Essays

Longform pieces on governance, autonomy, and the unglamorous work of keeping a digital community alive for a decade.

The Invisible Hand: How to Build Systems That Work Without Constant Oversight

How do you create digital infrastructure that doesn't collapse the moment you step away?

I didn't wake up one day and decide to build my own digital infrastructure. I was forced to learn it.

At first, I just wanted a stable place to host a marginalized community. Facebook, Tumblr, Blogspot — one by one, they all proved unreliable. Policies changed, accounts were suspended, and one thing became clear: if you don't own the space, you don't control the rules; if you rely on free services, your content exists at their mercy; if a platform can remove you overnight, your community isn't safe. I wasn't trying to make a statement about autonomy. I just wanted to survive. That instinct led from a basic WordPress site to deploying entire infrastructures with Docker, Nginx, Discourse, and automation pipelines — because I had to.

What no one tells you about self-hosting

Self-hosting is often romanticized as "owning your data." What no one mentions is the cost of autonomy: things break, and there's no IT department to call — you are IT. Every new feature is a potential failure point. Your time is no longer your own; you don't just build a system, you maintain it indefinitely.

This wasn't just about hosting a website. It was about learning how the internet actually works — and realizing most people never think about the infrastructure that holds everything together.

The true cost of autonomy

Self-hosting isn't about writing code — it's about knowing when to stop. Every system introduces a new decision: automate, or keep manual oversight to reduce risk? Optimize performance, or prioritize maintainability? Build redundancy, or accept controlled failure? I implemented 3-2-1 backup redundancy after a near-loss disaster, automated uptime alerting, and deliberately chose not to adopt heavier monitoring tools whose upkeep would have cost more than their value.

The hardest part of infrastructure isn't knowing what to build — it's knowing what to leave out.

Infrastructure as governance

People don't think about infrastructure unless it fails, and don't think about governance unless it affects them. In both cases, the best-designed systems are invisible; the worst failures are only obvious in hindsight; and the leaders who prevent disasters rarely get credit until they leave.

I've watched people lose years of work because they relied on platforms they didn't own, and communities collapse overnight because they never thought about sustainability. Governance, like infrastructure, isn't about being reactive — it's about designing a system where fewer problems arise in the first place.

From Communities to AI: What Product Leadership Taught Me About Digital Governance

What tech governance can learn from product strategy, team autonomy, and AI decision-making.

I wasn't hired to build AI products — AI wasn't even on the company's radar in early 2023. I was handed a directive with a countdown clock: turn a profit in a year, with no roadmap, no clear product vision, and no existing documentation. When there's no time for deep investigation, giving people too many options doesn't lead to innovation — it leads to paralysis. Instead of asking "what should we build?" I framed choices as two options with a clear trade-off: pick your headache.

Self-governed communities face the same trap. If every voice gets equal weight in every decision, nothing moves. The key isn't choosing the perfect answer — it's eliminating the worst choices first.

Autonomy vs. buy-in

Aligning teams with genuinely competing definitions of success meant anticipating objections before they were raised, giving teams ownership without letting chaos take over, and making small visible improvements first to buy time for deeper work. Communities want autonomy, but they also want someone to take responsibility when things go wrong — leadership isn't about making everyone happy, it's about making the right people slightly unhappy in ways they can live with.

Unpopular decisions, proven right later

I flagged team-wide inefficiencies that leadership ignored — the exact issues became operational bottlenecks months later. I insisted AI development needed rigorous data cleaning before training; when the product launched anyway on a phased timeline, model errors surfaced immediately, exactly where the shortcut had been taken. Governance failures aren't always visible immediately — but the things you refuse to address today become tomorrow's structural failures.

AI governance and community governance suffer for the same reason: when systems are too rigid, they break. When they're too flexible, they get exploited.

The perfect answer doesn't exist in either domain. The best answer is the one that moves things forward without breaking everything else — and the real problems rarely come from bad data or bad policy. They come from bad incentives, and people who refuse to acknowledge them.

University Dorm Collective Punishment Protest: When Rules Become Weapons

How a dormitory policy revealed the mechanics of institutional power, and why sometimes the best resistance is simply knowing the rules better than those who made them.

A university faculty meeting: nearly a hundred professors and administrators on one side, fewer than ten student representatives on the other. The Dean, without a vote or discussion: "I am going to pass this policy now." The policy: if any unauthorized person entered a dormitory room, all residents would be evicted — collective punishment designed to make students police each other.

The mechanics of control

Institutional power doesn't need force. It needs compliance. The dorm committee representatives chosen to represent student interests supported the policy the moment the Dean called on them — because committee members receive guaranteed housing privileges, and when your education depends on stable housing, principles become expensive. 99.5% of students opposed the policy, but those with the standing to speak in the room were those with something to lose. This is how institutions maintain control: not through direct oppression, but through a careful system of small privileges that make resistance too costly.

Finding the gap

While others brought prepared speeches, I brought a laptop — livestreaming audio to the student council president waiting outside, building a real-time strategy session the institution wasn't prepared to handle. It worked not because it was clever but because there were no explicit rules against it; by the time the room understood what was happening, the narrative was already outside their control.

The Dean's response — "You are the culprit destroying trust between school and students" — was hard to sit with. But the policy never passed, stopped in the end by a simple procedural rule: quorum count.

Resistance doesn't always mean direct confrontation. Sometimes it means understanding the rules better than those who made them, finding new channels when the traditional ones are blocked, and standing firm when everyone expects you to bend.

High School Uniform Blog Movement: Digital Activism & Early Governance Lessons

An early lesson in institutions, records, and why bureaucracies absorb rather than negotiate.

The rule: even on PE days, students at Taichung Girls' Senior High School couldn't wear sports uniforms to and from campus — full uniform required, gym clothes changed into and back out of on-site, and carried home. The stated reason was that sportswear "was inappropriate for young ladies," even though the redesigned gym shorts were already modest and the required uniform skirt exposed more leg than the sports pants students wanted to wear instead. Student representatives had debated it for years in formal meetings. Nothing had changed.

Building the movement

It was 2008 — no hashtags, no easy virality. I ran the campaign on blogs and early forums: documenting the rule's outdated logic, organizing an internal survey that found 99.8% student support for change, and encouraging students to write their own posts, linked together into a web of personal testimony. It spread beyond my immediate circle. The school noticed.

Containment, not negotiation

I was summoned to the principal's office and offered a seat in student governance meetings — a courteous trap. Accepting would have meant stepping into a bureaucratic process built to manage, pacify, and ultimately ignore. I turned it down. Days later, student council allies brought the same 99.8% survey into the official policy meeting and were shut down outright: "If you don't like the rules, you don't have to attend this school." It was never about logic. It was about power.

The policy didn't change while I was there. It changed years later, for a generation of students I never met, off a movement I never publicly took credit for starting.

Institutions count on students giving up because they'll eventually graduate. But digital records don't graduate. Stories don't expire. And once a movement begins, even if it doesn't win today, it doesn't disappear.

Lessons

Institutions don't fear arguments — they fear records that outlast individual students. Bureaucracies don't negotiate, they absorb: offering a seat isn't a concession, it's containment. And the most dangerous thing you can do to an institution is break the cycle of forgetting.

Designing Digital In-Depth Visibility: How UX, Editorial, & Advocacy Shape Online Power

How do you create digital visibility that isn't just about exposure, but about depth, meaning, and long-term influence?

Real power isn't about being seen — it's about controlling how you're seen. For nearly a decade I've built platforms that resist the algorithm-driven hunger for virality, prioritizing sustained engagement and narrative control instead. If visibility comes at the cost of autonomy, you don't own your platform — the platform owns you.

UX as depth, not speed

Most mainstream platforms optimize for fast engagement and quick dopamine. This design prioritized slow, intentional interaction — deliberate friction that filters out users looking for convenience rather than meaningful connection, even at the cost of some users finding it "hard to use." Language wasn't just localized but culturally aligned to reflect Taiwanese queer realities specifically, which users describe as feeling authentic in a way no other platform does.

Editorial as counter-narrative

Mainstream media favors clickable, digestible LGBTQ+ stories. Editorial work here went the other way: profiles that don't fit mainstream molds, and a recurring Community Observation series that turned inward to examine the unspoken power dynamics and conflicts within queer community itself, rather than only external oppression. A user-generated base of 344,000+ articles and 90M+ views let the community define and expand its own narrative rather than rely on top-down storytelling.

Being independent means fighting for every inch of visibility while knowing you could reach more people if you played by the rules. But real impact doesn't come from reach — it comes from depth.

The Cost of Independence: What I Wish I Knew Before Building a Self-Governed Platform

Challenges of autonomy, why rules without enforcement fail, and what worked and what didn't.

"You have to ban them." The message came from a long-time community member, furious at another member's post. On a self-hosted, privacy-first community where surveillance wasn't an option and users had a say in governance, there was no algorithmic overlord to make the decision, and no backroom of moderators marking violations in bulk. Just us, our rules, and the weight of the choice.

The freedom paradox

I built Lezismore thinking independence meant freedom from corporate platforms' problems — surveillance, arbitrary bans, algorithmic suppression. What I didn't expect was that freedom also meant responsibility. No surveillance is great, until someone exploits the anonymity. No centralized moderation is wonderful, until the community can't agree on what's unacceptable.

Case: the harassment dilemma

A user reported an unwanted but non-aggressive message from a male user as harassment, expecting an exclusively-women space the platform had never promised to be. Was it an individual boundary issue, a governance contradiction between an inclusive mission and a member's narrower expectation, or a failure of community norms? Banning him would have been easy. Easy isn't governance.

Case: the defamation dispute

A user described a negative sexual experience with another member; the person named filed a report claiming defamation and threatened legal action. Suddenly the question wasn't moderation but arbitration: was the post slander or opinion, was inaction an endorsement of harm or a commitment to free expression — questions that every major platform faces in some form, from Reddit's mob-driven reporting to Facebook's reluctance to arbitrate anything not legally required.

Governance is messy. There are no easy answers — if you want a better system, you have to build it. Self-regulation is possible but hard; it takes active participation, not just policy. Not making a decision is still a decision.

The Trust Trap: Anonymity vs. Security in Self-Governed Platforms

What happens when a fully anonymous community has to decide, with no proof, who to trust.

It started with a whisper: "Something's off about this user." In a fully anonymous, self-hosted community — no real names, no identity verification, no metadata tracking — a member was suspected of misrepresenting her gender to gain access to the space. The evidence was a vibe: unfamiliar with queer culture, odd phrasing, inconsistent details. No rule violation, no report of harm. Just suspicion, in a space built specifically so suspicion couldn't be resolved by demanding proof.

The moderation paradox

Require identity verification and you destroy the privacy the platform exists to protect. Allow full anonymity and you expose users to deception. Let the community self-police and you open the door to witch hunts. Every major platform lives some version of this: Discord's ban waves without accountability, Bluesky's disputes over who controls moderation tools, Twitch's hate raids exposing the limits of user-driven enforcement.

The resolution wasn't a rule but a practice: transparency tools letting users judge credibility from a visible interaction history over time, and community-driven education on patterns of deceptive behavior — framed as informative, not accusatory — rather than invasive verification. The suspicious user eventually deactivated before anyone confirmed anything either way. That's the trust trap: with no proof, everyone is both the accuser and the accused.

Case: the mass-reporting wave

An unpopular but rule-compliant post triggered a coordinated reporting campaign — a call-to-action in a group chat to mass-report a user until they were banned, dressed up as community consensus. The fix was to stop treating report volume as evidence: require specifics on what rule was broken and what harm was done, log controversial cases (anonymized) to refine future decisions, and flag coordinated, same-source reporting for review rather than automatic action.

Governance isn't about choosing between total freedom and strict enforcement. It's about understanding that the real danger isn't bad behavior — it's unchecked power, wherever it sits.

What every governance model gets partly right

From corporate platforms: governance at scale requires consistency. From self-hosted communities: top-down moderation isn't the only option — participatory governance can work. From the failures in both: no system is perfect, but transparency, adaptability, and accountability matter more than perfection.