Version 1.1
The 2025 edition of USENIX Security will implement a new approach to presenting accepted papers and fostering interactions at the conference. Some accepted papers will be presented as longer talks, tentatively set to 15 minutes; others will be shorter presentations, between one-half and one minute long. Accepted papers will additionally be presented as posters, during thematically organized discussion sessions that will run in parallel with talk sessions. Finally, authors of accepted papers will be invited to upload pre-recorded 15-to-20-minute video presentations, which will be published on the USENIX Security website.
The USENIX Security ‘25 program committee co-chairs and the USENIX Security steering committee welcome feedback from the USENIX Security community about the plans for this new model. We encourage you to share your feedback with us through GitHub issues or anonymously through this Google Form. We will regularly update the FAQ section with concerns and suggestions that are frequently shared or of broad relevance. We invite you to consult the FAQ first, as it may already address your feedback.
We will review comments received by April 22, 2024, and take them into account for the 2025 Call for Papers, which will be available in May 2024. We plan to host an evening session with the community during USENIX Security ’25 to hear from attendees about their experience with the new model.
Rising conference costs and increasing numbers of submitted and accepted papers necessitate USENIX Security shifting to a format with fewer parallel talk session tracks (see the 2023 USENIX Annual Meeting Report). In turn, this implies that the manner of presenting accepted papers needs to change.
In developing the plan for USENIX Security ’25, we strived to make the conference a great experience for both authors of accepted papers and the audience. We sought to achieve as many of the following goals as possible.
- Outreach, engagement, and promotion. Give the authors of every accepted paper the best opportunity to reach an interested audience. Facilitate interaction and discussion between the authors of accepted papers and the audience, and among the authors of accepted papers. Support junior researchers in promoting their research.
- Program quality and attractiveness. Allow the audience to gain comprehensive insight into all the research on topics of interest to them. Support high-quality talks. Allocate time for invited talks. Provide an attractive program to students, junior faculty, senior faculty, and industry.
- Fairness. Give all authors the opportunity to present their research to a broad audience.
- Logistics. Minimize logistics overhead for the audience, authors, and conference organizers during and surrounding the conference.
For USENIX Security 2025 we explored a range of options for paper presentation types and how to assign papers to different types of presentations.
A - Paper presentation types. - The smaller number of parallel tracks devoted to talks presenting accepted papers implies switching to one of the following options for talks, each with advantages and disadvantages:
-
All papers are presented via talks of the same length, but talks shrink to be as short as needed to fit in the available time (e.g., 5 minutes per talk).
- Main advantages:
- All accepted papers are treated equally.
- Every paper is presented orally in front of a large audience.
- Main disadvantages:
- Talks would have to be so short as to make it difficult to communicate technical content.
- Too many short talks in rapid succession may be hard for the audience to process or meaningfully engage with speakers.
- It would be logistically challenging to coordinate the delivery of a session full of talks of this length.
- Main advantages:
-
Only a small subset of accepted papers is presented via talks (e.g., similarly to NeurIPS). (Selection criteria for talks are discussed below.)
- Main advantages:
- Logistically easy to implement.
- Talks can be long enough to communicate technical content.
- Main disadvantages:
- If some papers are presented only via posters and not also as talks, people in the audience who weren’t looking for those specific papers are less likely to encounter them than if the papers were presented via talks. Conference attendees are less likely to learn new, unexpected things.
- Audience loses the ability to get an overview of all accepted work on a topic by going to a talk session (or sessions) on that topic.
- May create the perception of a tiered system, where papers with presentations may be seen as better than papers without presentations.
- Significant pushback in the community against some papers having no talks.
- Main advantages:
-
All papers are presented via talks, but talks are of different lengths. Papers are additionally presented via posters.
- Main advantages:
- All papers are presented as talks.
- Some talks are long enough to communicate technical content.
- Most disadvantages of option (2) are avoided.
- Main disadvantages:
- The majority of talks are too short to communicate technical content.
- Depending on implementation, could create the perception of a tiered system among papers, but to a lesser degree than option (2).
- Logistically more difficult than if only a subset of papers have talks.
- Main advantages:
B - Assigning papers to presentation types. - If not all papers are presented via talks of the same length, a critical decision is how to choose which papers will be presented in which manner. We considered several options.
- Multiple tiers of acceptance: whether papers are chosen to be presented via longer talks or via shorter talks is an editorial decision (made by, e.g., reviewers or PC chairs).
- Main advantages:
- Could lead to a more exciting program for the audience, higher quality talks, and papers deemed as notable by the program committee could be guaranteed to be presented via longer talks.
- Similar to the model used in the ML/AI community, e.g., at NeurIPS, and hence potentially familiar to the audience and to authors.
- Main disadvantages:
- Unfair: potentially biased in favor of better-known authors, “uncool” but scientifically valuable topics, etc.
- Likely to create the perception that papers presented as longer talks are better or more valuable than papers presented as shorter talks.
- Main advantages:
- Papers are randomly selected to be presented in one of two talk formats.
- Main advantages:
- Fairness. All papers, regardless of topic, seniority of authors, etc., have a chance to be selected for longer talks.
- Main disadvantages:
- Particularly excellent papers may not get longer talks, which could be detrimental to scientific discourse or harm the audience’s experience.
- Main advantages:
- Except for distinguished papers (i.e., those winning distinguished paper awards), all longer talks are selected at random from among all accepted papers; distinguished papers are guaranteed longer talks.
- Main advantages:
- Ensures papers that the PC identified as noteworthy are presented via longer talks.
- Fairer than other options that involve two talk lengths.
- Reduces perception that there are two clear tiers of papers compared to option 1, in particular, because of the element of randomness in choosing which papers are assigned to which talk length.
- Main disadvantages:
- May not create as exciting a program as option 1.
- Main advantages:
Weighing advantages against disadvantages, USENIX Security ‘25 plans to implement options A-3 and B-3: Each accepted paper will have a talk slot, but some papers get longer talks (tentatively set at 15 minutes) and some get shorter talks (~1 minute). (The two talk lengths will be finalized after we know how many papers are accepted.) Papers that win distinguished paper awards will be assigned to longer talks; remaining longer talks will be selected at random from among all other accepted papers. This option ensures that all papers have some probability to be presented as longer talks even if they don’t win distinguished paper awards; at the same time, distinguished papers are guaranteed longer talk slots, increasing the value of the conference to the scientific community and to the audience.
Discussion tracks. Additionally, all papers will be presented as posters in discussion tracks that will run in parallel with the talk tracks. All papers presented during a specific talk session, whether as shorter or longer talks, will also be grouped and presented collocated with each other in poster format during a discussion track session. We envision poster sessions taking place in parallel with talk sessions (though timed such that talks and discussions on the same topic do not conflict). This gives audience members multiple opportunities for how to engage with the papers presented at the conference, and we hope that it will provide an opportunity for deeper engagement between the audience and paper authors. We will also introduce awards to recognize papers whose presentations in poster form were particularly notable, as determined by aggregate feedback from the community collected at the conference.
Online videos. Also, all paper authors will be invited to submit a 15-to-20-minute video presenting their paper that will be posted on the USENIX Security ’25 website after the conference. In addition to enabling later viewing of paper presentations, this will allow anyone interested to watch a longer presentation of any paper that was assigned to be presented via a shorter talk rather than a longer talk.
Several additional options were suggested by the community as potential alternatives to the format described above. These options are briefly described below.
- Split USENIX Security into multiple conferences geographically or thematically.
- Pros: less travel time or more focused, smaller events.
- Cons:
- Not immediately feasible organizationally.
- The community will be partitioned and subcommunities might be more isolated.
- May reduce outreach, engagement, and promotion opportunities, especially for junior students.
- Join top conferences into fewer conferences.
- Cons:
- Not immediately feasible organizationally
- Doesn’t solve challenges that are forcing current change in conference format.
- Cons:
- Accept papers without presentation (and don’t require attendance at conference).
- Pros:
- Removes the problem of how to fit all accepted papers into talks.
- Allows authors who cannot attend the conference to nevertheless submit papers.
- Cons:
- Without some aspect of presentation, there is no conference. Many find talks valuable.
- Without mandatory presentations, many authors may be prevented from attending the conference (e.g., because they may have a harder time justifying the cost of attendance), and would lose the opportunity to network and discuss their work.
- Pros:
We encourage you to share your feedback with us through GitHub issues or anonymously through this Google Form. We will regularly update the FAQ section with concerns and suggestions that are frequently shared or of broad relevance. We invite you to consult the FAQ first, as it may already address your feedback.
A: The plan was informed by extensive discussions that involved the USENIX Security steering committee, the USENIX Security ‘25 program co-chairs, program chairs and co-chairs of other top security conferences, and organizers of conferences in other fields. Based on the data collected, the program co-chairs proposed an initial plan, which was then iteratively improved via feedback from the USENIX Security steering committee and early feedback from the USENIX Security community (i.e., potential attendees). The revised plan was approved by the USENIX Security steering committee.
A: At least one author per paper must participate in the discussion track, in person. The possibility for some authors to opt out of providing a talk or poster, with approval from the organizers, is under discussion.
Q: Will papers be selected uniformly at random for presentation via longer talks, or will a different type of random choice be used?
A: This is under discussion. Selection may be uniformly random or weighted random, e.g., based on audience interest in topics during the previous iteration of the conference. If we do use weighting, we will ensure that the weighting doesn’t significantly reduce the chance of any one paper being selected for a longer talk compared to uniform random selection.
Q: If a paper was submitted to the 2024 submission timeline and received a major revision decision and is accepted to 2025, what presentation rules and format will it fall under?
A: Such a paper would fall under 2025 rules for presentation.
A: We are considering whether it is logistically feasible for USENIX Security to support a combination of live and recorded shorter talks. If recorded shorter talks are supported, authors will still be required to present their paper via posters in person.
Q: How will USENIX Security distinguish between posters (for accepted papers) and posters (for works-in-progress works and other non-peer-reviewed posters)?
A: The details are still under discussion, but our goal is to keep these clearly separate in both reality and perception. Posters representing unpublished work may be required to include “Work in progress:” in the title, for example.
Q: What specific problem is the current conference format proposal trying to solve? Is it necessary to change? Could you explore conference format options that are fundamentally different?
A: Several factors drove the need for a new conference format, including that scaling the traditional format to match the number of accepted papers isn’t financially feasible. A particular challenge, both financially and logistically, with the traditional format is the large number of paper presentation tracks.
Here’s the definition of the immediate problem the new conference format is trying to solve:
Supposing that USENIX Security ’25 will have:
- three technical talk tracks over three days,
- a paper acceptance rate similar to that of recent previous conferences, and
- a format as similar to that of recent previous conferences as #1 and #2 permit,
How do we arrange the presentation of papers to fit the available time while keeping in mind considerations and constraints related to outreach, fairness, quality, and logistics?
Suppositions 1-3 are arrived at by a number of factors, including having to do with finances, mitigating the risks of conference formats that haven’t been tried or that USENIX has less experience supporting, the timelines for contracting space and releasing the call for papers, and other resource constraints.
We welcome suggestions about how to change the conference format more comprehensively, including what the goals of conferences should be, what specific events and activities should take place at the conference, etc. A fundamentally different conference format is unlikely to be implemented for USENIX Security ’25, but novel ideas and proposals could help shape future conferences.
| Version | Date | Description |
|---|---|---|
| 1.0 | Apr 2, 2024 | Release |
| 1.1 | Apr 9, 2024 | Added item to FAQ |
- The 2023 USENIX Annual Meeting Report: https://www.usenix.org/blog/2023-usenix-annual-meeting-report