Proof is built by contributors who run real trials and stand behind them.
This page is the contributor charter. It describes who Proof is for, what contributors commit to, what they receive in return, and what Proof refuses to offer them. Read it before applying.
Proof is infrastructure for serious field evidence. It is not a community, a creator platform, or a place to promote products. Applications are reviewed for accountability.
Four contributor groups. The same standards apply to all four.
Proof Records can be created by agronomists, farmers and farm managers, researchers and institutions, and trial partners or manufacturers. The contribution mode differs; the schema, the trust rules and the accountability are the same for all four.
- 01AgronomistsTreatment / control trials across client farmsIndependent and company agronomists run on-farm trials with paired treatment and control arms. Each trial becomes a Proof Record. Records can be assembled into Client Evidence Packs that the agronomist hands to the farms they advise.
- 02Farmers and farm managersRecording your own workFarmers testing a product or practice on their own land record what they did, what they compared it against, and what was measured. Farm identity stays in the private envelope. The evidence becomes part of the public ledger.
- 03Researchers and institutionsOn-farm trials and field studiesLevy bodies, universities, agricultural research institutions and independent research organisations record on-farm trials under the same schema as every other contributor. Each trial earns a permanent, citable record that survives long after project funding ends.
- 04Trial partners and manufacturersSponsored and manufacturer-affiliated trialsManufacturer R&D teams, sponsored trial partners, and contributors with declared distribution relationships record under the same rules as every other contributor. Affiliations and sponsorships are declared on the record itself. Conflict declaration changes how a record is read; it does not change which records get published.
No contributor group earns a different trust state. No contributor group is shown more prominently in the records explorer. No contributor group is granted the right to withhold a negative outcome. The schema does not distinguish between them at the moment of lock.
Contributors receive a permanent, citable record of their work.
Each Proof Record carries a permanent UID and a canonical hash. Once locked, it cannot be silently edited, withdrawn under pressure, or lost when a platform pivots. The contributor's work survives the lifecycle of their organisation, their funding, and the system Proof itself runs on.
A record can be cited by its UID. The hash confirms the citation points to the same record the reader is looking at now. This is what serious agronomic evidence has needed for decades and rarely had: a permanent, verifiable address.
Contributors retain attribution on every record they create. Verified contributors gain access to the Workbench, the Treatment / Control Builder, the Evidence Manifest, and the ability to raise Evidence Queries on records by other contributors.
The artefacts a contributor produces.
The unit of contribution is the Proof Record. A single trial: treatment, control, conditions, evidence, outcome. Each record carries the same seven tabs: Overview, Tested, Outcome, Field, Benchmark, Evidence, Ledger.
Agronomists can group their own records into Client Evidence Packs: a structured bundle handed to a farm client showing what was tested on the client's behalf, with consistent formatting and the same redaction rules as the public ledger.
As records accumulate, comparable records form cohorts. Cohort summaries appear once the privacy threshold is met (five records, three farms, two contributors). Cohorts describe shape and direction; they do not rank.
For the lifecycle of a record from draft to publication, see How it works →
What every contributor commits to at lock.
These are the structural commitments. They apply to every contributor regardless of group or affiliation. A contributor who cannot meet them at lock should not lock the record.
- Article IRecord real trials only.Every Proof Record describes a real on-farm trial that took place under the conditions stated. Hypothetical, simulated and demonstration records are not Proof Records.
- Article IIDeclare the comparison structurally.Treatment arm, control arm, and same-management declaration go in the Treatment / Control Builder, not in free text. Trials without a control are recorded as observational and earn fewer trust pips.
- Article IIIDeclare conflicts before lock.Sponsorship, manufacturer affiliation, distributor relationships, and any other commercial interest in the products being tested are declared on the record itself. Conflicts change how a record is read; they do not change which records get published.
- Article IVLock null, negative and mixed outcomes with the same care as positive ones.The outcome direction is recorded as it was measured. A null result is evidence. A negative result is evidence. A mixed result is evidence. The contributor does not choose which outcomes to lock.
- Article VAttach evidence honestly.Application photos, yield files, protocol documents, lab reports and field maps are linked to the parts of the record they actually support. Evidence is not stretched, recategorised, or substituted to make a record appear stronger than it is.
- Article VIAddress Evidence Queries.A query raised against your record is engaged with, not ignored. The contributor either addresses the query with an addendum, supplies further evidence, or accepts that the question becomes part of the record's history unanswered.
These articles are not enforced editorially. They are enforced structurally by the schema, and over time by the visibility of every record's history. A contributor who breaks them does so on the public record.
Public evidence preserved. Private farm data protected.
The redaction bar that appears on every public record marks the boundary between the public envelope and the private envelope. Farm identity, exact field boundaries, raw GPS, raw evidence files, EXIF metadata, invoices and private contributor notes stay in the private envelope.
The contributor's own identity is verified at application but the contributor controls how it appears on records. Records can be attributed to the contributor's name, to a role (Verified agronomist, Verified researcher), or to a pseudonymous handle linked to a verified Proof identity. The verification is held; the attribution is the contributor's choice.
For the full rule on what is public and what stays private, see Trust model §§ II–III →
Verified once. Attributed how the contributor chooses.
Every contributor passes a single verification at application. Proof confirms that the applicant is who they say they are, that their professional role and any organisational affiliation are real, and that the contact they nominate can be reached. The verification is held by Proof. It is not published.
On each record, the contributor chooses one of three attribution modes:
- NamedThe contributor's full name and organisation appear on the record.
- RoleA role-only attribution (Verified agronomist · independent, Verified researcher · institution, Verified farm manager) appears without the contributor's name.
- PseudonymousA persistent pseudonym linked to a verified Proof identity. Records by the same pseudonym are recognisably the same contributor; the verified identity behind the pseudonym is held by Proof.
Affiliations are separate from attribution. Sponsorship and manufacturer affiliation appear on every affected record regardless of which attribution mode the contributor uses.
Once a record is locked, it stays locked.
Locking is irreversible. The values, the evidence references and the canonical hash become permanent at the moment the contributor confirms the lock. From that moment the record is append-only: corrections and clarifications are added as addenda, dated and visible, sitting beside the original.
A contributor cannot withdraw a record because the outcome was unflattering, because a sponsor changed position, because a relationship soured, or because the contributor changed their mind. A contributor who cannot accept this should not lock the record.
For the immutability rule and how addenda preserve history, see Trust model § VI →
Contributors are accountable to one another, on the public record.
Verified contributors may raise an Evidence Query on any locked record outside their own organisation. The query is structured: a method clarification, a confounder, a measurement concern, a cohort applicability question. Adversarial framing is rejected at submission.
A query does not erase or invalidate the record. It opens a public, dated question that the recording contributor may address with an addendum. The query and its resolution become part of the record's history, visible alongside the original.
Repeated unaddressed queries are themselves visible. A contributor's pattern of engagement with queries is part of the public record of their work.
For why Proof uses the term "query" rather than "challenge", see Trust model § VIII →
§Failed and null results stay.
Some things Proof refuses to offer contributors.
These refusals are part of the network's integrity. A platform that offers them cannot be a Proof.
- 01Contributors do not get a leaderboard.Contributors are not ranked. Records are not ordered by contributor reputation, prolificness, or outcome direction.
- 02Contributors do not get to retract a published record.A record cannot be withdrawn for being inconvenient. Corrections happen through addenda. The original stays.
- 03Contributors do not get to publish a curated subset of their trials.If a contributor begins a structured trial and locks a draft, they accept the lifecycle. Cherry-picking which trials to publish defeats the evidence base.
- 04Contributors do not get a promotional channel.Proof Records are not press releases. Records are not styled, ranked, or surfaced to favour the contributor's preferred narrative.
- 05Contributors do not get followers, engagement metrics, or a feed.There is no social layer. A record is read for its evidence and its trust state, not for its audience.
- 06Contributors do not get to mark their records 'verified by Proof'.Proof verifies the contributor's identity and validates the record's structure. It does not certify that a product works.
A contributor who needs any of the above should look for a different platform. A contributor for whom these refusals feel correct is the kind of contributor Proof was built for.
From application to lock.
The contributor workflow has two halves. The first is application and verification, which happens once. The second is the record lifecycle, which runs for every trial.
For the full step-by-step lifecycle with artefacts at each stage, see How it works →
Four shapes of contribution. Each follows the same rules.
These are illustrative shapes, not real contributors. They show how the same schema and the same trust rules accommodate different modes of work.
- αIndependent agronomistLincolnshire and Norfolk · winter wheat and OSR
An independent agronomist runs paired-strip biostimulant trials across nine client farms in the 2026 season. Each trial becomes a Proof Record. Three trials show a positive yield response, two show no measured difference, one is recorded as observational because the control strip was lost to slug damage. The agronomist hands each client farm a Client Evidence Pack containing the records taken on the client's land, with the same redaction rules as the public ledger.
- βFarm managerYorkshire · winter wheat on heavy clay
A farm manager records a single biological-programme trial across two plots of winter wheat. The protocol is locked before drilling. The outcome is +0.13 t/ha against the farm-standard programme. The record sits in the explorer under role-only attribution; the farm identity stays in the private envelope. The trial joins a six-record cohort that has just passed the privacy threshold.
- γResearch institutionNational research body · maize across western counties
A research body running multi-site maize trials records each site as a separate Proof Record under the institution's attribution. Some sites show positive responses; one site is recorded as a negative outcome after the in-furrow treatment failed to deliver. All five records lock under the same schema. The negative result is published with the same prominence as the positive ones.
- δManufacturer R&D teamSponsored trials · biofertiliser on peat
A manufacturer's R&D team records a sponsored trial of their own biofertiliser on peat soils in eastern England. The trial is locked with sponsorship and manufacturer affiliation declared as conflicts. The recorded outcome is negative: the biofertiliser reduced yield by 0.18 t/ha against the untreated control. The record is published with the conflict declarations visible; the negative direction stays.
The records entered in Proof's first seasons will define what trustworthy on-farm evidence looks like for the decades that follow. The contributors who lock those records are the people Proof was built for.
Applications are reviewed for accountability. Verification takes one to three weeks.
Apply to contribute