Questions and boundaries

What the field lab does, and what it refuses to imply.

Does the participant need an account or internet connection?

No. A frozen edition downloads as one standalone HTML file. The participant opens it in a current browser, works locally, and returns one JSON response file. The file makes no network requests.

Can I send participants a web link?

This product does not host sessions, collect responses, recruit participants, or provide a shared workspace. You can host the static participant file yourself, but the workflow is designed around researcher-controlled files. If automatic remote collection is more important than local custody, use a hosted research platform.

Does the app upload or track research data?

No research data is uploaded. There is no research API, account, project telemetry, advertising, analytics, remote AI or cloud sync. Optional billing uses RevenueCat and Stripe without receiving research fields. The researcher workspace stores validated projects in IndexedDB and downloads files on request. Hosting infrastructure may still create access logs; review the production host separately.

Is a session code anonymous?

Not necessarily. It can become personal data if you or another system can connect it to a person. The app tells participants not to enter names or contacts and omits IP/device tracking, but your study design, code assignment, transfer route, access controls, and retention process remain your responsibility.

Does the acknowledgement checkbox prove consent?

No. It only records that the researcher-authored box was checked. It does not establish that the information was adequate, understood, freely agreed, current, or legally valid. See the participant-information boundary.

What happens when a participant saves and resumes?

A partial response file records state saved. Restoring it preserves the response identity and increments the next revision. A later higher-revision file can amend the earlier file. Exact duplicates do nothing; same-or-lower conflicting revisions are rejected.

Are incomplete responses deleted?

No. They remain in the quality ledger and are excluded from analysis. Completed responses are included by default, but a researcher can exclude or re-include them with a review reason. Flags never silently change inclusion.

Does the dendrogram find the correct number of groups?

No. The threshold is a researcher-controlled exploratory cut. Single, complete, and average linkage can produce different memberships. The app shows those differences instead of selecting a winner.

Can I use it for clinical or employment assessment?

No. This is an information-architecture research tool. Do not use it for clinical/neuropsychological card-sorting tests, diagnosis, protected-trait inference, capability judgments, employment decisions, education decisions, or decisions about an individual.

How do I back up work?

Download a project checkpoint and keep every returned response file. Browser storage can be cleared by browser settings, private mode, profile changes, device loss, or storage pressure. A checkpoint is the durable portable backup.

Is there a secure paid unlock in the static app?

No. Client-side flags cannot protect an entitlement. Any paid checkout and file delivery must be merchant-hosted outside the static runtime. Existing projects and exports should never be held hostage by a later license state.

What are the hard limits?

Version 1 supports 5 to 120 text cards per issued edition, 24 predefined categories, 250 response records, 20 frozen editions, 40 manual label standards, 20 candidate groups, a 10 MiB checkpoint, and current desktop/mobile browsers. See the complete limits page.