Vademecum GitHub

5. The Improvement Map and spaced repetition

The Improvement Map is where Vademecum models what you know and suggests where to go next. Everything on it is worked out from what you have already done, on your own device. There are no quotas and no due counts: it suggests, and you decide.

What Vademecum learns from

What you did What it tells Vademecum
Answered a board question You could, or couldn't, produce the answer to a vignette
Answered a flashcard You recalled it, or asked to see it again
Finished a Socratic session How well you explained it, and the gaps it named
Read a page You studied it (study, not yet recall)
Flagged a gap You said you are unsure here
Added an exam report An exam put this area below, at or above the mark

Recent evidence counts for more than old evidence.

Two estimates for each topic

Together these give each topic a state:

Where to go next

At the top of the map, Where to go next (tap to open it) suggests up to five steps. Each one names the topic, its state, why it was chosen, what the estimate rests on, and one thing to do:

Step When
Read the page You haven't read it yet; read once, then test yourself
Board questions Testing yourself fixes knowledge better than rereading
Flashcards It is fading; a quick recall now makes it last
Talk it through You missed questions; explaining why builds the understanding
Add a source You flagged it, but there is no page on it yet

Steps come from different specialties in turn. Mixing topics feels harder, but it helps you tell similar things apart.

Shapes and folded specialties

Each specialty has its own shape, so colour can show what you know. A specialty with three or more topics starts as a single node, with the number of topics inside it. Tap it to open its topics. Each opened specialty gets a Fold button above the graph that closes it again, and Fold all closes every one.

Colour by what you know

Above the graph, switch By specialty to By what you know. Each topic and page node is then coloured by its state. Tap a topic to see its estimate (understood, holding now, how much evidence) and the suggested step.

The same model, elsewhere

The reasoning and references are in ADR 0031.