Democracy Has a Latency Budget: Accounting for the Real Price of Public Decision-Making

There is a particular kind of silence that descends on an open-source issue tracker when a maintainer admits, in public, that they don't know the answer yet. Contributors read it as stalling. Managers evaluating the project read it as risk. Nobody reads it as what it usually is: a decision being made slowly, on purpose, in front of an audience.

At Asterix Project we have run our decision-making in the open for years — public RFCs, recorded maintainer syncs, a roadmap anyone can read without filing a ticket. We are not going to argue against any of that. But we are going to argue something less comfortable: transparency is a tax, it is paid in engineering hours, and most projects dramatically underestimate the bill.

The Invoice Nobody Files

Ask a maintainer how much time governance costs and you will get a shrug. Ask them to look at their calendar and the shrug turns into a wince. The cost doesn't show up as a single block labeled "governance." It shows up as a thousand small withdrawals scattered across the week.

Consider where the hours actually go in a typical month on a mid-sized project:

• Writing for readers, not for the record. An internal design note takes an hour. A public RFC that anticipates objections from strangers — legal, licensing, threat models, naming — takes three or four. The delta is not padding; it is the cost of writing for an audience you cannot see.
• Responding to good-faith disagreement. A well-run RFC thread on a contested proposal routinely draws fifty to a hundred substantive comments. Triaging, answering, and synthesizing them is real work, and it lands on the people least likely to have spare cycles.
• Meetings that must be legible. An open maintainer meeting is not just a meeting. It is a meeting plus notes, plus a recording decision, plus a written summary for anyone who couldn't attend across time zones. The meeting is thirty minutes. The legibility is another hour.
• Roadmap upkeep. A public roadmap that goes stale is worse than none, so it needs tending like a garden — statuses updated, dates corrected, promises quietly checked against reality.
• The apology tax. When a decision is made in public and then reversed, the reversal happens in public too, with a written rationale. That write-up is often longer than the original proposal.

Add it up conservatively and you are looking at ten to twenty percent of a core maintainer's time. On a project with three maintainers, that is most of a full-time engineer's worth of attention, donated.

Why the Tax Is Worth Paying — And Why Saying So Matters

None of this is an argument for moving decisions into a private Slack channel. Public deliberation is how a project earns the trust that makes adoption possible; a decision nobody watched being made is a decision everyone is free to distrust. The Apache-style meritocracy, RFC processes like Rust's, and the IETF's "rough consensus and running code" all survived because the overhead was treated as the product, not as friction around the product.

The failure mode is not transparency itself. It is unbudgeted transparency. When the cost is invisible, it gets paid out of the same hours that fund code review, and code review loses. Maintainers burn out quietly, then publicly, then not at all — because the project simply stops answering.

Budgeting the Open

If your project claims to be built in the open, treat that claim like any other infrastructure commitment:

1. Put a number on it. Estimate governance hours per release cycle the way you estimate feature work. Track it, even roughly, for one quarter. The result will surprise you.
2. Cap the surfaces. You do not need public deliberation on every decision. Define which classes of change require an RFC, which get a meeting agenda item, and which get a commit message. Not everything deserves a town hall.
3. Fund the scribes. The person who writes meeting summaries and synthesizes RFC threads is doing release-critical work. Treat them like it — with maintainer status, sponsorship, or at minimum, acknowledgment in the release notes.
4. Say no in public, kindly. A declined RFC with a clear rationale closes a thread. A declined RFC with silence breeds a follow-up issue, a ping, and eventually a resentment. The write-up is cheaper than the aftermath.

Openness is not free. It never was. The projects that last are not the ones that pretend the tax doesn't exist — they are the ones that budget for it, pay it deliberately, and say out loud what it buys. Trust, it turns out, has an invoice. Someone has to write it.