# Verified Research Lab: expanded site guide

> Independent testing for consequential research claims in computer systems and software.

## Purpose

VRL examines published and pre-publication research in operating systems, distributed systems, databases, networking, software engineering, programming languages, compilers, systems security, and performance. Initial work focuses on papers and computational artifacts available for independent examination. Evaluations support post-publication review, publication preparation, follow-up research, technical adoption, funding, and due diligence.

## Engagement process

1. Fit check: VRL reviews the question, technical area, materials, access constraints, conflicts, and timing.
2. Test plan: The claim, conditions, tolerances, exclusions, reporting format, confidentiality, and publication terms are agreed before work starts.
3. Independent work: VRL reconstructs, reruns, varies, and records the work without sponsor control over the method or finding.
4. Bounded conclusion: The deliverable states what was tested, what happened, where the conclusion stops, and what remains uncertain.

## Verification method

VRL uses **reproduction** for re-executing reported work with the supplied artifact or a reconstructed equivalent. It uses **replication** for testing the same claim with an independently constructed implementation, dataset, or experimental design.

- Specify the claim, decision, tolerances, conditions, and exclusions.
- Preserve code, data, versions, provenance, access conditions, and known gaps.
- Reconstruct the environment and compare independently obtained results with declared tolerances.
- Audit baselines, workloads, measurement, statistics, exclusions, and validity threats.
- Test reasonable variations in parameters, environments, datasets, workloads, and cost assumptions.
- Report the conclusion, uncertainty, limits, review status, and revalidation trigger.

## Evidence labels

- Artifacts identified
- Artifacts available
- Artifacts functional
- Result reproduced
- Method audited
- Robustness tested
- Independently corroborated
- Supported within tested conditions
- Not reproduced
- Inconclusive

Labels describe the completed work and conclusions for the claims examined, whether the review covers part or all of a paper.

## Services

### Research reproduction and replication

Reruns reported results from supplied materials, or tests the same claim with an independent implementation, dataset, or experimental design.

Canonical section: https://verifiedresearchlab.org/services/#reproduction-and-replication

### Claim and method review

Assesses whether the research design, analysis, and evidence support the stated claim.

Canonical section: https://verifiedresearchlab.org/services/#claim-and-method-review

### Artifact review

Checks whether code, data, environments, and instructions can be retrieved and executed independently, then assesses versioning, preservation, and hosting readiness. Teams can request a separate advisory preparation pass.

Canonical section: https://verifiedresearchlab.org/services/#artifact-review

### Benchmark audit

For researchers preparing performance, cost, security, reliability, or correctness comparisons. Examines baselines, workloads, resource parity, measurement, statistics, and operating conditions.

Canonical section: https://verifiedresearchlab.org/services/#benchmark-audit

### Independent verification report

For researchers, laboratories, journals, conferences, and public-interest sponsors. Publishes a versioned record of the claims, evidence, methods, subject response, review, and correction status.

Canonical section: https://verifiedresearchlab.org/services/#independent-verification-report

### Technical due diligence

For investors, acquirers, R&D leaders, and funders. Examines research-derived technology before investment, acquisition, procurement, or adoption.

Canonical page: https://verifiedresearchlab.org/services/technical-due-diligence/

## Fit-check brief

The contact form is a client-side email-draft helper. It validates the required name, email, and question fields, assembles the entries into a structured plain-text message, and opens the visitor's configured email application using a `mailto:` URL. The website does not receive, upload, or store the entries. The visitor must review and send the draft. If no email application is configured, the visitor can email contact@verifiedresearchlab.org directly.

## Governance commitments

- Sponsors may fund work but may not direct conclusions.
- Funding, conflicts, competence limits, and publication status are recorded.
- Subjects may correct factual errors and submit a response without gaining veto power.
- Public findings receive independent sign-off suited to their scope and risk.
- Evidence is preserved where legal, contractual, privacy, and security obligations permit.
- Amendments, supersession, and withdrawal remain visible.

## Current status

VRL is preparing its method and first evaluations of existing research. The public report registry contains no completed findings yet.

## Canonical links

- Website: https://verifiedresearchlab.org/
- Services: https://verifiedresearchlab.org/services/
- Method: https://verifiedresearchlab.org/methods/
- Reports: https://verifiedresearchlab.org/reports/
- Governance: https://verifiedresearchlab.org/governance/
- About: https://verifiedresearchlab.org/about/
- Contact: https://verifiedresearchlab.org/contact/
- Compact site guide: https://verifiedresearchlab.org/llms.txt