The method · published in full

The method is public. The panels never are.

Everything below is how every panel runs. We publish it because trust in a sealed process can only come from the process itself being inspectable.

Several honest reads, not one echo

What reaches you is several honest reads — not one confident voice the rest echo. Groups contaminate. In open discussion, the first confident voice anchors the rest, unique knowledge goes unmentioned, and five people converge on one opinion while feeling more certain than any of them started. Decades of decision research say the fix is structural: collect every view independently, before anyone sees anyone else's — then reveal together. That is the seal. It is not theatre; it is the mechanism.

Vantages, not verdicts

What comes back is a view you can stand beside, not a verdict to accept or refuse. No one on your panel is asked "what should they do?" They are asked how it looks from where they stand, and what they saw when they stood there themselves: what they did, and what they didn't see coming. Lived account first, view second, stated confidence last — so a quiet "I lived this twice" is never outweighed by a fluent guess.

The panel, step by step

1 · Intake, and the conflict screen

A paced private conversation. Alongside the decision itself we map two lists: who must never learn this exists, and who has a stake in how it lands. The first is a hard rule. The second is flagged on every name we ever propose to you.

2 · Redaction that attacks itself

Names out is not enough — content identifies. Before anything is shared, we run inference against our own redaction and show you the residue: "a reader could still infer: software, forty people, northern Europe." You approve with eyes open, or we blur further. The first outside reader confirms they cannot tell who you are before the panel widens.

3 · The panel

Three to five people matched on one criterion above all: they have made your decision — recently enough that the texture is intact, long enough ago to have perspective. Each candidate comes with the reason they're proposed and any stake flagged. You strike anyone, for any reason, unexplained.

4 · The sealed window

Each seat answers alone, in writing, against a deadline. Nobody sees another's answer. Nobody — including you — sees who has or hasn't answered yet; silence is invisible here by design, because visible silence weaponises refusal.

5 · The unsealing

Before the seal opens you record, privately, where you currently stand and what would change your mind — ninety seconds that measurably reduce the pull toward hearing only what you already believe. Then the views open one at a time, farthest from yours first, each with its author's stated confidence. Before you weigh anything, one sentence: "if the farthest view is right, it's because…"

6 · Independence, measured

Five answers are not five opinions if three of them share one line of reasoning. We measure the spread and tell you plainly: "five people, three independent lines." When a panel converges early, we offer one more perspective from further away — a seat vouched in by someone who knows someone who has stood there.

7 · Decision and waypoints

You decide — the panel never votes. The decision is recorded in your words, look-backs are scheduled while gratitude is fresh, and months later the ninety-second check-ins return something each time: which risks materialised, what you said you'd watch for. The people who showed up learn, with your approval, what their words changed. That closed loop is the entire reward economy of the bench.

Where machines are allowed

Software runs the plumbing: intake structuring, redaction drafting, clustering, scheduling. Every machine output is a proposal a person approves. Machines never answer your decision, never score a human's contribution, and never appear first — if a model could answer your question, we give you that answer at the door and no panel is convened at all.

The standing rule beneath all of it: nobody on your panel is ever scored on whether your decision "worked out." On rare, high-stakes calls, outcomes reward luck. People are recognised for one thing only — changing how you saw it, confirmed by you, over time.