Understanding CS Conference Tiers

Why Computer Science Publishes at Conferences

In almost every academic discipline, a researcher's output is judged by the journals they publish in. Biologists, chemists, economists, and psychologists all operate within a well-understood hierarchy of journals ranked Q1 through Q4 based on citation impact. A Q1 paper in Nature or Cell signals something. A Q4 paper signals something else. The system is imperfect, but it is legible.

Computer science does not work this way. It never really has. The field moves too fast for the traditional journal model, where a paper might sit in review for one to three years before reaching readers. By the time a deep learning paper clears journal peer review, the technique it describes may already be considered foundational and surpassed. So the field converged on a different primary venue: peer-reviewed conferences.

Top CS conferences accept submissions, distribute them to expert reviewers, and notify authors within a few months. Accepted work is presented in person, published in the conference proceedings, and immediately enters the research community's conversation. The result is that conferences became the prestige layer in CS, not journals. A paper accepted at NeurIPS carries more weight than one published in most CS journals. This is not informal consensus. It is the documented standard for tenure decisions, grant applications, and faculty hiring at every major research university in the world.

The Scale of Competition

To understand why a publication at a top CS conference is meaningful, you need to understand who submits to these venues and how many are turned away.

A submission to NeurIPS, CVPR, or ICML is not a high school science fair entry evaluated by a rotating panel of volunteer judges. It is a full research paper reviewed by three to five domain experts, each of whom has themselves published multiple times at the same venue, typically with h-indexes above 10 and years of graduate training behind them. The pool of submitters consists almost entirely of PhD students at institutions like MIT, Stanford, CMU, Berkeley, ETH Zurich, and Oxford, alongside faculty and researchers at Google DeepMind, OpenAI, Meta AI, and Microsoft Research.

Even at a 20 to 27 percent acceptance rate, consider the baseline: you are measuring against researchers who spent five to seven years in graduate training specifically to produce this kind of work. The acceptance rate is not the full picture of selectivity. The pool itself is the story.

For a high school student to contribute to work accepted at any CORE-ranked CS conference, let alone an A or A* venue, is categorically unusual. It is not a higher version of something familiar. It is a different kind of achievement entirely.

How This Compares to Everything Else at the High School Level

High school research competitions are genuinely prestigious, and reaching their finals is a remarkable accomplishment. But they operate on different criteria, with different evaluators, for a different purpose. The comparison is instructive.

This is not a critique of science fairs or olympiads. Reaching the ISEF finals from a pool of nine million students, or qualifying for USAMO from three hundred thousand AMC participants, reflects real ability and dedication. Those achievements matter.

The distinction is one of kind. High school competitions are designed for high school students. They apply standards calibrated to that population and evaluate work in that context. A CS conference does not know or care how old you are. Your paper is reviewed by the same people reviewing a Stanford professor's paper, using the same criteria. There is no separate category for young researchers. You either clear the bar or you do not.

The CORE Ranking System

With thousands of CS conferences across every subfield, the research community needed a way to distinguish between them. The Computing Research and Education Association of Australasia (CORE) maintains the most widely used ranking framework. It assigns one of four tiers to each conference: A*, A, B, or C. These tiers are not determined by acceptance rate alone. They reflect visibility, citation impact, the caliber of the program committee, the rigor of the review process, and how the broader research community actually uses and values the venue.

The rankings are publicly searchable. Any conference you encounter can be looked up by name. When evaluating a research experience, a publication claim, or a lab's output, the CORE tier of the venue is the most direct indicator of what the work actually went through to be accepted.

A* designates the flagship conference for a given area: the top venue, or one of a small group of equal-top venues, in that specific field of CS. These are not simply conferences with high standards. They are the conferences that define the standards. The research community at large treats an A* publication as a primary career credential.

Researchers who reach career milestones (tenure review, major grant applications, prestigious faculty positions) are expected to list their A* publications prominently. Being invited to give a keynote at an A* conference is considered a significant personal achievement even for established professors. Being invited to chair the program committee requires an h-index of 25 or above and a long track record of publishing at the venue.

The review process at A* venues is extensive. Submissions are distributed to reviewers who are themselves active researchers with publication histories in the conference. The period between submission deadline and notification spans multiple months. Reviews routinely run several hundred words with detailed technical objections. Papers are often revised through a rebuttal process before a final decision.

An A conference shares nearly all the characteristics of an A* venue. The papers published there are of comparable quality, the reviewers are domain experts, and strong researchers actively target these venues. The distinction from A* is typically one of breadth and visibility: an A conference may serve a narrower subfield, be less well known outside its immediate community, or have a shorter history than the flagship venues at the top tier.

Research leaders routinely submit their best work to A conferences. Some of the most technically significant papers in any given year appear in A venues rather than A* ones, often because the work fits the specific conference's focus better. A conference papers are explicitly favored by researchers compiling career highlights for grant applications and faculty promotions.

Acceptance rates may be marginally higher than at A* venues, but the pool of submitters remains the same. The review process is equally rigorous, conducted by people who have published in the area and can evaluate the work at a technical level. Being accepted at an A conference is a strong research result by any professional standard.

CORE's B tier is intentionally broad. Some B conferences sit very close to A level and would be considered strong publication venues by most researchers in the area. Others are weaker, with lower average paper quality and less rigorous review processes. The tier designation alone is not enough to evaluate a specific B conference without looking at the specific venue.

Leading researchers do publish at B conferences, but typically for specific reasons: proximity to a particular subfield, travel logistics, or supporting a graduate student's career development. They are unlikely to attend a B conference as the sole purpose of an international trip. Citations to B conference papers tend to be mid-range rather than field-defining.

For early-career researchers, B conferences represent legitimate publication venues that demonstrate the ability to complete and submit research. But within the field, they are understood to represent a different level of scrutiny and community engagement than A or A* venues.

C conferences are genuine academic venues. Papers are reviewed by academics prior to acceptance, which distinguishes them from predatory conferences and unreviewed workshops. But the standard applied is "acceptable" rather than "excellent," and individual papers within a C conference proceedings can vary widely in quality. Reviews are often brief and may not engage with the technical content in depth.

Some papers in C conference proceedings would not have cleared the A or B bar but represent valid work in a narrow subfield where no stronger venue exists. Others are there because the authors, often developing researchers, needed a publication record and accepted a lower-visibility venue. Leading researchers do not typically list C conference papers in career highlights and may not disclose rejections from them.

C conferences serve a real role in developing sub-communities and giving early-career researchers a publication pathway. But they should not be confused with the upper tiers when evaluating the significance of a research experience.

What to Do With This Information

When you encounter a research experience, a lab's publication list, or a claimed result, look up the conference in the CORE portal. A venue's name alone tells you very little. Its CORE tier tells you what the work actually went through.

Beyond CORE tiers, note that some venues are tagged as unranked (insufficient data to assess), unlisted (not primarily CS, or too small to have been submitted for review), or national/regional (held within a single country, typically on par with B or C venues). A conference that appears in proceedings but cannot be found in the CORE portal at all warrants scrutiny.

The goal of understanding this system is not to assign hierarchy for its own sake. It is to give you an honest picture of what original research output actually represents. There is a large and meaningful gap between completing a research project and having that project accepted by the community of researchers who work in that area for a living. The CORE tier system is how that community communicates what it has accepted.