Keeping study time and recall scores separate
A StudyPulse note about timer records, grading responses and reward calculations.
StudyPulse is the study tool I built for my semester. It records study time and recall results separately.
A timer tells me how long a session lasted. It does not tell me whether I can explain the topic. Keeping the two records separate avoids giving mastery credit only for time spent.
Grading and score calculation
A language model can grade rubric points. The server checks the returned values and calculates scores and rewards in code. I can test that calculation without depending on a new grading response each time.
This does not remove grading uncertainty. A model can still judge an answer incorrectly. The checks mainly control which values enter the calculation.
Imported quiz reports need another check. The Worker re-evaluates the report, requires credited topics to quote it and checks that the quoted text exists. It reduces credit when the evaluations disagree substantially.
The limit of a report
Those checks do not prove where an imported report came from. Someone could still invent the report. The feature is useful for personal tracking, but I would not use it as proof of a student’s result in a formal assessment.
The public architecture notes and domain tests explain the current rules. The project page covers the rest of the application.