Policy
The rules Solid Memo's code is based on: which public policy applies (none), which published rules from outside the project the code follows, and the project's own rules, written as data the code checks.
No public policy applies
Solid Memo implements no law, regulation or government policy. It is a volunteer project, not built for or by a public organization, and no public body has asked it to carry out a policy. If that changes, the policy and the code that implements it will be added here.
What the app does with personal data is in privacy.md.
Rules from outside the project
Each of these is a published rule the code follows as written, or with the departures listed. The formats Solid Memo reads and writes are in standards.md.
| Rule | Version | Published at | Implemented in | Checked by |
|---|---|---|---|---|
| SM-2, the SuperMemo 2 scheduling algorithm | 1990 | SuperMemo | sm2.ts (srs.md) | sm2.test.ts, row by row against srs.md's worked examples |
| FSRS-6, the Free Spaced Repetition Scheduler | 6, as fsrs-rs v6.6.2 | Open Spaced Repetition | fsrs.ts (srs.md) | fsrs.test.ts, row by row against srs.md's FSRS-6 worked examples, and against fsrs-rs's published test vectors |
| Unicode CLDR plural rules | the browser's CLDR | Unicode CLDR | Intl.PluralRules in i18n.tsx (i18n.md) | i18n.test.tsx: every plural message gives exactly its language's categories |
| BCP 47 language tags | RFC 5646, RFC 4647 | IETF | languageTag.ts, locale.ts (vocab.md) | languageTag.test.ts, locale.test.ts |
| Unicode normalization form C | UAX #15 | Unicode | the deck library check and search (deck-library.md, search.ts) | deckLibrary.test.ts, search.test.ts |
| DCAT-AP, the EU's profile of DCAT | 3.0.1 | SEMIC | the vendored shapes in packages/vocab/vendor/ (validation.md) | npm run library:check, and the pySHACL cross-check in CI (shacl_crosscheck.py) |
SM-2
Woźniak, P. A. (1990). Optimization of learning. Master's thesis, University of Technology in Poznań. SuperMemo publishes the algorithm's description as Application of a computer to improve the results obtained in working with the SuperMemo method. Solid Memo uses the ease factor and interval formulas as published: an ease factor of 2.5 to start, never below 1.3, intervals of 1 day, 6 days, then the last interval times the ease factor, and a lapse (a grade under 3) starting the card's repetitions again without changing its ease factor. It departs from the description in two recorded ways:
- Intervals are rounded to whole days, and counted in study days, not calendar days (srs.md).
- The description repeats every card graded under 4 at the end of the session until each scores 4 or more. Solid Memo repeats only a card graded 0 or 1 (Again, on the four-button scale), within the session (srs.md). A card graded 2 is a lapse and starts its repetitions again, but is not repeated that session, and a correct but hard answer (3) ends the card's day too.
The answer scales and a course's multiple-choice answers map onto the same grades; they do not change the algorithm (srs.md).
FSRS-6
FSRS-6 is the sixth version of the Free Spaced Repetition Scheduler, by Jarrett Ye and the Open Spaced Repetition project, described in The Algorithm. Solid Memo reimplements its formulas and its 21 default parameters as the reference implementation, fsrs-rs v6.6.2 (BSD-3-Clause), gives them; no code is copied from it. It fixes what FSRS-6 leaves to the app (srs.md):
- The parameters are the defaults, not trained on the learner's answers.
- Days are whole study days, and a second review the same study day takes FSRS-6's same-day formula.
- Intervals are rounded, from 1 day to 36,500, without fuzz.
- A prompt without a current FSRS-6 state has one estimated from its
SM-2 state, as fsrs-rs's
memory_state_from_sm2does at a retention of 90%.
CLDR plural rules
A plural message gives one form for each plural category of its
language ("1 card", "2 cards"). The categories, and which number falls
in which, are the Unicode CLDR plural rules
(by language).
The app does not ship them: it asks the browser's Intl.PluralRules,
so the CLDR version is whichever the browser has. The tests run on
Node.js, which has its own copy of CLDR, and the journeys
(strings.ts) pick each plural
form in the browser they drive, as the app does.
The project's own rules
The project's own rules are written as data, in formats a machine can read, and live in this repository beside the code that applies them.
| Rule | Where it is written | What applies it |
|---|---|---|
| The shape of every document in a pod and in the library | SHACL shapes in ns/shapes/ (shapes.md) | the app in the browser and npm run library:check (rdf-validate-shacl), and pySHACL in CI (validation.md) |
| The meaning of every Solid Memo term | the vocabulary in ns/vocab/ (vocab.md) | the constants generated from it, kept in step by npm run generate:check |
| What a library deck version must state, beyond the shapes: its place, its series, its publisher, the version before it and its license | libraryRules.ts (deck-library.md) | npm run library:check, and the Studio's check for the library |
What the repository's library asks of a deck: a title, description and keywords in every language the app speaks (LIBRARY_LANGUAGES, from LOCALES in locale.ts), the theme EDUC | repoPolicy in curationRules.ts | the Studio's check for the library; npm run library:check for the languages of each deck's newest version, and the library deck shapes for the rest |
| What a page may load and run | the Content Security Policy in apps/web/index.html (markdown.md) | the browser; build.test.ts checks it |
Every published version of a shape or of a deck is frozen: a change is a new version beside it (shapes.md, deck-library.md). So a rule the published shapes do not state, such as that a library deck version names its license, is added as a library rule, not by editing a shape.
No policy outside the project is based on this code. If one is, the code it is based on is the code in this repository.