cs-video-courses: three markdown files, no licence, and a table of contents where machine learning takes a quarter of the headings

The whole project is README.md, NOTES.md and CONTRIBUTING.md

There is no code here. The top-level listing is three markdown files, and the repository description is a list of computer science courses with video lectures. GitHub reports no primary language and no licence, and the repository publishes no releases.

The README delegates its own explanation. It asks you to check NOTES.md for general information about the list, and to read CONTRIBUTING.md for the contribution guidelines, which means the rules that govern the collection live in the two files that are not the collection.

Consequence for the reader: cloning this gives you a document to read, not a tool to run, and there is nothing to install, configure, or update. The last push was on 2026-09-25, so the list is still being edited, but the edit is a text change and the only way to consume an update is to read the diff.

An entry carries a course code, an institution, and a year that may describe the recording

The entries follow a house format. COMP6991 Rust Programming UNSW (2022), CS 10 The Beauty and Joy of Computing Spring 2015 Dan Garcia UC Berkeley InfoCoBuild, 6.0001 Introduction to Computer Science and Programming in Python MIT OCW, and CSE 142 Computer Programming I Java Programming Spring 2016 University of Washington.

The year in that line is ambiguous, and the ambiguity is the whole problem with a list like this. It can be the year the course ran, the year the videos were recorded, or the year the page was last published. Nothing in the entry says which.

Consequence for the reader: a 2015 recording of a course that is still taught is a different artefact from a curriculum that has since changed, and the two look identical in the list. If you are trying to match a course to a current syllabus, the year is the field you have to interpret, and interpreting it is left to you.

Some entries link a course page, some a playlist, and the pattern is not consistent

The link structure varies entry by entry. CS 106A Programming Methodology Stanford carries two links, a course page and a separate Lecture Videos playlist. CS 50R Introduction to Programming with R 2024 carries its own site and a playlist. IN2377 Concepts of C++ programming at TUM carries four, the current term plus Winter 2022, Summer 2022 and Summer 2021, all pointing at the same streaming host with different year and term parameters.

Others carry one. CS 61A links only cs61a.org. 6.001 Structure and Interpretation of Computer Programs links only the MIT OCW video lectures. CS 10 links only the Berkeley InfoCoBuild page.

Consequence for the reader: a single link does not tell you whether you are getting the lectures or only the syllabus, so a dead course page can take the playlist down with it and a live course page can hide a removed video. The four-way TUM entry is the clearest case of the maintenance cost, because keeping four terms alive is four things to re-check rather than one.

Machine learning takes twelve of the thirty table-of-contents headings

The table of contents has thirty top-level entries, and machine learning is the only one that opens into a dozen sub-headings: Introduction to Machine Learning, Data Mining, Probabilistic Graphical Modeling, Deep Learning, Reinforcement Learning, Advanced Machine Learning, Natural Language Processing, Generative AI and LLMs, Computer Vision, Time Series Analysis, Optimization, Unsupervised Learning, and Misc Machine Learning Topics.

Artificial Intelligence and Image Processing and Computer Vision each sit as a single top-level entry, and Computer Vision appears twice in the contents, once under machine learning and once as its own section. Systems Programming and Software Engineering are the other two that branch, into three and five sub-headings respectively.

Consequence for the reader: the taxonomy reflects where the collection grew rather than how computer science divides, and a duplicated Computer Vision heading is the visible symptom. Anyone using the contents as a map of the field will misjudge the balance, and the duplication means a reader looking for computer vision has two places to check and no indication which is fuller.

The contribution rule is a level threshold, written in one sentence

The README states the bar directly: contributors should create a pull request for actual college or university level courses, and should not add links for small MOOCs, basic tutorials, or advertisements for some sites and channels.

Three categories are excluded and one is required. That is the whole policy, and it is enforced through review rather than through a checklist in CONTRIBUTING.md that the README summarises.

Consequence for the reader: a course that is genuinely college-level but has no video lectures is out of scope, because the list is defined by having lectures, and a course that has lectures but is a beginner tutorial is out of scope for the opposite reason. The rule also excludes promotional channels by name category, so a link that is a genuine lecture series can be rejected for looking like an advertisement. Expect a pull request to be argued about on level rather than accepted on merit.

The issue rule is about contribution farming, and it is phrased as a threat

The second policy sentence is about issues rather than pull requests. The README invites genuine issues, and then says it has been noticed that few people open empty issues to raise their GitHub contribution on their account, and that such spammers will be blocked.

Consequence for the reader: the maintainer is telling you what the moderation policy is and why it exists, in the same paragraph as the invitation. A reporter who opens an issue titled something like a course typo, with nothing in the body, has been pre-classified. That is a deliberate low-cost filter rather than a review queue, and it means a real bug report in a dead link has to look like a real bug report to get past the first read.

No licence file means no stated terms for reusing the compilation

The top level holds three files, and none of them is a licence. GitHub identifies no licence for the repository, and the README carries no licence paragraph, unlike almost every other project in this batch.

Consequence for the reader: the individual course links belong to the universities that host them and are not the maintainer's to relicense, which is the obvious reason a list like this carries no grant. But the compilation itself, the selection, the categorisation, and the annotations are also unlicensed, and that is a gap rather than a decision anyone made. If you want to mirror this list on an internal wiki or ship it inside a product, nothing in the repository tells you whether you may, and the safe reading of an unlicensed document is that you may not redistribute it.

The NOTES.md file is the place to look for any terms, and the README does not summarise what is in it.

Editorial conclusion

Use this list as a starting index when you want university courses with recorded lectures, and expect to do your own currency check on every link you follow. Do not treat it as a syllabus, and do not mirror it: the repository has no LICENSE file and GitHub identifies no licence, so the terms for reusing the compilation are unstated. Before you plan a course from it, check three things: whether the year in an entry describes the recording or the curriculum, because entries carry both and the two drift apart, whether the course page or the playlist is the link that survives, and whether the topic you need is one of the twenty or so headings, since a field missing from the table of contents is a field the maintainers are not covering.

Frequently asked questions

Official sources

1. Official README
2. Project repository