NINeS 2026: Statement of Purpose

The goal of NINeS is simple: Create a conference whose reviewing process is welcoming of new ideas, constructive about flawed ideas, and humble about our ability to tell the difference.

Why is this goal important? As a research community, our goal should be to lay the intellectual foundation for future networked systems. Laying such a foundation requires entertaining a broad spectrum of new ideas, knowing that almost all of these ideas will never be realized in operational use, many will be hard to fully validate (because we have no operational experience with them), some may be difficult to deploy in the current ecosystem, and a few may eventually be seen as seriously misguided. We need a reviewing process that welcomes new ideas despite these many limitations. In addition, as a community we are responsible for training the next generation of researchers. The intellectual environment in which they mature should encourage them to imagine the future, not just perfect the present. In short, our task is to nurture the ideas, and the researchers, that will shape the future of networked systems.

Currently, we are failing at this task. New ideas have a hard time being published, which makes students worry about taking risks in their research. They also feel disrespected by the dismissive tone in many reviews. One might think reviewing is only a minor aspect of the community, but as David Patterson has observed “benchmarks shape a field” and the benchmark for our research community is publishing papers. Thus, the first step towards nurturing the ideas and researchers of the future must be a more welcoming reviewing process.

Why do we think we can accomplish this goal? Many PC chairs of the SIGCOMM conference have tried hard to make the conference more receptive to new ideas, but with little success; why do we think NINeS will fare better? The SIGCOMM conference is the flagship conference of the SIGCOMM SIG, and is therefore bound by the majority preferences of the SIGCOMM membership and the prevailing reviewing practices of its PC members. NINeS will not be sponsored by SIGCOMM, and has no such constraints. Our intent is for NINeS to slowly build its own community around a shared set of beliefs about the importance of the intellectual foundation and the nature of the reviewing process. Thus, the key difference between creating NINeS and previous efforts at reform is that we are not trying to change the minds of the current SIGCOMM membership nor the practices of its flagship conference, we are only trying to provide a home for those who are interested in the goals that are central to NINeS.

In addition, the presence of multiple conferences in the field of networked systems allows NINeS to prioritize papers that present new ideas over those that are refining more established ones. These more incremental papers already have access to several publication venues, so the research community will still have the benefit of their contributions, while NINeS can focus on the kinds of papers that have trouble being published today. Thus, we are not making a judgement that one kind of paper is better than the other; we think both are necessary but one faces too many challenges while the other does not, so we are trying to rectify that imbalance. We intend NINeS to be a highly selective conference with a different bar for acceptance, not a lower one, thereby broadening the spectrum of ideas presented to the research community.

How will we accomplish this goal? To be more welcoming of new ideas, NINeS must establish a different culture of reviewing, one that first focuses on the quality of the key ideas, and only if those ideas are found sufficiently new and interesting is the quality of the idea’s evaluation examined. It is often hard to evaluate new ideas, so the standard should be whether the conceptual and/or practical benefit of a new idea is sufficiently plausible, not whether it is thoroughly proven. Our evaluation criterion purposely does not include the deployability of an idea (i.e., as determined by its compatibility with the current infrastructure or the incentives of the incumbents), as it is not the job of the research community to limit future possibilities to those whose deployment we can imagine now; the Internet would clearly have failed that test.

It is easy to write general prescriptions like this, but by themselves they accomplish little because it is equally easy for individual reviewers to ignore them. To prevent this possibility, we first note that reviewing is not just an individual practice but is also a social one, both because the decisions involve the entire PC and because reviewing habits are learned from others by seeing the reviews of our papers and watching other people when we serve on PCs. Thus, our current culture of reviewing – and here we are using “culture” in the sense of the customs of a particular social group – is self-perpetuating and therefore hard to change. This is why it is crucial to not start with the current SIGCOMM membership as the social group, but instead build a new community around NINeS; the two groups will overlap substantially, but we will try to start NINeS with an initial group that believes in the mission and practices we describe here.

But choosing an appropriate starting community is not enough. We should design our reviewing procedures to help reinforce the community’s reviewing values. That is, the goal of the NINeS paper reviewing procedures is not just to make good decisions, but also to create an environment that will lead people to review in accordance with the values we hope NINeS will embody. We think two procedural principles will play an important role in realizing this goal:

1. PC members should focus more on the quality of decisions than the quality of feedback. When reviewers focus on giving feedback, they tend to become more arrogant about their opinions, when in fact they have spent far less time on the material than the authors. This does not imply that reviewers have nothing valuable to say, but current SIGCOMM reviews are far more opinionated than informed, and it has created an extremely toxic environment for young researchers. We must discard the fantasy that PCs are infallible oracles and embrace an attitude of humility in which reviewers are not seen as delivering judgements from a position of superior wisdom, but instead are merely trying to identify the good ideas in each paper.

2. The PC members should interact with each other before forming their final opinions. Rather than merely repeat this platitude in the PC instructions, the initial reviews at NINeS will be documents intended for other reviewers rather than for authors. This will allow reviewers to forthrightly express uncertainty and raise questions for other reviewers to answer, and make it natural for reviewers to respond to the thoughts of others. The reviews seen by authors can be created later in the process, and may resemble PC summaries more than individual reviews.

Is this enough? The NINeS PC cannot determine the merit of a paper, it can only render a highly subjective decision about its acceptance. To construct the intellectual foundation of our field, the research community must engage more actively in the discussion and exploration of ideas. Increasing this engagement will be the focus of the in-conference activities at NINeS.