Jake
26c7307607
Phase 9 of the rebrand cascade. Sweep covers everything the Phase 8
frontend pass deliberately skipped: docs/, root markdown, scripts,
Cargo.toml descriptions, code comments that survived earlier
word-boundary sed, plus a handful of identifiers caught on the final
verify pass.
transcription-app changes:
- README.md, HANDOVER.md, KNOWN-ISSUES.md, run.sh — magnotia/Magnotia
-> lumotia/Lumotia.
- docs/ — sweep across all subdirs except docs/handovers/ (preserved
as immutable audit trail). Includes architecture-map references
to magnotia_core::*, magnotia_storage::*, etc. now pointing at
lumotia_*; dev-setup.md tracing output examples (lumotia_startup
target); brief/ + superpowers/ + issues/ + whisper-ecosystem/ +
audit/.
- Cargo.toml descriptions on 9 crates (core, audio, cloud-providers,
hotkey, llm, mcp, plus referenced others).
- crates/core/src/{error,hardware,recommendation,paths}.rs +
crates/audio/src/wav.rs + crates/llm/src/model_manager.rs +
crates/cloud-providers/src/keystore.rs + crates/mcp/src/lib.rs —
doc comments and a model-manager user-agent string.
- Caught on final pass: BroadcastChannel("magnotia_task_sync") -> ...
("lumotia_task_sync"); magnotia_locale i18n localStorage key
renamed + migration shim added; CSS keyframe names
magnotiaPulse / magnotiaBar / magnotiaFade renamed in the design-
system kit; magnotia_viewer_item / magnotia_viewer_mode handoff
keys renamed in HistoryPage + viewer/+page.svelte; src/assets/
wordmark.svg text.
- src-tauri/src/lib.rs comment cleanup ("magnotia era" was sed'd
to "lumotia era" earlier — restored).
Preserved (intentional):
- crates/core/src/paths.rs — keeps "magnotia" / "Magnotia" / ".magnotia"
legacy detection strings in legacy_and_target_paths() so the
migration shim can still find user data from the magnotia era.
- src/lib/stores/{page,focusTimer}.svelte.ts + src/lib/i18n/index.ts
— migration call sites reference the legacy magnotia keys
deliberately.
- docs/handovers/ — historical audit trail.
cargo build --workspace passes. npm run check: 0 errors / 0 warnings
(3958 files). cargo test --workspace: 339 pass / 0 fail.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
29 lines
4.2 KiB
Markdown
29 lines
4.2 KiB
Markdown
<!-- Source: Lumotia Master Brief — Appendix A4: Latency, Working Memory Decay, and Software Architecture -->
|
||
|
||
## A4. Latency, Working Memory Decay, and Software Architecture
|
||
|
||
**Core finding:** 75–81% of ADHD cases show measurable working memory deficits (d = 1.63–2.03). Every millisecond of interface latency disproportionately taxes ADHD working memory. Local-first architecture is a cognitive accessibility requirement, not a technical preference.
|
||
|
||
**Working memory deficits in ADHD:**
|
||
- **Kofler et al. 2020** (*Neuropsychology*): 172 children, bifactor modelling. **Very large magnitude central executive WM deficits: d = 1.63–2.03**, affecting **75–81% of ADHD cases**. These deficits "determined consistent difficulties in anticipating, planning, enacting, and maintaining goal-directed actions."
|
||
- **Weigard & Huang-Pollock 2017** (*Clinical Psychological Science*): Applied the Time-Based Resource-Sharing (TBRS) model to ADHD. Children with ADHD experienced **higher cognitive load than controls in identical task conditions** because slower processing speed leaves less time for WM refreshing. Every millisecond of additional processing demand disproportionately taxes ADHD working memory.
|
||
- **Barrouillet, Bernardin & Camos 2004** (*Journal of Experimental Psychology: General*): The TBRS model — WM recall is a **negative linear function of cognitive load**, where cognitive load equals the proportion of time the attentional bottleneck is occupied by processing rather than refreshing memory traces.
|
||
|
||
**HCI response time thresholds:**
|
||
- **Miller 1968** (*AFIPS Conference*) and **Nielsen 1993** (*Usability Engineering*): Delays beyond **100ms** break direct manipulation feel. Beyond **1 second**: flow of thought disrupted. Beyond **10 seconds**: complete attentional disengagement. These are neurotypical baselines — effective thresholds for ADHD users are almost certainly shorter given reduced WM capacity.
|
||
- **Card, Moran & Newell 1983** (*The Psychology of HCI*): Expert users completed tasks **30–40% faster** with sub-second response systems vs. 2-second systems — a penalty amplified in ADHD populations with elevated switch costs.
|
||
|
||
**ADHD-specific latency vulnerability:**
|
||
- **Barack et al. 2024** (*Proceedings of the Royal Society B*): Pre-registered foraging study, **457 participants**. Those screening positive for ADHD **departed resource patches significantly sooner** — their exploration/exploitation trade-off is biased toward exploration. Every loading delay creates an artificial "depleting patch" that triggers the ADHD exploration impulse, manifesting as tab-switching, app-switching, and task abandonment.
|
||
- **Ardalani et al. 2020** (*Psychological Research*): Inattentive traits predict higher switch costs under working memory load — each navigation step imposes a disproportionate cognitive tax.
|
||
- **Madore et al. 2020** (*Nature*): Pre-encoding attentional lapses directly predict memory failure. Software that minimises attention-capturing events (loading screens, error states) directly supports better memory encoding.
|
||
|
||
**Applied studies (from earlier research):**
|
||
- **127 ADHD knowledge workers study (KLM + EEG):** 4.7 seconds cognitive overhead per app switch. 11.3 seconds context-reconstruction latency. Tools with >90-second setup increase cognitive load by 2.3x.
|
||
- **NIH study of 247 ADHD adults (8-week baseline):** Zero-friction AI tools achieved 31–47% reduction in task-switching latency, 58% reduction in off-task interruptions, 42% increase in on-time completion.
|
||
|
||
**Local-first as cognitive ergonomics:**
|
||
- **Kleppmann et al. 2019** (*ACM Onward! '19*): Seven ideals of local-first software. Ideal #1 — "No spinners: your work at your fingertips." Primary copy of data on the user's device means read/write operations at local disk speed (sub-millisecond), not network speed (50–500+ ms). Synchronisation happens asynchronously in background.
|
||
|
||
**Implication for Lumotia:** Local-first architecture keeps all interactions within Miller's 100ms direct-manipulation threshold, preventing the WM decay → exploration bias → task abandonment cascade. The 90-second setup threshold is a hard design constraint. Voice capture must work in under 3 seconds from app open.
|