Logseq/Q/What data does srs-of-matrix.edn store in a Logseq graph?

• What data does srs-of-matrix.edn store in a Logseq graph?

  • AI Answer

    • Short answer: It stores the shared SM-2 interval matrix — a lookup table mapping {repetition-count, easiness-factor} to interval-in-days — used by Logseq's built-in #card spaced repetition system.

    • Structure

      • The file is EDN (Clojure's Extensible Data Notation). Its shape is:
        • edn

          {repetition-count {easiness-factor interval-days, ...}, ...}

      • Outer key (integer 0, 1, 2, …): how many consecutive successful reviews a card has had (the repetition number n in SM-2).
      • Inner key (float e.g. 2.5, 2.36): the easiness factor (EF) for that cell.
      • Inner value (float): the calculated next interval in days.

    • Why it's called a "matrix"

      • SM-2 defines a 2-D lookup table of (n, EF) → interval. As cards are reviewed with varying quality ratings, new (n, EF) cells are populated and the matrix grows. The file is essentially a memoised version of that table, shared across all cards in the graph.

    • Relationship to per-card state

      • There are two separate pieces of state:
        • Per-card (block properties on each #card block): card-repeats::, card-next-schedule::, card-last-interval::, card-ease-factor::. This records this specific card's history — how many times reviewed, current EF, next due date.
        • The matrix (srs-of-matrix.edn): answers "for any card at repetition n with easiness factor EF, what is the next interval?" It does not know or care which card is asking.
      • At review time the two work together:
        • 1. Read this card's n and EF from its block properties.
        • 2. Look up (n, EF) in the matrix → get the interval.
        • 3. After rating, update both: the card's block properties (new n, EF, due date) and the matrix cell for the new (n, EF).
      • Two different cards with the same (n, EF) get the same interval from the matrix. The matrix is a graph-wide learned lookup table, not a per-card record.

    • Per-graph, local-only

      • Each graph has its own logseq/srs-of-matrix.edn. The file in logseq/docs (the official Logseq documentation graph, available at ghq/github.com/logseq/docs) is simply the SRS state for that graph — confirming the file is standard Logseq infrastructure, not custom to this garden.
    • Answer/Official from Logseq source — spaced repetition