The first submission to a new conference is always a lesson in institutional culture, and the question from the ICLR author in the original post captures that anxiety perfectly. They are coming from the familiar, closed-loop worlds of ICML and NeurIPS, where the review process is a private negotiation between authors and a handful of reviewers. ICLR, by contrast, throws the doors open, and for a newcomer, that exposure feels less like an opportunity and more like a vulnerability. The core tension is simple: when does your private work become public property, and what does that mean for your sense of ownership? This is the same tension that runs through our piece on Neural architecture search promised progress, but transformers emerged elsewhere, where the field's open, collaborative nature produced a landscape that no single lab could control.
The author's specific questions reveal the practical gaps in ICLR's documentation. The portal says papers are public "from the beginning of the review process," but the timeline is ambiguous. Reviews start in October, while the discussion period begins in November. Our read is that the "beginning" means the moment reviews are first posted, not when the discussion forum formally opens. So yes, your paper and its supplementary materials, including your code, become visible to the community in October. That is a stark difference from the private review you are used to. And yes, once a reviewer posts their initial feedback, both you and the wider community can see it immediately. There is no hidden embargo period. This is a deliberate design choice to foster transparency, but it demands a different kind of preparation. You are not just writing for three reviewers; you are writing for a global audience of your peers who can and will dissect your methodology in real time.
The concern about code being redistributed is legitimate, but it is also a reflection of the platform's philosophy. ICLR believes that open review accelerates science, and that the risk of idea "scooping" is outweighed by the benefits of rapid, collective feedback. For a first-time author, the practical takeaway is to treat your submission as a public release from the moment you click the button. If you are not ready for the world to see your code, consider whether you should submit it as supplementary material at all. You can still include a link to an anonymized repository that requires access, or you can describe the code's architecture and defer the full release until after acceptance. This is not about hiding your work; it is about controlling the narrative. The community benefits from your insights, but you should decide when and how the underlying assets are shared. This is a more nuanced approach than simply accepting the platform's default settings.
We would tell any reader facing this same situation to stop thinking of ICLR as a version of ICML with a public forum. It is a different beast, and it requires a different mindset. The process is not a hurdle to clear; it is a conversation to join. The fact that your work is public in October means you have an opportunity to engage with the community, answer questions, and refine your ideas before the final decisions are made. That is a powerful tool, but only if you are prepared to use it. As we noted in our exploration of Explore how Gemini simplifies shopping with Flipkart in India, the real value often comes from the interaction, not just the final product. The question you should be asking is not just "when does my work go public?" but "how do I make the most of this public window?" The answer to that will determine whether you see the open review as a threat or as a feature. Watch the forums closely on the day reviews drop, because the first hour of public feedback will shape the entire discussion period, and your response to it will set the tone for how your work is perceived.