after we build · part eight · the human side
Psychological safety is a chain requirement
The retrospective asks "which chain links broke?" The prediction check sometimes delivers the uncomfortable answer "wrong." The postmortem asks "which structural gap produced this?" None of these are possible in a team where people are afraid to name what went wrong — because naming what went wrong means admitting that someone's work was part of the gap.
Psychological safety is not a soft skill or a cultural aspiration. It is the precondition for honest chain operation.
- A team that cannot say "the brief was thin" will never fix a thin brief.
- A team that cannot say "the amigos session missed this case" will never add the prompt that prevents it.
- If people cannot name what broke, the system cannot improve.
Silence in a retrospective is not safety. It is system failure. The chain's learning loop runs on honesty. Without it, the loop produces feelings instead of corrections — and the same failures recur in different forms.
The discipline — trace to levels, never to people
"The scenario was missing" — not "you forgot to write the scenario."
"The observation was incomplete" — not "the PO didn't do their job."
The chain-aware root-cause labels exist precisely for this — they name the structural level, not the human who operated it.
When someone leaves — and when someone joins
When a team member leaves, what leaves with them is not their code — it is their understanding of the domain. The developer who built the billing integration understood why the idempotency key is structured the way it is.
That understanding lives in the ADR, the API contract, the test names, the commit messages — if the chain was kept intact. If it wasn't, the understanding leaves with the person, and the next developer reverse-engineers from the code.
This is the chain's most practical argument for its own existence. Every artifact in the chain is a hedge against the departure of the person who created it.
A well-written Epic description, a named scenario, a signal reading, an ADR — each one is a piece of understanding that survives its author. A team that keeps the chain intact has institutional memory. A team that doesn't has individual memory — and individual memory walks out the door.