Flagon: We build software in the open.

Most of a company's quality comes down to the quality of its decisions and how cheaply it can undo the bad ones. Here's how we try to make good ones without drowning in process.

Someone owns it

Every decision has an owner. Not a committee, not a vibe, but a specific person who's accountable for making the call and living with it.

The owner isn't a dictator. Their job is to gather the context, hear the objections, and then decide, including deciding against loud opinions when they think those opinions are wrong. A decision that requires everyone to agree isn't a decision, it's a hostage situation. We'd rather one thoughtful person be accountable than have a group diffuse responsibility until nobody is.

Who owns what usually follows who's closest to the work. The person doing the thing generally decides how it gets done. The bigger and more expensive a decision is to reverse, the more it makes sense to pull in the people it affects, and the more the reasoning needs to be written down where anyone can read it.

Disagree and commit

You're expected to argue. Bring your best objection, as hard as you can make it, before the call gets made. The time to fight a decision is while it's still being decided.

Once it's made, you commit, even if you lost. That's the deal. You don't get to relitigate it in side channels, drag your feet, or say "well, I told you so" when it wobbles. Disagree and commit only works if the commit part is real. It's also what makes the disagree part safe: people argue freely because they know a decision will actually get made and everyone will get behind it.

The disagree-and-commit deal

Argue like you're right, commit like you were wrong, and write down enough that the next person can tell which one you were.

Write down the why

When a decision matters, we write a short record: what we decided, what we were choosing between, and why we went the way we did. Lightweight, a few paragraphs, not a legal brief.

The why is the valuable part. Six months from now, someone will look at the outcome and think "that was obviously wrong." Maybe it was. But without the reasoning, they can't tell whether it was a mistake or a sane call given what we knew then, and they'll waste a week rediscovering a dead end we already walked down. The record is a gift to your future colleagues, one of whom is you.

One-way and two-way doors

The most useful question we ask about any decision: is this a one-way door or a two-way door?

A two-way door is reversible. If it's wrong, you walk back through and try again, cheaply. Most decisions are these, and they should be made fast, by the owner, without a big process. Deliberating for a week over something you could undo in an afternoon is its own kind of failure.

A one-way door is expensive or impossible to reverse: a public commitment, a data model everything depends on, an architectural bet. These deserve real care: more input, the reasoning written down, a night to sleep on it. The mistake we watch for is treating the two the same. Slow decisions on reversible things, and rushed decisions on irreversible ones, are both how you lose.

Bringing the community in

We build in the open, so a lot of our thinking happens in front of the community. We want that. Public reasoning gets stress-tested by people who don't share our blind spots.

But open isn't the same as decision-by-committee. We'll show the reasoning, ask hard questions, and genuinely change our minds when someone's right. What we won't do is put the direction of the company up for a vote, or pretend the loudest thread is a mandate. Input is a gift. The decision still has an owner, and that owner still has to answer for it.

When we're wrong

We will be wrong, regularly, in public. That's the cost of deciding fast and shipping real things, and it's a cost we'd rather pay than the alternative: being slow, vague, and wrong anyway with no track record to show for it.

When a call turns out badly, we say so, fix it, and update the record with what we learned. No spin, no quiet edit hoping nobody noticed. A company that can't admit a mistake can't correct one, and correcting them fast is most of what "good judgment" actually is over a long enough run. See why Flagon exists for why we're playing it that way in the first place.