stops drawing flowcharts and speaks in his own words instead, which metaphors does he reach for to convince himself, and others, that “this thankless job is still worth doing”? One thing up front: the three texts below are all personal essays or a talk transcript — they are the author's own argumentative voice, not the platform's official position. Wherever a point of view is quoted, it is attributed to the author himself.

From “market demand” to “survival demand”

In November 2025, 豆泥 gave a talk at the Ethereum Privacy Stack gathering in Buenos Aires, and its title was exactly this page's: is censorship resistance really indispensable? He put the question bluntly＊抗審查真的不可或缺嗎（演講稿）Today I want to raise a simple but uncomfortable question: if censorship resistance is so important, why is it so hard to build a durable product or organization around it?Source link ↗: is censorship resistance “really that important”? And if it is, why is building a lasting organization around it so hard? For the talk's opening slogan he chose “Privacy is not a crime,” stressing that for the Chinese-speaking community this is “something that has genuinely happened in the real world”＊抗審查真的不可或缺嗎（演講稿）My favorite slogan of the whole week was: 'Privacy is not a crime.' For us this is not just a slogan but something that has genuinely happened in the real world.Source link ↗ (the author's argument).

The first thing he takes up is whether “a real market demand exists.” The people who need censorship-resistant tools are those living at the margins: independent journalists, whistleblowers, grassroots organizers, human-rights lawyers, people inside authoritarian systems, or those touching highly sensitive issues within democratic societies. These people “do have a very strong need to protect themselves,” yet that demand is almost invisible in the market. The author states the gap very directly＊抗審查真的不可或缺嗎（演講稿）Marginalized communities have an enormous need for censorship resistance, morally and existentially, but the demand visible in the market is small, scattered, and often unable to pay.Source link ↗. For many of them, simply staying safe in the real world is already exhausting; to further expect them to become experts in Web3 A blanket term for a new generation of internet technology built on blockchains, emphasizing decentralization and users holding their own data. , key management, IPFS A network that breaks files up and copies them onto many computers around the world. As long as one still holds a copy, the content can be read — no single website required. , and privacy wallets is unrealistic. And the philanthropic funding, human-rights NGO A civil-society organization that pursues public benefit or advocacy rather than profit. , and democratic-government grants that have traditionally sustained such projects happen to be in a shrinking environment right now.

Here is a judgment that matters especially to policymakers: the supply side of censorship-resistant tools carries a “double risk.” Build a dating app and what you worry about is product–market fit; build a censorship-resistant tool and you also have to worry about prison, prosecution, and being watched＊抗審查真的不可或缺嗎（演講稿）If you build censorship-resistant tools for activists, on top of product-market fit you also have to worry about prison, prosecution, surveillance, and even being targeted by an authoritarian state or a powerful corporation.Source link ↗, with an adversary that may be a state-level actor. The cost of failure is compounded by the cost to personal safety, while what success can bring is “very limited.” This is exactly the first-person version of what the site's introduction calls “the builder's double risk.”

Following this thread, the talk offers its first answer — and the metaphor most often quoted from the whole piece. Censorship resistance is less like an ordinary market and more like an emergency service:

Perhaps censorship-resistance infrastructure will never look like an ordinary startup market. It is more like building fire escapes, seat belts, or public libraries. You don't ask, “How large is the total addressable market (TAM) for fire escapes?” You simply accept that, in a world where fires can happen, they are a necessity.

The metaphor is useful for policy because it splits the word “demand” into two layers. Measured with an ordinary market's ruler, censorship resistance never shows any scale: low purchasing power, a scattered population, and most mainstream users who “feel no pain until their own posts are deleted or their accounts banned.” But ask instead whether it is a piece of survival infrastructure, and the answer flips. At the end of the talk, the author brings the question back to where it started and answers it once more:

Finally, I want to return to the original question: does censorship resistance have market demand? My current answer is that it has a “survival demand.” People need it not for convenience, but because the alternative is silence, self-censorship, or losing one's voice in exile.

The single-word shift from “market demand” to “survival demand” is the most compact tool this talk hands policymakers. It makes two things clear at once. First, why the network effects of a censorship-resistant platform can rarely out-compete an ordinary platform's. Second, why the market alone will not hold this infrastructure up on its own — which is why public funding, philanthropy, ethical business, and community self-governance have to be pooled together to share the load. The author's ruler for success therefore looks beyond revenue, to “how many people can still speak, organize, and survive under pressure”＊抗審查真的不可或缺嗎（演講稿）We measure success not only by revenue, but by how many people can still speak, organize, and survive under pressure.Source link ↗.

The honeypot paradox: why censorship resistance must pay the price of governance

If censorship resistance were nothing but a good thing, this page wouldn't need writing. The sharpest passage of the same talk is what the author calls the “honeypot paradox”: when you build a censorship-resistant platform, it naturally draws people with sensitive things to say — political dissent, whistleblowing, taboo topics. That fits the mission exactly, but it also means you are concentrating risk.

You create a visible gathering point where a large volume of “dangerous” text concentrates in one place. Even with a more secure architecture, the concentrated attention is itself a source of danger.

This is no abstract worry. The author states plainly that there really are cases of “posts published on a censorship-resistant platform later used as court evidence”＊抗審查真的不可或缺嗎（演講稿）There are indeed cases where posts published on a censorship-resistant platform were later used as court evidence. So even if a platform protects the data, we cannot fully protect users from the consequences of their own speech.Source link ↗, so even when a platform protects the data, it cannot fully shield users from the consequences of what they said. The site's “Censorship resistance” chapter cites this too, but set in the context of a personal talk it reads more like an operator's self-warning: censorship resistance promises no safety; it merely swaps the risk of “being wiped out” for a different set of risks.

The other face of the honeypot paradox is spam. Because the platform “does not want to censor legitimate content,” it likewise attracts those who want to abuse it: scammers, bot farms, pornographic spam, coordinated disinformation campaigns. The author describes how “for a stretch of time” about 60% to 80% of new content was low-quality or malicious＊抗審查真的不可或缺嗎（演講稿）For a stretch of time we saw that roughly 60% to 80% of new content was low-quality or malicious: randomly generated spam, scam operations, or explicit material.Source link ↗ (the platform's own account, not independently verified by a third party). So the censorship-resistant platform falls into a contradiction it must resolve with its own hands:

This produces a painful contradiction: a “censorship-resistant” platform must, in the end, still actively filter and remove large amounts of content. You are fighting political censorship to defend freedom of expression, yet you still need some kind of content governance against abuse. Where is the line? Who decides the ban criteria?

Placed in a policy context, this passage is precisely the reason behind the site introduction's line that “the overwhelming majority of what the platform handles day to day is spam, while its reason for existing is that 1%.” The real meaning of the honeypot paradox is not “censorship resistance is dangerous, so don't do it.” What it warns is this: for a censorship-resistant platform, governance is not an option but a precondition for survival. Here, once again, censorship resistance and censorship become two sides of one thing. It is for this reason that the later behind-the-scenes note on the neighborhood watch writes about “sweeping up” as a kind of public duty.

The supermarket: when the public square is co-opted into shelving

The second essay, “Matters as I see it,” reaches for a completely different metaphor to talk about the same thing. Starting from his recent study of the rise and fall of various social platforms, the author offers an observation about “fate”: whether a social platform can escape the destiny of becoming a “supermarket.” He opens with a line that sounds like an aphorism but is really about how fragile platforms are:

All successful platforms are alike; each failed platform fails in its own way.

Right after comes the essay's core proposition. The author writes: “In a supermarket, if you are not the one paying, you are the product on the shelf. And the world right now is one giant supermarket.”＊我眼中的 Matters（隨筆）In a supermarket, if you are not the one paying, you are the product on the shelf. And the world right now is one giant supermarket.Source link ↗ Once a mainstream social platform latches onto network effects, the money-making loop cannot be stopped: advertising goes from side dish to main course, users go from consumers to product, and “the social platform is gradually left with only the platform, and no community” (the author's argument).

To break out of this frame, the author reckons there are “really only two ways.” One is to let users pay, turning them from products back into consumers so the community reclaims the platform; the other is to break the very frame of “the world is a supermarket”＊我眼中的 Matters（隨筆）The other is to break the frame of 'the world is a supermarket' itself: you are no longer merely a product or a consumer, and by reimagining the world you put an alternative operating model into practice.Source link ↗, to reimagine what the world looks like and then build out a different operating model — a cooperative-style supermarket, say, or a B Corp. Both paths set a high bar, and the author believes Matters' partners chose the second, harder road, “keeping an idealistic spirit alive in the hard mode of the capitalist world, and treading on thin ice as they search for a way out” (the author's argument).

This essay's contribution to the governance discussion is that it pulls censorship resistance back from a “technical choice” to “the power relationship between platform and community.” The author lists a string of Matters features that are “commercially unproven yet socially beneficial”: combining IPFS for censorship-resistant resilience, tipping built on a content economy for writers' autonomy, the Billboard that restructures ad revenue-sharing, and the Seven-Day Book that encourages everyone to pick up a pen — “none of these are the kind of experiments a commercial company in the conventional sense would encourage” (the author's argument). Censorship resistance here is not an isolated engineering feat; it is one technical consequence of the stance that “users are not shelf goods.”

The author is also honest about the cost of this road. He admits that Matters “has perhaps never been a supermarket aimed at maximizing profit, and may not even have done that part very well,” but immediately turns it into a reason for existing:

If there is even the slightest possibility of showing us that a social platform need not be a supermarket, then that, I think, is what Matters exists for.

The essay closes on an aphoristic note: “Long live matters.”＊我眼中的 Matters（隨筆）Long live matters.Source link ↗ The pun cuts two ways — it is the platform's name, and it is “may what matters endure.” For policymakers, this piece offers not a number but an angle on the problem: when we debate “why we need a censorship-resistant platform,” what we are really asking is “whether we are willing to preserve a kind of digital public space that does not treat people as commodities.”

The neighborhood watch: how a censorship-resistant platform does its own “sweeping up”

The first two pieces are about “why it's worth it”; the third, “Behind the Matters neighborhood watch,” is about “it's worth it, so how do we do it.” It lands the honeypot paradox's line that “governance is a precondition for survival” on one concrete little tool: a feature that lets citizen members “sweep up” (the author's term; technically it hides rather than deletes) pornographic and abusive comments, leaving a public record of every action.

This note supplies the original wording for the real-world ratio the site cites again and again. The author puts the density of spam very concretely:

I can't speak for other platforms, but on Matters, out of every 5 pieces of content, 4 are violating ads — 4 in violation! Not 1. … In short, in a place where 80% is junk, a platform that does no content filtering would probably just collapse.

These figures are the platform's own account, not independently verified by a third party, and should be marked as such when cited. They explain a design motive: the ops team had already removed more than thirty thousand ad items by hand, and the filtering algorithm introduced two years earlier “was cracked by dark magic”＊Matters 守望相助隊幕後（隨筆）The ops team has already manually deleted more than thirty thousand ad items, and the filtering algorithm we introduced two years ago has been cracked by dark magic — new forms of abusive advertising always manage to outdo the algorithm.Source link ↗, since new forms of abuse always slip past the algorithm — so the team hit on the idea of opening up part of the governance floor for readers and creators to work on together.

The neighborhood watch is by design tiered and deliberately conservative. The author spells out several self-imposed limits: to avoid collapsing right out of the gate, the scope is first confined to “the most urgently needed abusive-ad comments and pornographic-ad comments”; to keep members watching one another for reckless deletions, they built a public handling bulletin board where every removal (which the author refers to as “sweeping up,” though it technically hides rather than deletes) is logged as evidence for later review; and to avoid “secondary spread,” the ad content is even blurred out. The author openly says he worried about “jerks who go around deleting other people's posts”＊Matters 守望相助隊幕後（隨筆）For instance, I was very worried about jerks who go around deleting other people's posts, or ad content being too hard to classify, so we first confined the scope to the most urgently needed abusive-ad comments and pornographic-ad comments.Source link ↗. This caution about one's own power echoes exactly the site's concern with “due process.”

What needs careful handling is the boundary of this tool's “actual force.” As shown on the neighborhood watch's service page and related screenshots (community-watch.matters.town), the handling flow is drawn in four stages: a pornographic or spammed ad comment appears; a trusted citizen helps make the call; the comment is replaced and a public trace is left; and finally the commenter can request an admin re-review, with the platform retaining final authority over re-review, appeals, and permission adjustments. In the screenshot, that case's re-review status is marked “pending review.” This is a mechanism that is tiered by design and whose removal action still requires governance sign-off. So the more careful description is this: the neighborhood watch separates citizens' collaborative judgment from the platform's final decision, and does not claim that a citizen's report amounts to an immediate, final removal.

What most rewards a policymaker's attention in this note is how it defends the seemingly contradictory act of “a censorship-resistant platform doing censorship.” Borrowing a concept from the book 《困在社群平台》, the author first acknowledges the tension, then takes it apart:

Matters is at its core a censorship-resistant digital platform, and what it resists goes without saying. Yet introducing a collectively self-governed censorship (deletion) process on top of that censorship-resistant foundation sounds, at first, like Thoreau's positive conception of freedom and Berlin's negative conception colliding head-on like two trains — when in fact it is addressing problems at different levels.

“Addressing problems at different levels” is exactly what the whole site is trying to say, put once more in his own words: fighting state censorship (not being wiped out) and maintaining public quality (not being drowned in spam) are two uses on one and the same scoring line, not an either-or choice of sides. The author even pulls the stigmatized word “algorithm” back to neutral: he argues that an algorithm “does not necessarily mean being made to scroll helplessly downward” and can also serve to improve a good digital experience＊Matters 守望相助隊幕後（隨筆）The word 'algorithm' doesn't necessarily mean being made to scroll helplessly downward until you end up trapped in a social platform; it can also be about improving a good digital experience. In this era we cannot crudely demand that every platform strip out its algorithms.Source link ↗, and that “in this era we cannot crudely demand that every platform strip out its algorithms” (the author's argument). For any policy language that wants to draw a blanket “algorithm = evil” line, this is a pragmatic reminder.

Three metaphors, one argument

Put the three pieces together and you see they are really three angles on one argument. The fire escape is about “is it worth it.” The demand for censorship resistance is a survival demand; to measure it with the total addressable market (TAM) is to ask the wrong question in the first place. The supermarket is about “why this kind of platform.” Censorship resistance is just one technical consequence of the stance that “people are not shelf goods”; it shares the same set of values as the content economy and community autonomy. The honeypot and the neighborhood watch are about “what the cost is and how to bear it.” Censorship resistance concentrates risk into a honeypot, so governance (sweeping up) becomes a precondition for survival; and that governance power, in turn, has to rein itself in through public traces, tiered re-review, and a deliberately narrowed scope.

This also loops back to one direction from the end of the talk: rather than building “a big, beautiful platform,” build “invisible infrastructure,” refashioned into “composable components” that many small communities can reuse＊抗審查真的不可或缺嗎（演講稿）We should build composable components: censorship-resistant storage, privacy-preserving identity, safer funding rails, reputation systems, and interoperability protocols, so that different communities can assemble and use them in different ways.Source link ↗. The neighborhood watch is a living example of just such a “component you can take and rework”: a governance experiment a small platform built in a week, with public traces and tiered re-review, that in principle any mid-sized platform can borrow. The three metaphors finally settle on one pragmatic line: censorship resistance need not be a grand walled city; it can be a fire escape that every neighborhood can afford to install.

A note on citation and licensing

The three texts cited on this page are all essays and a talk transcript that the author, 豆泥, published on Matters in a personal capacity, each marked with a CC BY-NC-ND 4.0 license; they are an argumentative voice, not the platform's official documents, and wherever a point of view is involved this page attributes it to the author himself. The three complement one another in register: the talk lays out the argumentative skeleton of “demand, contradiction, direction” more fully, while the two essays each contribute a more down-to-earth metaphor — “the supermarket” and “sweeping up.” The figures that appear in the text — 4/5, thirty thousand items, 60–80% — are all the platform's (author's) own account, not yet independently verified by a third party, and are marked as such in the prose and in the stat cards. The neighborhood watch's removal action is tiered by design and still requires the platform's re-review; this page describes it neutrally on that basis, and does not equate a citizen's report with a final removal.