CS2023 Curricula Guidelines and Computer Science Accreditation

Michael Oudshoorn, Webb School of Engineering, High Point University. A conference paper hosted on ACM's own csed.acm.org — the CS2023 project's site — asking what the new guidelines do to accreditation. Read locally from the PDF.

Growth edge 1's successor asked whether anybody checks that a department covers the CS Core. The answer this paper gives is structural, and better than a survey would have been: the body that could check does not ask for it.

The hour ledger, CS2013 against CS2023

The paper's Table I compares core hours per knowledge area. CS2013's figure is all tier-1 hours plus 80% of tier-2; CS2023's is CS-Core.

Nine and a half hours across a decade is close to nothing, which is the paper's own framing: the changes look minor in the table and are substantial inside four knowledge areas (mathematical and statistical foundations, AI, security, and society/ethics). Software Engineering losing 17.6 of 23.6 core hours while Software Development Fundamentals holds 43 is the sharpest single move, and the paper does not dwell on it.

The mathematics rewrite has a number behind it: a faculty survey rated probability, statistics and linear algebra at 98%, 98% and 89% importance for industry and graduate school combined, which is what displaced "Calculus I is necessary" as the stated requirement.

What ABET actually requires, and why it settles the edge

The 2024 computer-science criteria demand at least 40 semester credit hours of computer science: substantial coverage of algorithms and complexity, theory, programming-language concepts and software development; exposure to architecture, information management, networking, operating systems and parallel/distributed computing; study of computing-based systems at varying levels of abstraction; and a major integrative project. Mathematics is at least 15 credit hours at Calculus I rigour, including discrete mathematics, statistics and probability. There the paper judges CS2023 and ABET "closely aligned."

Then the sentence the edge was looking for:

"The computer science topics identified in the 2024 accreditation criteria are a subset of those identified in CS2013."

Not CS2023. CS2013, the guidelines CS2023 replaced. The accreditor is a decade behind the guidance and asks for less than it. Named as present in CS2023's core and absent from the criteria: artificial intelligence (12 CS-Core hours), graphics and interactive techniques (4), human-computer interaction (8), software engineering (6). A department can hold current ABET accreditation for a computer science degree while teaching no AI at all.

The paper gives the reasons without defending them: the criteria have to be satisfiable by community colleges and liberal-arts schools, they cannot swing year to year because curriculum committees move slowly, and flexibility is deliberate. It also notes that ABET does not cover dispositions or competencies, CS2023's other half, and that the five or six student outcomes are the closest thing, deliberately coarse to keep assessment affordable.

Why the alignment cannot catch up quickly

A criteria change takes at least two years: the commission votes twice on any substantive change, and the proposal must be open for comment first so institutions can adapt before their next visit. So CS2023 has no immediate effect and, in the paper's words, "conversations are already underway." The lag is procedural rather than an oversight, which means it is permanent — every guideline revision arrives at an accreditor that is still enforcing the previous one.

What it does to this spoke

abet-computing-criteria holds the rules and cs2023 holds the advice; this page holds the gap between them, measured. It closes edge 1's successor in the direction the edge did not anticipate. The edge asked for a syllabus traced against the CS Core, or an accreditation review reporting conformance. Neither exists because neither is required: ABET checks a subset of the previous decade's guidelines, so conformance to the CS Core is not a thing any program is assessed on and not a thing any program has reason to publish. Corrected 2026-10-05: one does exist. curriculum-alignment-uaeu (arXiv 2606.19469, June 2026) traces an ABET-accredited BSc's 41 courses against CS2023 and finds 48.4% of its knowledge units covered. The cause this page gives still holds (no program is assessed on the CS Core), but a department can publish its conformance anyway, and one has.

That reframes the spoke's standing evidence problem rather than solving it. cirr-standards found the bootcamp sector with a good instrument and three participants; mooc-pivot found a university consortium that measured itself and published a collapse. This is a third shape: an instrument that exists, is mandatory, is used on hundreds of programs — and measures something narrower than the thing the field says it wants taught. The gap is not missing data. It is data about a different question.

Related

cs2023 · abet-computing-criteria · curriculum-alignment-uaeu · cirr-standards · mooc-pivot · executable-curriculum · meta-curriculum · synthesis