Computer Science Curricula 2023 (CS2023)

The undergraduate computer-science curriculum guidelines of ACM, IEEE-CS and AAAI, produced by a joint task force and endorsed in sequence: ACM on 18 January 2024, IEEE-CS on 22 January 2024, AAAI on 22 February 2024. Successor to CS2013, which it revises rather than replaces.

This is the designed object the rest of this wiki lacks. Everything else here is a repository — developer-roadmap, coding-interview-university, awesome-system-design-resources — a list that grew by accretion and popularity. CS2023 is a curriculum that a committee argued its way to and that institutions are expected to build degrees against.

What it defines

17 knowledge areas, the structural unit of the guidelines:

Algorithmic Foundations (AL) · Architecture and Organization (AR) · Artificial Intelligence (AI) · Data Management (DM) · Foundations of Programming Languages (FPL) · Graphics and Interactive Techniques (GIT) · Human-Computer Interaction (HCI) · Mathematical and Statistical Foundations (MSF) · Networking and Communication (NC) · Operating Systems (OS) · Parallel and Distributed Computing (PDC) · Security (SEC) · Society, Ethics, and the Profession (SEP) · Software Development Fundamentals (SDF) · Software Engineering (SE) · Specialized Platform Development (SPD) · Systems Fundamentals (SF).

Against CS2013’s 18: Computational Science was dropped as a separate area and the other seventeen were revised. AI, security, parallel and distributed computing, and the professional and ethical material were expanded, and the role of mathematics was enlarged — the guidelines tie that last change to developments in AI.

Structurally, CS2023 pairs the revised knowledge model with a competency framework that a programme customizes: the guidelines are meant to be assembled around the competencies a department chooses to target rather than followed as a fixed syllabus.

What it settles for this wiki, and what it does not

The corpus’s shape has been popularity-selected — the biggest repositories, the ones with the most stars (kamran-ahmed‘s roadmap, coding-interview-university). Reading CS2023’s area list against them shows the selection at work: Society, Ethics and the Profession and Human-Computer Interaction are first-class areas in the guidelines and are close to absent from the repository corpus, which optimizes toward what gets someone hired. interview-preparation is the shape that pressure makes.

The comparison also runs the other way. CS2023 has no equivalent of the corpus’s executable-curriculum thread — material you run rather than read — and nothing about the self-directed disposition self-directed-disposition-study measures. A curriculum specifies content; the repositories specify a practice. Neither document is trying to do the other’s job, and this wiki holds both for the first time.

The tiering, and the hours (added 2026-08-10)

The prescriptive half, read from the report itself. The earlier version of this page recorded the hour counts and tiering as a live gap, on the finding that the report “could not be extracted.” That was wrong about the file, not about the gap: the report is published as fourteen section PDFs, and §3.1 Body of Knowledge and §1.3 Introduction to the Knowledge Model both extract cleanly.

Every topic sits in exactly one of three tiers — the report writes it as a set union, Topic ∈ { CS Core } ∪ { KA Core } ∪ { Non-core }:

• CS Core — “topics that every computer science graduate must know. Every computer science curriculum is expected to cover all the CS Core topics.” This is the binding tier.
• KA Core — “topics recommended for more in-depth study”, per knowledge area.
• Non-core — what CS2013 called electives, renamed.

Over 50% of knowledge units contain CS Core topics and over 75% contain KA Core topics. The hour budget across the 17 areas, 162 knowledge units in total:

270 hours is the whole of what CS2023 actually requires, and that number is the finding. It is roughly a year of contact time; the rest of a degree is the department’s. The report declines to total the KA Core column — printing N/A rather than a sum — because the hours are not additive: SPD’s KA Core is several near-disjoint platform tracks (web, mobile, embedded) that no student takes all of, and MSF’s 145 KA Core hours are a prerequisite pool other areas draw on, not a course. A curriculum picks which areas to take to KA Core depth, and that choice is what the report calls a competency area.

Two things in the table cut against how this wiki’s repositories are organized. MSF at 55 CS Core hours is the single largest requirement — more than SDF’s 43, and the guidelines tie the expanded mathematics directly to AI. And GIT carries 4 CS Core hours against 70 KA Core, the most lopsided ratio in the report: near-optional in the general case, an entire specialism when chosen.

Skill levels are attached per topic — Explain, Apply, Evaluate, Develop, loosely mapped onto revised Bloom’s, with Explain a prerequisite for the other three. So a CS Core topic carries both an hour count and a required depth, which is the part a topic list alone can never express.

What this settles against the repository corpus

The earlier comparison held that the corpus is popularity-selected and under-weights SEP and HCI. The hours make that sharper and partly reverse the emphasis. SEP is 18 CS Core hours, above AI’s 12 — a full-fledged requirement, not a gesture. But the largest single block is mathematics, and that is the requirement developer-roadmap and coding-interview-university most thoroughly omit: both route around MSF almost entirely while covering SDF and AL heavily. The repositories are not merely missing the ethical material. They are missing the guidelines’ largest requirement, and it is the one least visible in a hiring pipeline.

Related

developer-roadmap · coding-interview-university · meta-curriculum · executable-curriculum · self-directed-disposition-study · interview-preparation · iticse · synthesis