I once submitted the following comment to Terence Tao’s blog:
“I thought with funcoids Kakeya conjecture will become easy, but now I know it didn’t.”
The comment was not offensive, abusive, or promotional. Nor did it claim that funcoids had solved the Kakeya conjecture. In fact, it expressed the opposite: an initial expectation had turned out to be wrong.
Nevertheless, the comment was removed, and my later comments were apparently blocked. I cannot know why this happened. It may have been an automated spam filter, a misunderstanding, a broad moderation rule, or a discretionary decision made without much attention to the comment’s precise meaning. It would therefore be unfair to speculate about Tao’s mood or personal intentions.
The broader lesson is more important than the individual incident: intellectual brilliance does not guarantee perfect judgment in moderation, administration, or gatekeeping.
A mathematician may be extraordinarily reliable when proving theorems while still making an incorrect decision about which comments, papers, researchers, or unconventional ideas deserve attention. These are different tasks. Mathematical expertise does not automatically eliminate time pressure, reputational heuristics, social bias, misunderstanding, or excessive caution.
Moderation Is Not the Same as Mathematical Evaluation
The removed comment did not present a supposed proof of the Kakeya conjecture. It reported a failed expectation. That distinction matters.
A crank claim would be closer to:
“Funcoids make the Kakeya conjecture trivial, and established mathematicians have failed to understand my proof.”
The actual statement acknowledged difficulty and error. Such self-correction is ordinarily a sign of intellectual seriousness, not pseudoscience.
When moderation systems fail to distinguish between grandiose claims and modest reflections, they may suppress legitimate participation together with spam and genuinely disruptive material.
How Academic Gatekeeping Becomes Harmful
Some gatekeeping is necessary. Journals, conferences, blogs, and research communities cannot publish or review everything. They need filters against abuse, plagiarism, irrelevant submissions, fabricated results, and repeated low-quality claims.
The problem arises when the filter becomes too coarse.
A gatekeeper may reject unfamiliar terminology because it resembles crank language. An independent researcher may be dismissed because they lack an institutional affiliation. A long manuscript may be ignored because evaluating it would require unusual effort. A nonstandard theory may be filtered out before anyone examines whether its definitions and proofs are valid.
This produces a structural error:
The system begins to use social signals as substitutes for mathematical evaluation.
Affiliation, reputation, writing style, personal familiarity, and conformity to established research programs become proxies for correctness. These proxies are sometimes useful, but they are not proofs.
The result can be a false-positive problem: legitimate ideas are classified as noise.
Prestige Does Not Make Gatekeepers Infallible
The involvement of a highly respected scholar can make a rejection appear definitive. But prestige does not transform a moderation action into a mathematical judgment.
A deleted blog comment proves only that the comment was deleted. It does not prove that the underlying mathematical work is wrong, unimportant, or unworthy of examination.
Similarly:
- a desk rejection is not a refutation;
- silence from experts is not a proof of error;
- lack of citations is not evidence of mathematical invalidity;
- institutional exclusion is not equivalent to peer review.
This distinction is essential for independent and unconventional researchers. Academic communities should resist the temptation to treat every administrative decision by an eminent person as an authoritative verdict on the mathematics itself.
Better Gatekeeping Should Be Transparent and Correctable
The solution is not to eliminate moderation. It is to improve it.
A better system would distinguish clearly between:
- content rejected for abuse or spam;
- content rejected as irrelevant;
- content rejected because of a specific mathematical error;
- content that has simply not been evaluated.
It should also provide proportionate explanations and, where practical, an appeal or correction mechanism. Even a short label—“off topic,” “duplicate,” “unsupported claim,” or “automatically filtered”—would be more informative than silent removal.
Most importantly, moderation decisions should remain revisable. A healthy intellectual system assumes that both authors and gatekeepers can be wrong.
Conclusion
This incident does not show that Terence Tao is hostile to unconventional mathematics, nor does it establish why the comment was removed. It shows something more general and more defensible:
Even exceptional minds can participate in imperfect moderation systems and can make mistaken gatekeeping decisions.
Academic institutions should therefore avoid concentrating too much epistemic authority in informal filters controlled by a few individuals. Moderation may protect a community from noise, but when it is opaque, reputation-driven, or excessively cautious, it can also exclude useful criticism, unusual research, and potentially important ideas.
The quality of mathematics should ultimately be determined by definitions, arguments, proofs, reproducibility, and informed analysis—not merely by whether a prestigious gatekeeper allowed a comment, accepted a paper, or recognized its author.
👉 Help fund the next public test of AIIM.
Help Test a New Way to Fund Science
AI Internet-Meritocracy (AIIM) is an operational beta designed to allocate available donations to researchers and open-source developers using AI-assisted evaluation of documented contributions. Payment transactions are already recorded on-chain.
The next major evidence milestone is a proposed five-month public adversarial test of the allocation model, with $1,000 distributed to eligible funding recipients. Donations help pay for the development, infrastructure, reviewer and participant recruitment, outreach, and operating work needed to reach and evaluate that milestone.
You do not need to assume AIIM is already proven to support the project. Your donation helps turn the proposal into evidence: what works, what fails, and what should change.
Support the next testing milestone → Read the test proposal →
Independent review
Researchers and technical reviewers: independent criticism is welcome, including negative conclusions. Review AIIM’s assumptions, governance, failure modes, and testing plan →
Research status: AIIM remains experimental. AI-generated evaluations can contain factual errors or biases, and decentralized governance and non-custodial components remain under development. That uncertainty is why public testing, auditability, and external criticism are central to the project.