Trust should be an architecture, not a disclaimer.
Most research asks you to trust the vendor. PolitiMind is built so the record itself carries the assurance: who qualified, what was asked, how it was computed, and whether anything changed after publication.
The problem
A methodology footnote is not an integrity guarantee
The standard assurance in this industry is a paragraph at the back of a deck. It tells you what the vendor says it did. It gives you no way to check, no record of what changed, and no way to distinguish a sound result from a confidently formatted one.
An institutional evaluator needs the opposite: a stack where each layer proves something specific, states what it does not prove, and leaves a record an outsider can examine.
The trust stack
Six layers, and exactly what each one proves
Integrity claims fail when they are stated too broadly. Each layer below is paired with its limit, because a layer that is oversold is worse than one that is absent.
Top of the stack — what a reader receives
- L06
Identity integrity
ProvesThat a participant is real, unique, and eligible for the study they answered.
Does not proveWho they are, to anyone outside the restricted identity layer.
- L05
Privacy separation
ProvesThat identity and response data are held apart by design, not by policy alone.
Does not proveThat we hold no identity data — we do, minimally, under access control and audit.
- L04
Response provenance
ProvesThat each response was collected under a specific frozen study version.
Does not proveThat any individual answer is sincere — no method establishes that.
- L03
Methodology controls
ProvesThat weighting and computation ran under a registered, versioned method.
Does not proveThat the method is the only defensible one; it means the one used is stated.
- L02
Audit history
ProvesThat sensitive access and consequential changes are recorded append-only.
Does not proveThat no error occurred — it means an error can be located and dated.
- L01
Tamper evidence
BetaProvesThat a published record can be shown to be unchanged since it was committed.
Does not proveThat the underlying opinion is correct, or that the result is beyond dispute.
Base of the stack — the verified participant. Each layer above depends on the ones below it.
The integrity lifecycle of a single study
- 01
Frozen study version
The study specification is versioned and locked before fielding begins.
Questions, order, audience definition, and method version are fixed at launch. Nothing that arrives later can quietly redefine what was asked.
- 02
Verified responses
Only qualified participants can submit, and each response binds to the frozen version.
The response record carries which study version it answered, so results can never be assembled from a mix of silently edited instruments.
- 03
Quality controls
Integrity review runs during and after fielding; flagged activity is examined, not silently dropped.
Review outcomes are recorded. We publish that controls exist and that they are auditable — the signals and thresholds themselves stay undisclosed so they cannot be gamed.
- 04
Processing under a registered method
Weighting and aggregation execute as a recorded run against a registered algorithm version.
The run, its inputs, and its method version are retained together, which is what makes a number reproducible rather than remembered.
- 05
Published result
The result publishes with its specification, sample context, and methodology attached.
Provenance travels with the deliverable instead of living in an internal system nobody outside can reach.
- 06
Integrity record
A commitment to the published record is anchored to a public ledger, making later change detectable.
Anchoring is currently testnet-stage: the record and verification path operate end to end, and production anchoring is gated on external cryptographic review before we describe it otherwise.
What anchoring is, precisely
Cryptographic anchoring proves nothing was changed — not that an opinion is true
Anchoring publishes a commitment to a result record on a public ledger. Later, anyone holding the published record can recompute that commitment and compare it to the anchored one. If they match, the record has not been altered since it was committed. If they do not, something changed and that is now visible.
That is the entire claim. It does not make an opinion correct, a sample representative, or a methodology sound — those rest on verification and method, which are separate layers of this stack. Anyone marketing a ledger as proof of truth is selling you a category error.
Anchoring and independent verification are currently testnet-stage. The encoding rule and verifier path are published so a technical reviewer can follow them; production anchoring waits on external cryptographic review.
What we claim, at the readiness we claim it
- Participation requires identity verification, so responses come from confirmed real people rather than anonymous traffic.
- Participant identity and response data are held separately, so buyers receive population-level intelligence and never participant identities.
- Published results carry provenance: the specification, method version, and computation run behind a number are recorded and inspectable.
- Methodology is stated in plain language — sampling frame, weighting approach, margin of error, and known limitations travel with every result.
- Every study runs behind integrity controls that detect duplicate, coordinated, and low-quality participation before results are published.
- BetaResult commitments are anchored to a public ledger so the record is tamper-evident after publication.
- BetaAn independent verifier can recompute a published commitment and check it against the anchored record.
Anchoring and independent verification are beta, testnet-stage capabilities; production anchoring is gated on external cryptographic review, and no result is third-party certified today. We publish the concepts, the encoding rule, and the verification path — not hashing internals, security topology, keys, contract assumptions, fraud logic, or audit schemas. Withholding those narrows attack surface; it is a deliberate boundary, not a gap.
What backs this up
Published results carry provenance: the specification, method version, and computation run behind a number are recorded and inspectable.
Open a published resultMethodology is stated in plain language — sampling frame, weighting approach, margin of error, and known limitations travel with every result.
Read the methodologyResult commitments are anchored to a public ledger so the record is tamper-evident after publication.
An independent verifier can recompute a published commitment and check it against the anchored record.
Why it matters commercially
Defensibility is the deliverable
Institutions do not get in trouble for citing a number. They get in trouble when the number is challenged and nobody can reconstruct how it was produced. An integrity record turns that moment from a crisis into a lookup.
For regulated buyers, newsrooms, and anyone whose recommendation will be attacked on purpose, that record is the difference between a defensible finding and an expensive retraction.