03 / Plan the next pass
Read a practice record without overselling it
Separate a completed pass, your own labels and a later recall check. Keep the limits visible when deciding what to practice next.
Finishing a deck produces a record of actions and judgments. It does not produce an objective learning score. That distinction matters most when a tidy summary makes a small exercise look more conclusive than it is.
Use a practice record to identify prompts worth checking and to preserve the conditions of a pass. A useful record can show both what happened and what it cannot establish.
Read the denominator first
A fictional three-card deck might have two recorded assessments because the person stopped before the final card. The summary should say two of three recorded and mark the session incomplete. The missing card has no assessment; it should not be treated as remembered, forgotten or failed.
The Recall Deck reports counts for Remembered, Partly recalled and Needs another look. It does not convert them into a mastery percentage or compare one person with another. Exported records also say whether the pass was complete.
Keep the original attempt beside the label
A label is a quick index into the record. To understand it, inspect the prompt, the text written before reveal, the reference and the comparison note. An answer marked Remembered with no written attempt tells you less about the original response than a preserved attempt does.
People can judge their own recall inaccurately. In a foreign-language vocabulary experiment, Karpicke and Roediger found that participants’ predictions did not track their later performance. That is a reason to avoid treating confidence or our self-assessment labels as objective measurements; it is not a diagnosis of an individual reader.
Sources: Karpicke and Roediger (2008): The Critical Importance of Retrieval for Learning
Separate an immediate repeat from a later check
Right after revealing “drying tray beside the window,” repeating those words may be straightforward. That immediate response and an attempt after a delay occur under different conditions. Give each pass its own record rather than presenting the later-looking total as a timeless statement about memory.
The 2008 vocabulary experiment found benefits from repeated retrieval on a delayed test under its study conditions. It does not provide a universal schedule for every deck. The tool deliberately makes no scheduling recommendation and sends no reminders; decide when to return based on the task and reliable guidance for that context.
Sources: Karpicke and Roediger (2008): The Critical Importance of Retrieval for Learning
Record changes that make passes different
Before comparing two exports, check whether the questions, answers or available cues changed. Also note whether the source was open, whether answers were written or spoken, and whether a pass followed immediately after viewing the deck. These are practical record-keeping questions, not an attempt to turn a personal exercise into a controlled study.
In the fictional museum example, rewriting “What about blue?” into a precise shelf-label question changes what the cue provides. A later, more favorable self-assessment could reflect that edit. Preserve the revised deck title and describe the change in your own notes.
For skills that require doing something, include an appropriate task outside the deck. Recalling the parts of a paper model is different from building it. CMU distinguishes practice of component skills from combining those skills and selecting when to use them.
Choose a modest conclusion and a next step
A supported conclusion is “I marked two of three cards Remembered during this pass.” Another is “The model-card question needs a clearer reference.” Neither statement claims that you mastered a subject, improved your general memory or will receive a particular exam result.
Use the downloadable record sheet for conditions, gaps and your next check. Keep filled copies in your own storage. To return to the exercise, open the Recall Deck, import a saved JSON deck and start a new pass. Previous Markdown results remain separate files; they are not imported or silently merged.
Reproduce a label that is not a score
Use the fictional paper-museum starter deck for this deliberate mismatch exercise. On the first card, write Every model may be handled. Reveal the reference: a blue label marks a model that visitors may handle. The attempt has broadened a conditional rule into a rule for every model. For this demonstration only, choose Remembered anyway and add the note Deliberately mismatched label for this fictional check. This creates a record worth questioning, not a recommended assessment.
On the second card, leave the written attempt blank, reveal the reference and choose Needs another look. Record No written attempt; check the drying-tray rule. Stop before assessing the third card and export the session notes. The third card remains unrecorded. A missing written attempt also cannot tell a later reader whether you answered silently, guessed or did not attempt the prompt.
The direct session-function check on September 12 produced two recorded cards out of three, with one Remembered and one Needs another look. The Markdown export preserved the first attempt beside its conflicting label and rendered the second attempt as (No written attempt). It marked the session incomplete. The tool did not inspect answer meaning or override the selected label.
Use that contradiction to choose the next action: review the blue-label rule against the reference, then make a separate pass if useful. Keep this deliberately mismatched record separate from ordinary practice notes. These observations verify the session and export logic for supplied inputs; they are not a browser interaction test, participant study or evidence of learning.
Continue with Recall Deck, then Download the working sheet.
| Record detail | Observed result |
|---|---|
| Recorded cards | 2 of 3; session incomplete |
| Self-assessment totals | Remembered 1; Partly recalled 0; Needs another look 1 |
| First original attempt | Every model may be handled. |
| First stored label | Remembered; no automatic correction |
| Second attempt in exported notes | (No written attempt) |
| Third card | No assessment recorded |