Drift table (2026-10-23)

Shared patterns are real. Shared implementations often are not. Domain overlap is small: pages/OCR vs speakers/audio.

TX pins cited in Transcribe docs: analysis 50a0ede…; settings notes also cite da742f9….

Naming pairs

Idea	TranscriptX	Transcribe
UI package	src/transcriptx/web/	src/transcribe/ui/
Nav registry	web/navigation.py (section="settings")	ui/navigation.py (section="system")
Workspace roots	PathSettings	RuntimePaths
Analyse launcher	Run Analysis	Analyse
Consume text	Transcript	Reading
Correction surface	Corrections Studio	Review (OCR lanes)
People/places	Speakers → Locations (Folium)	View → People & Places (st.map)
Tag rules	io/tag_validation.py	tagging/kernel.py + catalog store
Geocode cache	location_cache.json	data/cache/geocode.json
Doctor	DiagnosticsController	DoctorService + CorpusDoctorService
Analysis layout	core/analysis/<module>/	analysis/modules/<id>.py
Backup format	transcriptx.workspace-backup	transcribe.workspace-backup
Default port	8501	8510

Do not port (from analysis_module_porting.md)

interactions, pauses, transcript_quality, llm_speaker_summary, contagion, and the voice family (voice_features, voice_mismatch, voice_tension, voice_fingerprint, voice_charts_core, voice_contours, prosody_dashboard).

Deferred (not this programme): reinterpretations such as politeness, echoes, temporal_dynamics, momentum, and special-case ocr_quality.

Drift

Area	TX	Transcribe	Severity
Nav skeleton	primary / workflow / view / settings	Same groups; notebook pages	none on skeleton; large on page set
Settings tabs	Storage, Watcher, Speakers, Questions, Keywords, Corrections; Profiles and Dashboard Builder as pages	Configuration, Analysis, Detection, Tags, Prompts, Interface, Models, Profiles, Export	medium (intentional)
Tag catalog	String sanitize; no personal_corpus.tag-catalog	Full kernel + colours + Settings → Tags	large (docs claim a share; TX never got the kernel)
Analysis ids (core 25)	Present plus many more	Same 25 registered	none on ids
Analysis depth	Multi-file packages, sklearn/spaCy/viz	Mostly adaptations (TF-IDF, seed buckets, baseline clouds, optional extras)	large vs “ported” wording; keep for v1
Result envelope	TX pipeline objects	transcribe.analysis-result	medium (intentional)
Export	Transcript bundle / library-by-tag	transcribe.notebook, export-bundle, finetune	medium (intentional)
Backup ZIP	Role roots under transcripts/config/data	Role roots projects/data/inbox/exports	small (same pattern)
Doctor / Diagnostics	Diagnostics page	Diagnostics + notebook + corpus doctor	small
Places	Folium; default geocode on	st.map; opt-in Nominatim	medium (intentional)
Review vs Corrections	Transcript/speaker corrections	OCR lanes	large (different jobs)
First-run / sample	Fixtures + workflow docs; demo mode removed	Home CTAs; no samples/	large (Transcribe UX)
OCR / corpus import	Absent	Core product	none as a gap-to-port
Voice / speakers	Shipped	Out of scope	none (intentional)
PR CI	Smoke-focused; nightly owns contracts/coverage	Matrix 3.10–3.12 + blocking release-checks	medium (different shape; keep Transcribe’s PR gate)
Nightly / issue templates / Docker CI	Present	Missing (I6)	large
Release extras	clean_env_audit, image_pip_check, perf envelopes	Core I2 kit only	medium
Pre-commit	black + ruff fix + mypy	thinner critical ruff	small
Coverage floor	70%, omit web/	70%, omit ui/	none
Docs hosting	Pages + Sphinx + RTD + docs/workflows/	Pages + Sphinx + RTD; no workflow media	small
Runtime coupling	n/a	Must not import TX	none (policy held)

Transcribe-only (keep)

Page/notebook domain; OCR lifecycle; Prompt Hub; Detect; corpus ImportRun; British Analyse; port 8510; project OCR precedence over env; organisation tag catalog.

Doc authority for intentional divergence

settings_tx_alignment.md, places_tx_alignment.md, analysis_module_porting.md, analysis_visual_compare.md. Treat tag-catalog.md / INTEGRATION_SEAM.md “shared with TX” as planned copy-boundary.