A1 — Explain It Without the Explainer
Learn something unfamiliar, close every source, reconstruct it from memory, then test the reconstruction against the authority you closed.
Due: 29 March 2027 · Weight: 30%
The brief#
Pick something you don’t yet understand, learn it with the source open, then find out what you actually kept once it closed.
This is a self-measurement task, not a research report. Choose material narrow enough to reconstruct in full inside the time you have — a mechanism, a procedure, an argument — and unfamiliar enough that you can’t already recite it. Read or watch with the source open, close it, and write your explanation from memory alone. Then reopen the source and mark your own reconstruction against it, in public, in the submission itself.
The mark isn’t for getting it right the first time. It’s for locating, precisely, where the reconstruction diverged from the source, and for the corrected explanation you produce once you’ve seen the gap.
What you submit#
A single document with three parts: the reconstruction (written with the source closed, timestamped or otherwise evidenced as such), the comparison against the reopened source, and the corrected explanation. Note what your instrument — this exercise — could and couldn’t tell you about whether you understood the material, as distinct from whether you remembered it.
The spec
The brief poses the problem and leaves room for a response. The spec is the fixed part: what a marker checks. Some lines can be checked mechanically; the rest call for judgement.
- a source consulted while it was open, and a plain record of when it closed
- a reconstruction produced with the source closed, not edited afterward
- a comparison against the source that names at least one specific place the reconstruction was wrong, not just that it was "mostly right"
- a short corrected explanation, written after seeing where it broke
How it is marked
| Criterion | Weight |
|---|---|
| Reconstruction accuracy against the closed source | 40% |
| Candour about the weak spot | 30% |
| Quality of the corrected model | 30% |