Independent review of our science-funding initiative and SciWard
We invite independent scrutiny, including critical or negative assessments. Reviewers do not need to endorse our science-funding initiative, SciWard, or any related project. We encourage publication on websites, journals, forums, repositories, or other venues that our science-funding initiative does not control.
Who this is for
Researchers, engineers, AI-safety specialists, economists, nonprofit and governance specialists, open-science researchers, security researchers, software developers, and other people with relevant expertise are welcome to review the project.
What to review
- SciWard: whether its proposed evaluation and funding mechanisms are coherent, testable, and resistant to manipulation.
- Testing and evidence: whether experiments, preregistrations, metrics, and reported results support the claims being made.
- Governance: incentives, accountability, appeals, concentration of power, and transition toward centralized administrative governance.
- Technical implementation: architecture, source code, blockchain components, security assumptions, and operational limitations.
- AI safety: assumptions behind the Symbiote initiative, human-in-the-loop mechanisms, prompt-injection risks, and failure modes.
- Nonprofit and financial transparency: whether public descriptions of legal control, spending, fundraising, and project status are sufficiently clear.
Useful source material
- SciWard
- Testing & Evidence
- Adversarial testing plan
- Current product status and limitations
- Transparency
- Key Contributors
- The Strongest Arguments Against SciWard
Questions reviewers may investigate
- Are the project’s central claims stated precisely enough to be falsifiable or empirically tested?
- Do the proposed evaluation mechanisms measure what they claim to measure?
- What incentives could cause gaming, Goodhart effects, collusion, or strategic self-presentation?
- Are the planned adversarial tests strong enough to reveal important failure modes?
- Which decisions should remain subject to independent human judgment?
- Are the project’s current limitations and development status disclosed clearly?
- Which technical or governance assumptions are most fragile?
- What evidence would materially increase or decrease confidence in SciWard?
- What risks are missing from the current documentation?
- Which claims should be weakened, strengthened, or tested differently?
Independence policy
our science-funding initiative does not require a favorable conclusion. Reviewers retain editorial control over their work. We may point out factual errors or provide source material on request, but reviewers are not expected to adopt our wording or conclusions.
Substantive external reviews may be linked from our science-funding initiative’s evidence pages regardless of whether they are supportive, mixed, or critical. Inclusion of a review does not imply endorsement of its conclusions.
Conflicts and compensation
Reviewers should disclose any material relationship with our science-funding initiative, Victor Porton’s Foundation, project contributors, or funders. If compensation is ever offered for a review, that fact should be disclosed publicly and payment must not depend on the review being favorable.
Publish independently
The preferred outcome is a review published somewhere our science-funding initiative does not control. Suitable venues can include an academic or personal research website, a journal or repository, an AI-safety or open-science forum, or another independently administered publication platform.
Submit your independent review
Published a substantive independent assessment? Submit it here. Supportive, mixed, and critical reviews are welcome under the same policy. Pseudonymous reviews are accepted; reviewers who want their credentials privately verified may provide them confidentially.
Published independent reviews and criticism
No substantive external reviews are listed here yet. When they are available, this section will link to them with the author, affiliation where applicable, publication venue, date, and a clear note that our science-funding initiative does not control the reviewer’s conclusions.
Related contingency plan: What SciWard would change if adversarial testing reveals vulnerabilities (paid independent checks, human challenges and retesting).
Related contingency plan: How SciWard would respond if adversarial testing reveals vulnerabilities (paid review, challenges and retesting).
Help Turn SciWard’s Beta into Evidence
SciWard remains experimental. The next proposed milestone is a five-month public adversarial test, with $1,000 planned for distribution to eligible funding recipients. Donations support the work required to run and evaluate that test.
Fund the public test → Test proposal → Independent review →
AI-generated evaluations can contain factual errors or biases. Public testing, auditability, and external criticism remain part of the project.
