Let’s GoLearn anything, in the openAboutDocsSign in
← Catalog

About / the point of the tool

A course is an argument, not a shelf of notes.

Let’s Go is a place to read a course somebody wrote, do the work it sets, and have that work marked against what the author was actually asking for. It is not a video library and it is not a wiki: every course here has an author who made choices about what matters and why.

01 / the loop

A course makes a case, then gives you somewhere to try it.

AAuthorchooses the sequence
BLearnerreads the lesson
CWorkanswers or builds
DEvidencefeedback and progress

Anyone can read published material. An account is for enrolling, submitting work, recording progress, and writing a course of your own. That split keeps the front door open without pretending that private progress or a submission can be anonymous.

02 / what is distinctive

The useful constraints are visible.

01

Courses that argue, rather than list

A lesson here is written to make a case for something, the way its author would explain it to somebody in the room, not a bullet list of facts to memorise.

02

Marking that has to cite your own words back to you

A written answer with a rubric is marked by a language model against criteria the author wrote. It is not trusted on its own: every point it claims has to be backed by a quotation from what you actually wrote, and that quotation is checked in code before any mark is kept.

03

Four kinds of work, not one kind of learner

Authors can ask for multiple choice, a written response, a project submission, or runnable code. A course can mix those shapes instead of forcing every subject into a programming exercise.

04

Seven courses, written or ported for this platform

Four language courses (TypeScript, Go, C++, Rust) ported in from earlier tutorials, plus Tech Literacy, Programming Concepts, and Systems Design, written for this workspace.

03 / operating limits

What it does not do yet

The honesty is the point, so here is what to expect and what not to.

  • Multiple choice is graded by a fixed rule, not a model, so it is exact. A written answer with a rubric is graded asynchronously by a model with an evidence check; a missing rubric or an unavailable marker leaves the attempt recorded for a person rather than lost.
  • A project submission is recorded with the artifact types the author requested and is left for a person to mark. A code challenge runs in a sandbox, and only a passing verifier result awards its points.
  • An answer that is genuinely wrong scores zero, the same as an answer the marker could not verify. The two are recorded differently on the server, but a learner sees the same outcome: no marks, and feedback saying why.
  • Installing the app gets you a home-screen icon and keeps lessons you have already read available without a connection. It does not let you take a course offline you have not opened yet, and it does not let you submit work while offline.
  • Publishing a course is the owner’s decision alone. Co-editors can write and revise everything in it, but only the owner can put it in front of learners or take it down.