The Disclosure Ladder
Two things this table encodes that are worth stating plainly.
The template
The Disclosure Ladder
Five levels of showing your work, from nothing to the full trace. The design decision is not how much to show, it is which level is the default for which user in which moment, and what one action moves them up a rung. Showing everything is not transparency. It is a way of making the user responsible for checking work they did not ask to supervise.
The five levels
Level 0. The answer only. No reasoning, no sources, no trace. The output stands on its own.
Level 1. The answer, with its provenance. What it drew from. A cited document, a named data range, the records it looked at. One line, one action to open.
Level 2. The answer, with the shape of the reasoning. A short account of how it got there, in the user's language. Two or three steps, not a transcript.
Level 3. The full chain, on request. Every step, every retrieval, every intermediate decision, laid out and readable. Available in one action, never the default.
Level 4. The raw trace. Prompts, tool calls, tokens, timings, model versions. Built for debugging, not for reading.
Who gets which by default
| Reader | Default | One action away |
|---|---|---|
| First-time user | 1 | 2 |
| Routine user, low-stakes task | 0 or 1 | 2 |
| Routine user, high-stakes task | 1 | 3 |
| Reviewer or approver | 2 | 3 |
| Admin or auditor | 2 | 4 |
| Support and engineering | 3 | 4 |
Two things this table encodes that are worth stating plainly.
Level 0 is legitimate. A product that shows its reasoning for everything is exhausting, and the exhaustion is not neutral: it teaches people to stop reading, which is worse than never having shown them. Reserve it for outputs the user can check at a glance or that cost nothing when wrong.
Level 4 is never a default for anyone. It is a tool, not an interface.
When showing more costs trust
This is the part that surprises teams, so it goes in its own section.
When the reasoning reveals the method is thinner than the confident tone implied. A three-sentence answer delivered with authority, whose trace shows it looked at one document, reads worse than the answer alone. The right fix is not hiding the trace. It is fixing the tone.
When the volume implies rigor that is not there. Forty steps of visible reasoning feels thorough and is often just verbose. Users read length as effort, which is exactly the wrong signal to send.
When it exposes the user's own data back to them in a way they did not expect. "I read your last two hundred messages to answer this" may be true, accurate, and a serious mistake to say out loud in that moment.
When the reader cannot evaluate it. Showing a chain of reasoning to somebody with no way to judge it does not create trust, it creates anxiety with extra steps.
The test: does the disclosure change what a reasonable reader would do next? If not, it is decoration, and decoration in this position costs attention that the important disclosures need.
The ladder in the interface
Each rung needs a real place to live. Fill this in per surface.
| Level | Where it appears | What triggers moving up |
|---|---|---|
| 0 | ||
| 1 | ||
| 2 | ||
| 3 | ||
| 4 |
Rules that hold across products:
Moving up a rung is always one action, never two. Reasoning that takes three clicks to reach does not exist.
Moving up is never destructive. Opening the trace does not lose the answer, close the panel, or reset anything.
The rung is remembered per user, not per session. Someone who always opens the reasoning is telling you their default, and re-collapsing it every time is a small daily insult.
The escalation trigger
The most useful thing on this page. Set the conditions under which the product raises the default level on its own, without being asked.
Raise a rung when at least one is true:
- Confidence is below your stated floor.
- The action is in an irreversible class.
- The output contradicts something the user said or the system said earlier.
- The user reversed a similar output recently.
- Money, permissions, external communication, or deletion are involved.
Your triggers: _______
Write them down and put them in the eval set. A disclosure ladder that only moves when the user climbs it is a ladder nobody climbs on the day it matters.
The one thing to do this week
Pick the surface where your product currently shows the most reasoning. Ask three people who use it what they do with that information. If the answer is "nothing, I skip it," you are spending attention you will need later, and dropping it a rung is the highest-value edit available to you this week.
From "Disclosure Without Overwhelm", chapter 8 of The Design Operating Model. falkster.com/design/disclosure-without-overwhelm