Start at module 01 — why anyone built a quantum computer at all — and work down the path. Each module climbs the same six rungs: something you already know, then the new idea as a picture, and only then the symbols. No equation appears before the diagram that gives it meaning.
This is the whole of quantum computing in one control. Two routes reach each outcome; the dial sets the phase between them. Turn it to π and the two arrows to |0⟩ point in opposite directions, add to nothing, and that outcome stops happening entirely. Everything on screen is computed by the same engine that runs your circuits.
A classical probabilistic bit has no phase, so nothing you can do to it will make two routes to the same outcome cancel. That single difference — not "trying every answer at once" — is where quantum speedup comes from.
Challenges are graded by running your circuit and comparing its output distribution to the target within 3%. There is no answer key to memorise — many different correct circuits pass, and the score is how few gates you used.
Write or fix real Qiskit code in the editor — graded the same way, by running what your code actually produces. "Debug" challenges start from code that parses and runs, but produces the wrong answer on purpose.
Graded by execution, not by comparing to a stored sequence: your order is built into a real circuit and run.
Add gates and watch the arrow move. The target ring is the state you are aiming for.
Your line is parsed into a circuit and executed. There is more than one correct answer — anything that produces the target distribution passes.
Every wrong option encodes a specific documented misconception. Which one you pick is recorded — that is what makes this a diagnostic rather than a scoreboard.
Mastery is tracked with Bayesian knowledge tracing: one probability per concept, updated after every attempt, with separate parameters for slipping and guessing. Four numbers per concept, and all of them explainable.
Knowledge fades if it is never revisited — these are concepts you once demonstrated real mastery of, whose score has since drifted down only because time passed, not because you got something wrong.
The prerequisite graph the recommender walks. A node unlocks when everything it depends on reaches 45.
One class, 24 students, six concepts. You appear in the roster live — the row marked "You" reads the same mastery scores the learner view writes. The other 23 are sample data, generated from a fixed seed so the class is stable.