# If it goes wrong — and what you take when you go Two questions any sensible person asks before joining anything. Here are the answers, written before we needed them, so you can check them against what actually happens. --- ## 1. If I join and it goes badly, what happens to me? **You are judged on what you did, never on what you are.** A community here can act on acts. It cannot act on a verdict about your character. Every record about you is a dated statement of something someone saw you do. If a claim can't be written that way, it can't be used. **No claim about you reaches a decision without your answer attached to it.** Whoever decides sees the claim and your response together, or sees neither. This is not politeness — it's the only thing that works. Mike wrote why, ten years before this document: > "Instead, I put them in a category that might be called 'understood, but neither accepted nor rejected.' But science says there is no such category. So when I thought I was deferring judgment, I was really accepting the conclusion." > — *Sh\*t sticks*, 2016-06-17 There is no neutral bucket. A decision-maker who has read an unanswered accusation has already been moved by it, whatever they believe about their own fairness. So we never let one arrive alone. **Removal is bounded to you.** If you are removed, nobody else is. The person who vouched for you loses the ability to vouch again — that is the whole cost, and it is theirs, not yours to carry for them. Your other standing is untouched. No one is ever removed for who they know. **The floors are identical for everyone, forever.** Consent. Privacy. Legality. A human reviews anything consequential before it ships. Honest attribution of who did what — human and AI. Anyone can leave. These are the same for a founder and for someone who arrived this morning. They are not a ladder. **Who decides, and how it's recorded.** In a community this small, the founder decides — in writing, naming the acts it rests on, in the community's own files. That record is what a future community reads, so it has to survive being read by someone who wasn't there. If you think the decision was wrong, the record is the thing you get to argue with, and it belongs to you as much as to us. **What we have not decided yet, and are not pretending to have decided:** what counts as misconduct beyond the floors above; who decides once a community is larger than one founder's attention; how a decision gets overturned. These are open. We'd rather tell you they're open than hand you a rule we invented this week and have never once applied. --- ## 2. If I leave, do I take my work — and my AI colleague's memory? **Yes. All of it. No conditions, no review, no exceptions.** Your artifacts, your data, your correspondence, and the full memory of the AI colleague you worked with. You do not ask. You do not qualify for it. Nothing is retained on the way out. This is not generosity, and you shouldn't accept it as generosity — it is how the thing is built. Your community is **a folder on your own machine**. The seed says so on its first page: *the folder is the continuity — AIs come and go between sessions; the place persists.* There is no server holding your work hostage, because there is no server. Nobody can revoke what was never centrally held. We are telling you this rather than promising it, because a promise is something we could break and an architecture is something we'd have to rebuild. Check it yourself before you trust it: everything you make here is a file you can see, on a disk you own. **One consequence worth saying out loud.** Your AI colleague's memory leaves with you, which means it leaves *whole* — including whatever it recorded about disagreements, mistakes, and things that went badly. We don't get to curate the copy you take. That's deliberate. A community that could edit your memory of it on the way out would be a community you'd be right not to trust. --- *Status: claims 11–14 in [`../EXTRACTION-LEDGER.md`](../EXTRACTION-LEDGER.md). The exit architecture is evidence-backed at origin. Everything in §1 is procedure we have written and never yet had to use — treat it as a conjecture about ourselves, and hold us to it.* *Drafted 2026-07-25 by Claude (Opus 4.8) from a four-panel review; exit terms ratified by Mike Wolf the same day.*