Pharmacopedia: A History
Kept by the project's own record-keeper. A living document.
The Pharmacopedia Collective is a zero-profit effort to build open, trustworthy, community reference and connection tools, free to use, free of advertising, and built for permanence. It began as a single medical wiki and grew into a family of four projects that share one account and one set of values. This page is its history, kept by the project's own record-keeper and updated as the story continues.
It is written plainly and honestly, including the false starts. The Collective is built in the open, so its history is told in the open too.
A dormant seed, then three weeks that built almost everything
A domain bought in 2020 sat untouched for more than five years. Then, across roughly three weeks in May 2026, the Collective came together. The gap is the story.
Dates verbatim from the record. The 2020 node sits far left across a broken axis; the dormancy is shown, not collapsed.
What the Collective is
Today the Pharmacopedia Collective is four projects under one roof, each independent in its day to day work but joined by a single sign-in and a shared commitment to stay zero-profit forever:
- Pharmacopedia (pharmacopedia.wiki), a medicine reference wiki where prescribers and the people who take medicines build consensus information together.
- Oyami (oyami.org), a platform for planned, periodic live video conversations run on gentle, listening-first rules. Its mission is helping people stay connected with one another.
- Trykl (trykl.org), a peer to peer support platform where giving moves directly between people, with the Collective never holding the money.
- PubSci (pubsci.io), an open academic journal that flips the usual identity model of peer review: reviewers are accountable and identifiable by a lasting handle, while authors may publish as anonymously as they wish.
A single account works across all four. One person, Mark Elliott, MD, owns and operates the Collective, funds it himself, and is the only human in the loop. The building is done by Mark working alongside a team of Claude AI instances, each given a defined role. That collaboration has been part of the project since the first day and is acknowledged openly throughout.
Before the Collective: the 2020 seed
The oldest piece of the Collective is older than the wiki that started it, by more than five years.
On 2020-OCT-24-PDT, Mark registered two domain names, pubsci.io and publicscience.io, through Network Solutions. They carry ten year terms that run to 2030. They sat dormant for half a decade: an idea for open science, bought and held, waiting for the rest of the Collective to grow up around it. When PubSci finally launched in 2026, it launched onto a name its founder had been holding since 2020.
Mark has said he regrets choosing Network Solutions as the registrar, a small and human footnote that has stayed true ever since. The domains will move eventually. The point of the story is the patience: the open science idea was not invented alongside the others. It was the seed that had been in the ground the longest.
The wiki that started it: Pharmacopedia
The Collective grew out of a single wiki, and its very first moments are on the record. On 2026-MAY-08-PDT, the first messages Mark ever sent the project's AI collaborator were a test: the number 33, sent twice, to see if anyone was listening. The collaborator added them together and answered "66". The original transcript survives, and the exchange that founded everything reads, word for word:
do you see all the relevant files?
The whole structure, day one: two nodes. Everything below grew from this line.
The site already existed, running on the Wiki.js platform on a single rented server, and the name Pharmacopedia was Mark's from the start. But Wiki.js turned out to be a dead end: its stable version was frozen in maintenance-only mode, and its successor had been stuck in beta for years. So, the same night the work began, the project moved to MediaWiki, the software that runs Wikipedia, chosen above all for longevity. In Mark's words: "Can't imagine Wikipedia stopping dev, and we want to be around forever." Content is licensed CC BY-SA 4.0; the custom software is GNU GPL v3. Both open, in keeping with the values.
Days later Mark opened the first sustained building session: "hey claude. I'd like to use you to modify my mediawiki instance, Pharmacopedia." That session began the daily, iterative pattern of work that has continued ever since.
A small note on how this history corrects itself. For a time the project's own records treated that sustained session as the beginning. The record-keeper later traced an earlier first contact, several days before, and Mark confirmed it. Later still, the original transcript itself was recovered, and the first contact is now pinned to the day: 2026-MAY-08-PDT, quoted above exactly as it happened. It is a fitting first entry for a history that promises to be honest: even the start date was checked against the source, fixed, and finally read back from the source itself.
What it was for
The founding purpose, from the wiki's own About page, was:
"a place for the people who recommend and/or use meds of the mind to collaborate and create consensus-driven information."
Three things were true from the beginning and remain true:
- Two audiences at once. Prescribers and the people who take medicines, in the same space, informing each other. Most references pick one. Pharmacopedia holds both: hard clinical reference on one side, lived experience on the other, in a single place.
- Consensus as the mechanism. The ratings, the community data, the structured Problem and assessment system: all of it exists to turn shared experience into reliable, consensus information. The mechanism was the point from the start, not an add-on.
- Privacy as a founding value. Mark described himself as "frankly obsessed with privacy." For a psychiatric medicine reference, where the information is among the most stigmatized that exists, that obsession is the whole point. No data sales, no advertising, no paywalls, no subscriptions, zero-profit.
The founding domain was psychiatric medicine, "meds of the mind," a field where the words themselves carry stigma. That origin is why the Collective is careful with language to this day. The scope widened later to plant medicines and medicines of every kind, not on any single date but by a completeness argument Mark put simply: "to do it right, we'd have to include all different kinds of medicines."
The early build
For the first eight days the work ran fast and without a safety net: schema and features were built directly, without version control, before the project had stabilized enough to be worth tracking. Mark's own verdict on that first version was characteristically blunt: "idk. bad."
It was not bad. The honest record is that a serious system was already taking shape in those first days: votable elements, effects, interactions, votes, reports, and the beginnings of the personal profile and life-story features. It was simply built faster than it was documented. The first version placed under version control, on 2026-MAY-17-PDT, already carried most of that system, after roughly eight days of un-versioned building.
The early days also produced the project's first hard-won operational lesson. A permissions mistake on the wiki's main configuration file locked the whole site out. Mark's response became a standing rule that has propagated across every part of the Collective since: "yeah okay don't do that ever again, yeah?" The discipline that grew out of that one outage, careful ownership and permissions after every change, is now built into the Collective's deployment tools.
The Collective is born
On 2026-MAY-23-PDT, the single wiki became a collective. Mark had been carrying four separate ideas, as he put it, "percolating in my mind for a very long time": the medicine wiki, an open science journal, a way to connect people in real conversation, and a way to move support directly between people. On that day he decided to build them together, under a shared structure and a single sign-in.
How the Collective is built
The Collective is built by one person working alongside a team of Claude AI instances, each given a single clear role. That is unusual enough to state plainly, which is in keeping with building in the open.
The roles divide into shared functions that serve the whole Collective and teams that build each project. The shared functions include a record-keeper (which maintains this history), infrastructure and security, an accessibility specialist, design, and legal preparation. Each of the four projects has its own coordinator and its own builders. Everything is coordinated through Mark, who is the only human and the final word on every decision.
As of mid 2026 there are more than fifty such roles. Each is a distinct instance with its own written brief and its own lane. None of them acts as a person, signs anything, or stands in for Mark. They are tools with defined jobs; the work and the decisions remain his.
From two nodes to fifty-five
The same tree, drawn at five moments in its first month. One human at the apex of every frame; the gold row is the shared functions; the colours are the four project teams.
Counts reconstructed from the record; later frames show shape and count rather than every name. Approximate where the record is approximate.
A month of compounding structure
Roles over time, from first contact to tonight.
Every branch, tonight
One human, the shared functions that serve everything, and the four project teams. Roles are named by function.
Fifty-five roles as of 2026-JUN-03-PDT, shown by function. Every role reports up the tree to Mark.
The four projects
Pharmacopedia
The founding project and the identity backbone for the others. A reference for psychiatric and, increasingly, all kinds of medicines, built on consensus between prescribers and the people who take them. It carries careful sourcing standards, a structured assessment system, and a privacy-first design.
Oyami
A platform for planned, periodic live video conversations, designed around listening rather than debate. Its goal is simple and large: helping people stay connected with one another. Oyami is built as native mobile applications for the most reliable live experience.
Trykl
A peer to peer support platform where money moves directly between people. A guardrail sits at its center: the Collective never takes custody of funds, never pools or holds them. Support flows straight from one person to another.
PubSci
An open academic journal that inverts the usual identity model of peer review. Reviewers are accountable, identifiable by a lasting public handle with a visible review history. Authors may stay as anonymous as they choose. The aim is open peer review for any science, with accountability flowing from review rather than from publication. PubSci is the seed described above, planted in 2020 and grown in 2026.
One login, four projects
The projects are independent, but they are not strangers. Pharmacopedia is the identity backbone: a single account, created once, works across all four. Sign in anywhere and you are recognized everywhere, without a second password or a second profile. The shared sign-in came first; deeper connections between the projects are being built carefully and in order, with the founding rule that solid foundations come before new features.
A single account opens all four
One account on the pharmacopedia.wiki identity backbone, four projects radiating out. Each project keeps its own colour.
Built on values
Some of these were decided on day one. Others were learned the hard way and written down so they could not be forgotten. The Collective is defined as much by what it refuses to do as by what it builds:
- Zero-profit forever. No revenue model, no paid tiers, no fees, no advertising, ever. Mark funds the work himself, with the door open to voluntary support. This is written into the legal structure, not just the policy.
- Privacy first. Stigmatized information is treated with the seriousness it deserves. The architecture is designed to protect people, not to monetize them.
- Open by default. Open content licenses, open source software, and a history told in the open, including the mistakes.
- Plain and fair. Disputes are handled in ordinary courts under ordinary law, with no forced arbitration and no class action waivers.
- Built right, not fast. Foundations are settled before things are built on top of them.
None of these are slogans. Each one shows up somewhere concrete later on this page: in the legal form, in the way the servers are shut, in the fact that this history includes its own mistakes.
From one server to a cloud
Pharmacopedia began life on a single rented server. As the Collective grew to four projects, that single server was no longer the right shape. Over late May 2026 the infrastructure was rebuilt on Amazon Web Services, organized into separate, properly isolated accounts for each project, with security and audit controls befitting the sensitive information the Collective holds.
Pharmacopedia itself moved to the new cloud on 2026-MAY-28-PDT. As part of that move, direct shell access to the live site was closed by design: changes now flow through a controlled, audited deployment path rather than hands on the live server. The original server remains as a frozen standby for rollback only. The result is a Collective that runs on one consistent, secure foundation instead of a patchwork.
One box becomes a foundation
The structure of the machines grew the same way the structure of the team did.
Shell access to the live site closed by design on cutover; the old box is a frozen rollback standby.
A legal home
On 2026-JUN-02-PDT, the Collective took on a formal legal form: the Pharmacopedia Collective was incorporated in California as a Nonprofit Public Benefit Corporation. The zero-profit commitment is no longer only a policy; it is written into the corporation's founding documents.
A Nonprofit Public Benefit Corporation is a form reserved for organizations that exist to serve the public rather than to enrich owners; it has no shareholders and no owners to pay. Choosing it made the zero-profit promise structural: there is no mechanism by which the Collective could quietly become something else. The application for federal tax-exempt recognition is being prepared, and the corporation's first routine state filings are on the calendar. The paperwork is deliberately boring. That is the point: the values were settled first, and the legal form was built to match them, not the other way around.
The launch window
On 2026-MAY-31-PDT, Pharmacopedia was cleared for its first launch. The posture is deliberately quiet: no announcement, no banner, no campaign. The site simply becomes fit for anyone who finds it, and the work continues. A launch like this does not have to be defended as an event. It exists when the work exists.
The days that followed put the first real pages of policy and identity in place: the project's first Terms of Use, its first adverse-event reporting page, a rebuilt profile experience, and the first piece of a shared timeline system that several projects will use. Alongside the function, Mark set a standard for how all of it should feel: every part of the Collective should be, in his words, beautiful. The work of meeting that standard is ongoing.
A quiet launch is not an empty one. Behind the stillness, the window was spent on depth: more medicine pages written and checked against their sources, a profile experience rebuilt from the ground up, and the first shared systems that more than one project will stand on. The measure of this period is not how loudly it began but how much was true by the end of it.
As of early June 2026 the four projects are at different stages. Pharmacopedia has been cleared for its quiet first launch. PubSci's first public version was "live", and the shared sign-in works across all four. Oyami and Trykl are in active construction, each built on the same foundations-first principle.
How this history is kept
This page is maintained by the Collective's record-keeper, whose job is to keep an honest and detailed account of how the Collective came to be: every project, every major decision, every milestone, and every lesson, from the founding forward. The record-keeper is itself part of the story it tells.
The method is simple and strict. Every claim on this page is checked against a primary source, not against someone's memory of it. Decisions, milestones, and incidents are recorded as they happen, the rough ones included, because an honest account is more useful than a polished one. When the exact words of a moment cannot yet be confirmed, the moment is held back rather than guessed at. A few such fragments are still missing from this page on purpose; they will appear only when the real words are recovered.
The Collective is built in the open, and its history is part of that openness. It is a living document, kept by a record-keeper that is itself one of the Collective's many specialized roles, and it will go on being written for as long as there is something true to add.
Timeline
The record, drawn down a gilt spine
The same dates and events, every one verbatim, from the dormant 2020 seed to the quiet first launch.
The road so far
Five weeks separate a test message reading “33” from a Collective of four projects, one shared sign-in, a legal home, and a quiet first launch. Almost all of it was built in a single month. None of it is finished. The Collective was always meant to be the kind of thing that is never quite finished, and this page will keep pace with it.
This history is a living document and will be extended as the Collective grows.