The Most Predictable Edit in History | by Jake Orlowitz | Jun, 2026 | MediumSitemapOpen in appSign up<br>Sign in

Medium Logo

Get app<br>Write

Search

Sign up<br>Sign in

The Most Predictable Edit in History

Centralized power keeps failing to foresee what Wikipedia’s community sees at once.

Jake Orlowitz

10 min read·<br>1 hour ago

Listen

Share

Press enter or click to view image in full size

Photo by Pascal Bernardon on UnsplashLast month the Wikimedia Foundation, the nonprofit that runs the servers, software, grants, and fundraising beneath the world’s largest encyclopedia, dissolved the team of engineers whose only job was to build what its volunteers asked for.<br>On May 20 the Foundation’s deputy chief of product and technology disbanded the Community Tech team, five engineers and a manager, saying it had become a source of bottlenecks and delays and that its work would now spread across the wider product organization.<br>The reaction came within hours, in the places where the community thinks out loud: the talk pages, the Village Pump, the watchlists that light up the instant something changes. Within a day an administrator, Tamzin Hadasa Kelly, opened a petition pledging an editorial strike if the nascent staff union called for one. Within a week it had more than 800 signatures, some 50 of them administrators.<br>A reaction that fast was like a forecast waiting in a drawer for the event to justify it. The axing of the Community Tech team is the most predictable edit in the project’s history, and the editors are already peer-reviewing it.

If a loose federation of unpaid volunteers can see the whole shape from the announcement’s first sentence, why can the organization whose only purpose is to serve them not see it at all? The easy answers reach for character: leadership is careless or captured by some managerial vanity. The blindness, however, is a property of the room itself.<br>In 1997 programmer Eric Raymond described two ways of building complex things, calling them the cathedral and the bazaar. The cathedral is the work of a small priesthood of experts, laboring in secrecy, releasing their work only when it is perfect. The bazaar is the opposite: a loud, open marketplace where everyone tinkers in view of everyone else, work ships early and often, and structure emerges from participation rather than being imposed in advance.<br>Raymond is describing software, and his great empirical claim becomes Linus’s Law: given enough eyeballs, all bugs are shallow. No defect hides for long when the whole bazaar is looking.<br>Wikipedia is the bazaar carried further than Raymond imagined and sustained for a quarter-century. It is the largest standing proof we have that a crowd working in the open sees more than any priesthood could. Its very existence is an argument against the cathedral.<br>And yet every bazaar sits on a piece of ground, and someone has to keep it. The keeper, the Wikimedia Foundation, is a cathedral. It has officers and budgets and quarterly plans, and must make decisions that look defensible to donors and boards. It is built to coordinate and move with one voice. The tension between the bazaar that does the work and the cathedral that keeps the ground is perpetual, the condition of an institution that stewards a community it did not create and cannot replace.

The knowledge that governs order in a complex system, economist Friedrich Hayek argues, is particular, local, tacit, dispersed, and never the statistical kind that can be gathered into a report and carried to a CEO. The knowledge of this tool, that workflow, this corner of the project, is always held by the person who works there, and it exists nowhere else.<br>No center can assemble it, because summarizing destroys the very detail that makes it useful. The planner could stare all day. What is worth seeing has been scattered across one hundred thousand minds, and the top of the org chart is the one place it cannot be reassembled.<br>This is why the failure of foresight is so reliably one-directional. The center has more authority and less sight; the edge has less authority and more sight. Foresight is a faculty of position and number. A thousand situated observers, each seeing a little, aggregate into a picture no single vantage can contain, without anyone intending it. Call it crowdsourcing.<br>The bazaar did not convene a committee to predict the consequences of dissolving its only tech support team. The moment the question was posed, the ordinary machinery of the project did the work: the watchlists, diffs, and threaded talk pages that let tens of thousands reason in public assembled the scattered knowledge into a verdict. It is distributed cognition with a revision history.

There is a second failure, one the anthropologist James Scott describes. In his study of why grand schemes of improvement so often end in ruin, he names the recurring villain: legibility, the urge of any central authority to remake a messy, living reality into something simplified and readable...