For supervisors, editors, publications, and administrators
Know what is proven. In one screen.
A verification page is a short list of exact claims, each one backed by a signature you can check yourself. No trust in the author, no trust in us. Read it in a minute, audit it in an afternoon, or let your systems read it through the API.
Exact claims only
The page says what it proves and what it does not.
Both lists are printed on every page, in the same words. The point is not to make fieldwork look good. The point is to make it checkable.
What the page proves
- Distinct real humans, each with a humanity score, attested to being in this meeting for the span shown.
- Each of them endorsed a specific commit of the record, and the page names it.
- Each witness statement was signed with the approval, and the human-involvement grade is recomputed from them.
- Everything was anchored in a published daily batch, timestamped into Bitcoin through OpenTimestamps, which you can verify independently.
What it does not claim
- That the transcript is accurate to what was said.
- That a provenance declaration is true, only that it was made and when.
- That participants were not paid or coerced.
- That a participant’s real-world identity matches claims made in the meeting.
Human involvement
A, B, or C. Computed, never asserted.
Every document carries a declaration of who made it, when, and how much AI was involved. Every participant who approves it states what they witnessed: they authored it, saw it made, agree it is representative, or were only told about it. The grade is the weakest of those statements, capped by the declaration. If a participant claims more than the declaration allows, the page flags the mismatch.
- A: human-made during the meeting, witnessed by the people in it
- B: representative of the meeting, AI-assisted or made afterwards
- C: made before the meeting, or known only second-hand
- Ungraded: no declaration, or a participant declined to state
History
Every version, every approval, in order.
A workspace has one append-only chain of commits. Each commit hashes its parent, its tree of documents, its author, and its time. A reader sees who approved which commit; a late correction shows as a newer commit that the earlier approvals do not cover. You can recompute every hash from the canonical bytes we publish.
- Package approvals name the commit and every document in it
- Exclusions are signed too, so a disagreement is on the record
- Diffs between commits show which segments changed
For ethics boards and audit
Check consistency without learning identities.
Audit readers see what they need and nothing more: how many participants, what scores, which attestations exist, rounded times, and whether the chain and batches are intact. No handles, no profiles, no transcripts. Consent records export in a form your board can file.
- Anonymized read role across a whole study
- Consent-record export per participant, without identities
- Independent anchoring receipts for every attestation and commit
For universities and newsrooms
One standard for every student and every desk.
An institutional license covers everyone. Supervisors get a uniform way to read fieldwork, administrators get an admin console with SSO and data residency, and students get the tool for free as part of the program.
Every seat covered
Annual licensing per department or institution. Students and staff sign in with your SSO and never see a bill.
Methodology reports from VerifyMyResearch
A study rolls up into a disclosure at the level each reader is entitled to. Committees read the full report; the public sees output metadata only.
Verification through your systems
Submission portals, editorial systems, and research-integrity platforms read verification state through the API and SDK. Plugging in is a one-day job.
Residency and retention
Choose where sealed data lives and how long. Proofs are immutable; personal data is erasable on request, and the page says when it was.
“I can tell fabricated fieldwork from real fieldwork in the time it takes to open a link.”
Questions reviewers ask
Can a researcher fake participants?
Not cheaply. Each participant is a distinct Cubid human with a humanity score, and the page shows the score. Collusion with real people is possible in any system; the design makes it expensive, visible, and signed.
Do I need an account to read a page?
No. Public and link-visible pages open without signing in. Pages that reveal more to viewers ask you to sign in with Cubid so the owner can grant you that level.
How do I verify a page independently?
Every attestation, commit, and anchoring receipt is published with the bytes that were signed. The open SDK recomputes the hashes and checks the signatures; the protocol spec describes every step.
What happens if a participant withdraws?
Their attestation becomes a tombstone: the signature is removed, the time it existed is kept, and the page says a participant withdrew. The remaining proofs stand.
Bring verification to your department or desk.
Tell us whether you review theses, run a desk, sit on a board, or administer a program, and we will set you up with early access and a pilot workspace to read.