The Boiler We Bought Twice
March 2023. Our Sanctuary spent €6,200 on a second-hand biomass boiler that a member's cousin swore was basically new. The Proposal passed on a Tuesday evening with fourteen consents and two stand-asides. Eleven weeks later the heat exchanger cracked. The installer, who was very kind about it, told us the unit had spent two winters in a damp barn outside Ghent. We got through the next five months on three oil-filled radiators and a great deal of jumpers.
That's a failed decision. Nobody lied. Nobody broke an agreement. The room followed its own process exactly and the outcome was still bad.
What happened afterwards did more damage than the boiler ever did. Jonas, who'd brought the proposal, apologized so many times it got uncomfortable for everyone in the room. Two members quietly stopped coming to Council. Six weeks later, during an unrelated Proposal about bike storage, somebody said "well, we all know how the last thing we bought turned out," and the room went cold for a solid minute.
The boiler cost us €6,200. The recovery cost us most of a year, and we did the recovery badly.
Why the Repair Costs More Than the Mistake
Your community will make a failed decision. Probably about something duller than a boiler. An insurance policy. A membership fee change. A builder who came recommended.
The failure itself is survivable. Money comes back. Most operational mistakes are, in hindsight, boring.
What's harder to survive is what a badly-handled failure teaches the room. Three lessons get learned, and all three are wrong.
Proposing is dangerous. When the proposer absorbs the full weight of a bad outcome, the next six months produce fewer proposals from fewer people. Your most careful members stop bringing things forward first, because they're the ones who felt the shame most keenly. Your least careful members don't notice and carry on. That's a catastrophic filter to install in your governance.
Consent starts to read as liability. Members begin treating the consent round as a register of who gets blamed later. They hedge. They stand aside on proposals they actually support, purely to keep their name off the risky line. Your record stops telling you what people think.
The past becomes a weapon. Once a failed decision turns into a rhetorical device, every future debate has a free trump card sitting in it. "Like the boiler" ends conversations without addressing anything.
Those three compound quietly. A community can absorb a €6,200 loss more or less indefinitely. It can't absorb a Council where nobody brings anything forward.
Name It Out Loud Within Thirty Days
The most common failure in the repair is silence.
Everybody knows. Nobody says. The decision sits in the middle of the community like a piece of furniture people walk around, and each week that passes makes it heavier to mention.
Put it on a Council agenda within thirty days of the outcome becoming clear. Fifteen minutes. Say the plain thing out loud: this decision didn't work, here's what it cost, we're going to look at how it happened, and we're going to do that next month rather than tonight.
That last part matters. Don't run the review in the same meeting where you first name the failure. People are still warm. The proposer is still braced. Announce it, schedule it, close the item, and let a few weeks pass.
The announcement itself does most of the work. It tells the room that the thing everybody's thinking about is allowed to be discussed, that a process exists, and that the process has a date on it. Half the corrosion of a failed decision comes from people privately rehearsing arguments for a conversation that never gets scheduled.
The Autopsy: Three Questions, No Names
When the review meeting comes, run it on the decision. Keep it off the decider.
That distinction sounds like a nicety. It's the whole thing. Aim the review at the process and you get a community that learns. Aim it at Jonas and you get a community that stops proposing.
Three questions, in order, forty-five minutes total.
What did we actually know at the time? Reconstruct it honestly from the record rather than from memory. Memory is contaminated by the outcome. The Proposal thread, the attached quotes, the messages in the week before the vote. In our case the honest answer was uncomfortable: we knew the boiler was eight years old, we knew nobody had inspected it, and we knew a member's family was involved in the sale. All three were visible in the thread.
Did we follow our own process? Usually yes. That's what makes these decisions painful. Occasionally you find a real gap: a proposal that skipped its two-week reading period, a purchase that exceeded a threshold requiring a Treasurer sign-off nobody remembered existed.
What single change would have caught it? One change. Not a new committee, not a six-page policy. Ours was a rule that any purchase over €2,000 requires an independent inspection before the vote, paid for by the community, and the inspection report gets attached to the Proposal. It costs about €150 and it would have saved us €6,200.
Write the answers down as a Scroll in the Library the same evening, while people still remember what they said. The Quiet Power of Documenting Decisions covers why that record earns its keep years later; a failure record earns it faster than any other kind.
Be Careful With the People Who Stood Aside
There were two stand-asides on the boiler Proposal. Kirsten was one of them. She had written, on the record, that she wanted an independent inspection first.
She was right. Everybody knew she was right. And for about three months, that made her socially radioactive.
This is the most delicate part of the whole recovery, and most communities handle it badly in one of two directions. Either the person who warned you gets frozen out for being right, or they get handed a permanent veto they never asked for, where their future concerns arrive pre-validated and the room stops thinking.
Both are bad. What worked for us was doing it once, explicitly, in public, and then stopping. At the review meeting, our facilitator read Kirsten's stand-aside aloud, noted that the concern was specific and correct, thanked her, and moved on to the next question. Ninety seconds. No lingering.
That's the entire ritual. Acknowledge it, then let her be an ordinary member again. If your record makes stand-asides visible in the first place, you get this for free. If it doesn't, you'll spend the review arguing about who warned whom, which is a bad use of an evening. Vetoes, Blocks, and Stand-Asides is the vocabulary underneath all of this.
Reversing Without Humiliating Anyone
Sometimes the review ends with a reversal. You undo the decision, sell the thing, cancel the contract, go back to the old arrangement.
Reversals are healthy. Communities that treat their decisions as permanent are communities that make fewer of them.
Two rules make a reversal land well.
First, the reversal gets proposed by somebody other than the original proposer, and ideally by somebody who consented the first time. Jonas offering to undo his own boiler would have read as penance. When Henrik brought the reversal instead, it read as ordinary governance.
Second, the reversal Proposal says explicitly what changed. Avoid "this was a mistake." Write the facts instead: "We bought the unit on the assumption it was serviceable. The inspection in June found the exchanger cracked and the warranty void. We propose selling it for parts and installing the heat pump quoted at €9,400." New information, new decision. That framing keeps the original vote from becoming evidence that the room is stupid.
The Sunset Clause Is the Cheapest Insurance You'll Buy
Most of this pain is avoidable upstream, and the fix takes one sentence.
Put an expiry date on decisions that carry real risk. "This arrangement runs until March 2027, at which point it lapses unless Council renews it." "We'll trial this guest fee for two seasons and review it in October."
A sunset clause changes the emotional weight of a vote enormously. Members consent more readily because they're agreeing to a trial rather than a permanent commitment. The review is scheduled in advance, so nobody has to be the person who raises it. And when something turns out badly, the decision expires on its own rather than requiring somebody to stand up and unmake it.
We now sunset roughly a third of our Proposals. Purchases over €2,000, any change to fees, any new role, and anything a member specifically asks to have time-boxed. Each Proposal in ARA's Council module carries a review date, and the ones with a sunset resurface on the Council agenda automatically when they're due. The half of governance that used to depend on somebody's memory now depends on a date field, which is a considerable improvement.
For the deeper pattern, if a failed decision has left actual bad feeling rather than merely a bad boiler, that's a conflict, and it needs the conflict tools instead. Conflict Is Not the Problem. Avoidance Is. is the piece for that.
Eighteen Months On
We sold the boiler for parts in August 2023 and recovered €900. The heat pump went in that October and has been faultless. The two members who stopped coming to Council both came back, though one of them took until spring.
Jonas is still with us. He proposed four things last year, including the heat pump, which he researched obsessively and which came in €300 under quote. He'd tell you the boiler taught him more about proposing than his previous nineteen proposals combined.
The thing I'd want you to take from all of this: a failed decision is normal. It's evidence that your community is deciding things, which is more than a lot of groups manage. What determines whether it strengthens or corrodes you has very little to do with the size of the mistake. It comes down to whether you name it on a schedule, review the process rather than the person, thank the people who warned you and then let them go back to being ordinary, and put an expiry date on the next risky thing you agree to.
We got the boiler wrong. Getting the recovery wrong was the expensive part, and that one was entirely optional.