Researcher responsibilities & SciBLIND limits of liability

What SciBLIND provides

• A study-builder and participant-runner that implement seven paradigms with published, auditable timing and scoring logic (see our /learn articles and open-source ranking/ELO/Bradley–Terry modules).
• Infrastructure engineered for research integrity: daily backups, structured logging, exception capture, end-to-end tests across every paradigm, and a regression-test harness that catches the kinds of bugs we have already fixed.
• Full transparency: every study gets an exportable methodology report, an RNG seed per session for deterministic replay, and a quality-flag column on every trial so anomalies are never hidden from the researcher.

What the researcher owns

Before you collect live data

You are the person responsible for confirming that your study behaves exactly the way your hypotheses require it to before a single real participant sees it. Concretely, that means:

1. Run at least one pilot session end-to-end.Click through every trial type, submit every response path, trigger the timeout path, and then open the resulting CSV / JSON export. Verify the columns, the RTs, the correctness flags, and the condition-assignment variables look the way you expect.
2. Verify in both light and dark mode, desktop and mobile if your study is open to both. A stimulus that renders correctly on a 27-inch monitor may not render correctly on a 5.4-inch iPhone SE.
3. Read the RT quality column on export. Flags like backgrounded, sub_physiological, or lapse_suspected are informational, never filtering. It is the researcher's call what to do with flagged rows.
4. If anything looks off, contact us before you go live. We would much rather spend a morning fixing your study with you than try to recover data afterwards.

Ethics and legal

• You are responsible for obtaining IRB / ethics committee approval (or your jurisdiction's equivalent) before recruiting participants.
• You are responsible for the consent text on your study page, for ensuring it matches what your protocol was approved for, and for honouring participant withdrawal requests within the timelines your approval set.
• You are responsible for participant compensation: MTurk / Prolific / Bakker completion codes are minted by SciBLIND, but the obligation to pay is yours.
• You are responsible for any data that identifies participants — names, emails, phone numbers, Telegram handles — that you collect through SciBLIND. We encrypt at rest and limit access to you and SciBLIND super-admins, but we cannot retroactively remove data that was collected in violation of your IRB approval.

What SciBLIND does not warrant

SciBLIND is offered as a platform, not as a custom-built instrument for your specific hypothesis. To be explicit:

• We do not guarantee that a given study configuration will capture the effect you are looking for. That is a methodology question only you can answer.
• We do not guarantee that the defaults we ship are the right defaults for every paradigm. We document them; you decide whether they fit your protocol.
• We do not accept liability for data loss caused by researcher error — for example, unpublishing a study mid-collection, or misconfiguring a between-subjects variable after participants have already been assigned. We log changes (ExperimentChangeLog) and warn before destructive actions, but the final action is yours.
• We do not accept liability for data loss caused by participant-side behaviour (closing the tab, losing internet, running the study in a browser that has disabled cookies or JavaScript storage). Where we can, we capture this as a session flag and surface it in exports so you can make an informed decision.
• Infrastructure-side data loss — e.g. a Hetzner-wide outage or a bug that corrupts a column the researcher relied on — is our responsibility. Our commitment is: we back up nightly, we have a public status page at scaientist.statuspage.io, and if something we shipped breaks your data we will tell you plainly and help you recover what is recoverable.

How to reach us before a live run

If you are about to go live with a study that matters — a thesis, a paper deadline, a funded project — email hello@scaientist.com with a link to the study and what you want to verify. We will click through it with you and flag anything that looks off before participants arrive.

This page supplements our main Terms of Use and Privacy Policy, which cover the ScAIentist® ecosystem as a whole. SciBLIND-specific responsibilities live here.