The claim
Requests in the notation and engraving theme did not receive more total reactions on issues in theme than every other theme on the held-out set: they received 215 reactions, against 250 for the input, workflow and interface theme.
Scope: population: Issues labelled "feature request" in the GitHub repository musescore/MuseScore; period: Issues created up to the download on 2026-11-05; data: GitHub issue titles and bodies (first 2,000 characters); conditions: themes defined by fixed keyword rules, held-out set: even-numbered issues; not for: requests made in forums, Reddit, app reviews or by e-mail, users who do not file GitHub issues · Data: E2 (T1.r1, T1.r2, T1.r3, T1.r4, T1.r5, T1.r6, T1.r7, T1.r8, T1.r9) · Limitations: L1, L2, L3 · Extends: B1 · Planned: lab plan PL2, planned 5 November 2026 · Outcome: computed from the recorded results · Weaknesses: the held-out data was used elsewhere
Evidence
Count of MuseScore feature requests matched by the arranging and instrumentation keyword rule, on the 654 even-numbered issues labelled 'feature request' in musescore/MuseScore (held-out set), downloaded through the GitHub search API on 2026-11-05 (lab runs RN2, RN3 and RN6; plan PL3).
Setup: studied arranging and instrumentation theme · cases held-out set, MuseScore feature requests · procedure theme classifier · lens share of issues in theme
- R1 share of issues in theme = 0.0459 (theme: arranging_instrumentation, set: held-out)
Total GitHub reactions on MuseScore feature requests in each of nine keyword-defined themes, on the 654 even-numbered issues labelled 'feature request' in musescore/MuseScore (held-out set), downloaded through the GitHub search API on 2026-11-05 (lab runs RN2, RN3 and RN7; plan PL2). The computation repeats that of E1 under a second plan: the lab record version used allowed one decision rule per plan, and H1 and H2 needed different rules.
Setup: studied notation and engraving theme · cases held-out set · procedure theme classifier · lens total reactions on issues in theme
| Row | theme | total reactions on issues in theme |
|---|---|---|
| r1 | arranging_instrumentation | 44 |
| r2 | parts_and_linked_staves | 117 |
| r3 | transposition | 13 |
| r4 | generation_assistance | 12 |
| r5 | notation_engraving | 215 |
| r6 | playback_sound | 185 |
| r7 | platform_install | 76 |
| r8 | import_export | 154 |
| r9 | input_workflow_ui | 250 |
Limitations
Themes are assigned by keyword rules on issue titles and bodies; some matches are incidental words, and issues about a theme that use other words are missed.
GitHub feature requests come from people who file issues on GitHub, mostly experienced users and contributors; they do not represent all MuseScore users. A study of Moodle found that developers write most issue-tracker feature requests while users write most forum requests (KC and Hosseini 2026, arXiv:2606.20876).
Reactions are sparse: the median exploration-set issue has none, so theme totals are driven by a few issues.
Definitions
Methods 1
D4 theme classifier: Fixed case-insensitive keyword rules (regular expressions in themes.py, artefact A6) applied to an issue's title and body; an issue may belong to several themes.
Data 2
D1 MuseScore feature requests: Issues in the public GitHub repository musescore/MuseScore labelled 'feature request', downloaded with the GitHub API; pull requests excluded.
D3 held-out set: MuseScore feature requests with even issue numbers; not read during exploration; reserved for planned tests.
Variables and measures 4
D5 arranging and instrumentation theme: Feature requests matched by the theme classifier's arranging_instrumentation rule (arranging, orchestration, reduction, implode/explode, instrument change, ensemble, instrumentation, playable range).
D6 notation and engraving theme: Feature requests matched by the theme classifier's notation_engraving rule (slurs, ties, accidentals, stems, beams, tuplets, clefs, signatures, lyrics, chord symbols, articulations, dynamics, engraving).
D7 share of issues in theme: Issues matched by a theme rule divided by all issues in the set.
D8 total reactions on issues in theme: Sum of GitHub reactions (all reaction types) on the issues matched by a theme rule, at download time.
Context
References
- musescore/MuseScore GitHub repository,… musescore/MuseScore GitHub repository, issues labelled "feature request", https://github.com/musescore/MuseScore, retrieved 2026-11-05 Used by: A7
Artefacts
- A1 · data Every result of the lab record (117 results of 6 runs) at full precision: the lab result's ID, its run, the quantity as the run's script named it, its conditions and its value (both as JSON). The evidence's tables give the values to four significant digits.
- A2 · lab_record Lab record of the study: questions, hypotheses, plans, runs, results and decisions, as an append-only, hash-chained event log (Claimset lab record specification v0.15, begun 2026-11-05; exported with claimset 0.41.2).
- A3 · code Python script fetch_issues.py: downloads the feature requests through the GitHub search API, one creation year per query (run RN2).
- A4 · code Python script split_parity.py: splits the downloaded feature requests by issue number, odd numbers for exploration and even numbers held out (run RN3).
- A5 · code Python script explore_top.py: lists the exploration set by reactions (run RN4).
- A6 · code Python script themes.py: theme classifier: fixed keyword rules and per-theme counts (runs RN5, RN6, RN7).
- A7 · data · musescore/MuseScore GitHub repository,… The 1,289 issues labelled "feature request" in the GitHub repository musescore/MuseScore, as downloaded through the GitHub search API on 2026-11-05 (run RN2): titles, the first 2,000 characters of bodies, creation dates, states, reactions, comments and labels.
Score
Scores are computed by rules from the claimset file (EntryLib rubric 0.2): they check the record, not the science, and may change as the rubric and reviews develop. The rubric.
C2: 77, Medium
Novelty: new, by the authors · no close match on EntryLib
- −12 Open problems Limitation L1 (could bias magnitude, not mitigated): Themes are assigned by keyword rules on issue titles and bodies; some matches are incidental words, and issues about a theme that use other words are missed. To remove it: Mitigate it with evidence, or narrow the claim's scope so that it no longer applies.
- −5 Plan The lab record flags the test: holdout used elsewhere. To remove it: Rerun the test as planned, or explain the flag in a limitation.
- −3 Plan The plan's time is anchored by a git push only. To remove it: Anchor plans with the platform, a timestamp authority (RFC 3161) or a registry.
- −2 Open problems Limitation L3 (reduces precision, not mitigated): Reactions are sparse: the median exploration-set issue has none, so theme totals are driven by a few issues. To remove it: Mitigate it with evidence, or narrow the claim's scope so that it no longer applies.
- −1 Open problems Limitation L2 (reduces transferability, not mitigated): GitHub feature requests come from people who file issues on GitHub, mostly experienced users and contributors; they do not represent all MuseScore users. A study of Moodle found that developers write most issue-tracker feature requests while users write most forum requests (KC and Hosseini 2026, arXiv:2606.20876). To remove it: Mitigate it with evidence, or narrow the claim's scope so that it no longer applies.