SIGOPS’ Response to Community Feedback on the Frequency of the SOSP and OSDI

In late 2019 some members of the systems research community came together to present a proposal that SOSP, the ACM Symposium on Operating Systems Principles, and OSDI, the USENIX Symposium on Operating Systems Design and Implementation, be held annually rather than biennially. Their goal was to support the publication of more work produced by the growing number of researchers in the field. Although the number of publication venues has grown since the inception of both SOSP and OSDI, the number of active researchers in the field has increased at a greater rate.

USENIX and ACM SIGOPS carefully considered this proposal, taking into account the needs of the community, the load placed on reviewers, the existing publication venues, the different options to scale up OSDI and SOSP, and the financial commitments created by contractual obligations to venues. Ultimately SIGOPS has decided that a measured response to the community’s suggestion is to support USENIX’s initiative to hold OSDI in both 2021 and 2022 as a two-year exploratory experiment, while maintaining SOSP’s biennial frequency during this experimental phase.

The PC chairs for SOSP 2021 and OSDI ’21 have collaboratively agreed on coordinated submission and review policies, and we will continue this collaboration moving forward.

The addition of an OSDI submission and review cycle into the calendar year will likely affect reviewer load and submission volume for all systems research venues. In partnership with USENIX, SIGOPS will closely monitor these impacts on the community, review the available data from the experimental phase, and will make any needed changes in advance of our 2023 conferences.

Like USENIX, SIGOPS understands the motivation for scaling up conference publication capacity given the growth of our systems research community. As one would only expect when considering a change of the magnitude of doubling the frequency of SOSP and OSDI, SIGOPS sees both potential benefits and risks. Clearly, matching conference publication capacity in the systems area to the volume of strong work benefits the community, both through timely dissemination of new ideas and facilitation of career advancement for community members publishing in those conferences.

Yet community members also voiced potential risks during the SOSP 2019 Business Meeting. Increasing the number of peer review cycles each year increases reviewing load for community members, at a time when many community members are already expressing concern about load. Increased reviewing load could potentially have a knock-on negative effect on review quality: while SOSP and OSDI are widely regarded today as offering high review quality, other communities within CS that have seen increased reviewing load have reported struggles with review quality. More frequent conferences may unintentionally give rise to community fragmentation: while today the whole systems community assembles for SOSP and OSDI, and devotes its attention to all the work there (and to the students presenting that work), many attendees may not find it feasible to accommodate travel to both SOSP and OSDI every year. To maintain program quality, and thus the quality of conference attendees’ experience, aggregate program size should grow enough to accommodate the community’s strong work (i.e., the caliber of work that today appears in SOSP and OSDI), but not significantly further. We do not profess to know the exact correct aggregate capacity in advance: this provisioning problem appears to be approachable only through search. SIGOPS acknowledges these potential risks raised by the community out of a desire to steward changes to our community’s conferences with care, so that we extract as much of the potential benefit as possible, while mitigating the potential risks.

After careful consideration and discussion of the above pros and cons with USENIX and to allow a gradual transition and experimental approach to changes for our community, the SIGOPS Board and SOSP Steering Committee have decided to keep SOSP biennial during OSDI’s initial experiment with annual frequency. Doing so will allow us to engage with USENIX to gather evidence of the effects of OSDI’s new annual pace before deciding whether SOSP will continue as biennial or move to an annual cadence. We believe this measured, evidence-based approach will best serve our community.

SIGOPS will hold SOSP in 2021 and 2023 at the customary Fall time. We anticipate reaching a decision on whether to move SOSP to annual frequency by early 2023. We welcome further input from the community, and particularly invite more discussion of this topic during the SOSP 2021 business meeting.

With thanks for the community’s positive engagement,
The SIGOPS Board and the SOSP Steering Committee