Courses

Courses I have built to learn a topic properly. Some track your progress as you work through them; sign in on a course to carry that progress between devices.

Alongside the chess courses there is a chess glossary: the vocabulary they use, with the words you have to see marked as such, so they are not mistaken for words you can simply look up.

  • Instrument Log

    Classical Psychometrics for Eval Rubrics: test theory, the kappa family, ICC and generalizability theory — measurement thinking applied to eval rubrics.

    Progress saved

  • Probabilistic Record Linkage

    Deciding when two records describe the same thing, built on Fellegi & Sunter (JASA, 1969).

    Progress saved

  • Context Graphs

    How context graphs connect people to the priorities that matter. A crash course in systems and data modeling.

    Reading

  • Teeline Shorthand to 100

    Six weeks to take a passage in Teeline and transcribe it accurately twenty- four hours later.

    Progress saved

  • Copying Morse

    Learn to copy an unfamiliar plain-text transmission in Morse code onto paper.

    Progress saved

  • Sending Morse

    Advance to sending messages with Morse code via sound or light.

    Progress saved

  • Graphical Perception

    Which visual encodings a reader actually decodes accurately, from Cleveland & McGill (JASA, 1984) to a chart you can defend with measurements you took yourself.

    Progress saved

  • Color Science

    OKLCH, display gamut, and why the 4.5:1 your linter enforces is a model rather than a measurement. Ends in a token palette and a contrast policy you can defend.

    Progress saved

  • Map Projections

    Flattening a sphere is impossible by Gauss's theorem, not by clumsiness, so every map is a choice about which invariant to break. Tissot's indicatrix makes the cost computable; the deliverable is a projection decision memo.

    Progress saved

  • Task Abstraction

    Not whether a mark can be read but whether it should have been drawn at all, one level upstream of graphical perception. Brehmer & Munzner's why/what/how typology (2013) and the argument around it; the artifact is a task abstraction memo.

    Progress saved

  • Chess: Board Vision

    The four questions a position asks before tactics do — what square is that, what does that piece attack, how many pieces hit it, and what is undefended — each drilled alone until it is automatic, then shuffled, because a game never tells you which one it wants. Five units, every one gated at eighteen of twenty held-out positions inside their cap.

    Progress saved

  • Seeds of tactical destruction — scanning a position

    Board vision's sequel. Heisman's twelve seeds of tactical destruction — loose pieces, overworked defenders, an exposed king — named from memory under a cap, then read as evidence rather than an answer: four or more present is 85% tactical, two or three is the same as not looking. Gated at eighteen of twenty, except the calibration block, which cannot be: the features top out at 78%, against 29% for guessing quiet.

    Progress saved