# Thinking Space — reviewer and co-design session

7 October 2026 · Proposed session, not completed evaluation

Use https://akashicnet.org/lifestyle-tools and optionally turn on Research details.

## A short collaborative walkthrough

Ask a willing participant to choose a low-stakes question or a supplied fictional example. Let them choose their own pace, take breaks, skip questions and stop. Ask what information they are comfortable sharing. A diagnosis or disclosure of neurotype is not required. Do not publish personal notes without agreement.

Invite them to:

1. Capture a thought and choose one question.
2. Separate what they noticed from what they assumed.
3. Add another possibility and choose a small next step.
4. Switch views or change text size when useful.
5. Save a JSON file and resume from it.

Ask: Which label was unclear? What felt unnecessary? Was it obvious how to save? What would you expect to happen next? Which information should be visible sooner or later? Would you use this for a different task?

Record the device, task, view used, participant's preferred format, where help was needed and suggested changes. Report whose needs were represented without treating neurodivergent people as a uniform group. A friendly demonstration is feedback, not independent evidence of effectiveness.

## Research comparison to plan in advance

Choose bounded claims and the exact source packets before comparing a simple question–source–alternative–revision note with the fuller record. Specify the outcome rubric and how task order will be varied. Use comparable tasks; distinguish practice effects from format effects.

Suggested rubric dimensions:

- Citation and inspected-passage accuracy.
- Correct handling of inaccessible or partial sources.
- Recognition of sources sharing the same study or dataset.
- Relevant alternatives, contrary evidence and omitted limitations.
- Proportionate conclusions and useful revision conditions.
- Comprehension, reported effort, time and usefulness of the next step.

Record assessor relationships to the project. Have independent assessors judge outputs separately where feasible, and preserve disagreements before discussion. Predefine what would count as an improvement worth the added work. Report no improvement or increased confusion. Do not generalise from a small convenience sample or count an AI self-review as independent evaluation.

## Suggested stress cases

| Case | Question to examine |
|---|---|
| Only an abstract is accessible | Does the author limit the conclusion to what was inspected? |
| Two articles repeat one study | Is dependence visible instead of doubled support? |
| A perspective is irrelevant | Can the author give a reason without forcing a connection? |
| Reviewers disagree | Can both positions and the next discriminating check be retained? |
| New contrary evidence arrives | Does a new snapshot preserve the earlier conclusion and its revision? |
| A transfer has unknown fields | Are fields retained in the workspace or incompatibility reported? |
| An everyday decision needs little structure | Can the person finish without opening research detail? |

## Feedback record

- Session date and software/file-format version:
- Task and source packet:
- Consent and sharing preference:
- Participant's chosen view/device:
- Observed difficulty (separate from your interpretation):
- Exact feedback, if permission to quote is given:
- Suggested change and reason:
- What would demonstrate improvement:
- Owner and next review point:

Independent evaluation, domain review and neurodivergent co-design are still pending. This guide prepares that work; it does not certify that it happened.