About Corpus

Philosophy

Corpus is organized around one belief: some skills decay fast, and others barely decay at all. “How to structure a technical argument” ages differently than “which prompt pattern gets the best result from this month’s model.” Treating both the same way — mixing durable technique with whatever tool happens to be current — makes the durable material harder to trust and the timely material harder to keep honest. So the site is split: Human Skills hold what stays true independent of any particular tool; the tool-dependent material, genuinely tied to the current moment and explicitly expected to need refreshing, comes in two forms — AI-Assisted Workflows, which a human follows themselves with an AI tool’s help, and Agent Skills, downloadable files handed to an agent that does the work directly. (The practical version of this idea — how to size up any specific thing you’re learning — lives as its own technique on Knowledge Half-Life.)

That architectural choice reflects a broader philosophy the course operated on for years, worth stating plainly here:

Guidance is descriptive, not prescriptive. Assignments were rarely spelled out in exact detail. Figuring out how to define and complete the work was part of the point — closer to professional life than a rubric-driven classroom.
No minimum page or word counts — the opposite, if anything. The class motto was “enough words, no more” — originally “Use enough words, no more,” trimmed down over the years because apparently even the motto had to follow its own advice :) In professional life, almost everything you write is capped, not padded.
Topics get repeated and refined, not written once and abandoned. Writing and presenting the same material several times across a semester builds real depth, and rewards the discipline to keep improving something rather than moving on the moment it’s “done.”
You own your topic. The specifics of what to study and how to study it were the student’s responsibility, not a professor’s script — mirroring the latitude (and responsibility) that grows with seniority in any professional role.
A good topic is a real intellectual challenge, and that’s the point, not a flaw. Struggling with material that takes real time and effort to understand is exactly what building expertise looks like.
Content gets packaged differently for different audiences. A presentation, an abstract, and a full paper aren’t the same content resized — each has a different job.
Self-learning is the actual long game. The habits that matter most are the ones that keep working after graduation, when there’s no more syllabus telling you what to learn next.
New tools create new problems along with new solutions — expect that, and stay curious anyway. A growth mindset toward new tools isn’t optional in a field that keeps producing them.

None of this is really about computer science specifically. It’s a philosophy of how to keep getting better at something after the structured part of learning it ends — which is exactly what Corpus is trying to preserve.