Anonymous GitHub

When Anonymous Links Break, Your Workflow Needs Resilience

Anonymous links that worked yesterday are failing today, right when ICLR rebuttals are due.

3 min readMachine Learning

Anonymous links are a fragile foundation for critical work, and the researcher who discovered their ICLR rebuttal links had broken mid-process learned this the hard way. Our opinion is straightforward: if your workflow depends on sharing access through ephemeral, unmanaged links, you are building on sand, and the academic review cycle is the worst possible place to discover that. This isn't a minor inconvenience; it's a systemic vulnerability that undermines the very transparency and collaboration that open review processes aim to deliver.

The timing here matters enormously. With 42,000 Submissions Push ICLR to Reconsider Its Acceptance Rate flooding the system, the pressure on researchers to produce clear, accessible rebuttals has never been higher. A broken link during rebuttals doesn't just frustrate the author; it erodes the reviewer's ability to evaluate revisions fairly. When reviewers cannot access supplementary materials, the entire premise of a transparent, evidence-based rebuttal collapses. This is not a hypothetical edge case. It is a predictable failure mode of relying on sharing mechanisms that offer no persistence, no version control, and no audit trail. The researcher's post asking "Anyone else having the same issue?" reveals a deeper truth: this problem is widespread, yet we treat it as an individual technical glitch rather than a structural flaw in how we manage research data.

What this means for our readers is practical and immediate. If you are preparing a rebuttal, a supplementary dataset, or any collaborative research output, ask yourself whether your sharing method survives a browser restart, a permission change, or a platform update. The solution is not to abandon digital collaboration but to adopt tools that treat data as a first-class citizen, tools that manage access, track changes, and persist links regardless of how they were generated. Clarity in your ICLR rebuttal can help reviewers see your revisions is only possible when the underlying infrastructure supports that clarity. A well-written rebuttal is useless if the evidence behind it is inaccessible.

The specific takeaway here is direct: audit your sharing links before you submit anything time-sensitive. Test them from an incognito window, from a different device, and after logging out. If they fail any of those tests, your workflow is broken. The broader question this incident raises is whether the research community will continue to accept ad-hoc sharing as a standard, or whether we will demand resilient, AI-native systems that keep data accessible and workflows intact. The answer will determine not just how we manage rebuttals, but how we build trust in the entire peer review process.

From Machine Learning

Mid ICLR rebuttals and I found out the links I made are not working. Anyone else having the same issue?

Read the original at Machine Learning