Several Mid-Central judges told me that I held too many meetings, sometimes arrived with my preferred answer, and did not always respect problem authorship carefully enough.

They were right.

As Chief Judge, I am responsible for delivering a correct problem set on time. I had begun to treat that responsibility as a reason to remain too close to every decision. The team experienced the cost: more meetings, less room to form an independent view, and changes that could make an author feel displaced from the problem they had created.

The lesson was not that I needed a better explanation for the same process. The process needed to change.

Accountability does not require control of every decision

The Chief Judge should know where the contest is exposed. Which problems lack an independent solution? Which test suites are still provisional? Which statement change could alter the intended problem? Which deadline has no credible owner?

That visibility is different from making every technical choice.

If I give my view before the other judges have formed theirs, I narrow the discussion. A reasonable suggestion from the Chief Judge can become the default simply because of the role attached to it. The team may reach agreement without testing whether it is genuine.

In 2027, I will ask for the judges’ assessments first. When views differ, we will name the disputed claim and decide what evidence could settle it. My responsibility is to close the decision, assign the remaining work, and protect the quality bar. It is not to make every problem look as though I wrote it.

Use meetings for decisions

The new plan concentrates judging into a three-month sprint with six one-hour meetings. Problem collection can continue throughout the year. The shared meeting clock should be reserved for work that benefits from the team being present together.

Each session has one stage of the release to close:

  1. Select provisional problems and backups. Assign authors and reviewers.
  2. Review reference solutions, generators, validators, and the main correctness risks.
  3. Require a runnable draft for every active problem.
  4. Resolve statement ambiguities and obtain author approval for substantive changes.
  5. Choose the final set and drop problems with unresolved critical defects.
  6. Verify the package, complete external review, and confirm submission.

Agendas should arrive at least 48 hours in advance. Assigned reviews happen before the meeting. Every session ends with a written decision, an owner, and a deadline.

This structure gives the meetings a test. If a session ends with a broad discussion and no changed state, it did not do its job.

Respect authorship as part of contest quality

A contest problem passes through many hands. Reviewers find ambiguous language. Testers expose missing cases. Editors improve the statement. The Chief Judge has to decide whether the problem is ready to ship.

The original author still owns the mathematical object being reviewed.

A substantive statement edit can change the problem’s interpretation, intended difficulty, or proof obligation. In the revised process, such a change requires the author’s approval. If the disagreement cannot be resolved in time, we should consider dropping the problem. Shipping a different problem under the author’s name is a worse outcome than using a prepared backup.

This rule also improves discussion. Reviewers must explain the defect they are trying to remove. Authors must respond to a concrete risk. The question becomes whether the current version can be defended, rather than whose wording wins.

Independent validation is the release gate

Consensus in a meeting does not prove that a problem is correct.

Every selected problem needs an independent solution and enough test coverage to reject plausible wrong approaches. Generators and validators need review. Limits need evidence. A final package should contain no unresolved defect that could change a team’s result.

Backup problems matter because a release gate is credible only when “drop the problem” remains possible. Without a backup, the team is tempted to reinterpret a critical issue as acceptable because the schedule has removed every alternative.

Measure whether the team changed, not whether the plan sounded good

I will look for evidence after the next cycle: six consistent meetings, complete drafts by the third, final selection by the fifth, independent validation for every selected problem, and author approval for substantive statement changes.

I will also ask the judges for feedback within two weeks of the contest. A process can deliver on time while making capable people unwilling to return. That is a real failure for a volunteer technical organization.

The feedback I received was uncomfortable because it described behavior I could recognize. I am grateful that the judges said it plainly.

My next responsibility is to make the correction visible in how the work runs.