Editorial Guide

Editorial Guide

Page types

Person pages

Namespace: Main (e.g. Jane Doe)

Encyclopedic article about a person. Documentary voice: third person, past tense, factual. The person page is a hub that links out to episode pages.

Lead paragraph: Biographical identity first, relationship to wiki owner in one sentence, arc in one more. No statistics in the lead — save those for a dedicated section. No emotional framing.

Jane Doe (born 3 May 1997) is a Berlin-based photographer and former classmate. She and the wiki owner exchanged 6,200 Instagram DMs between March 2021 and May 2022, the largest one-on-one thread in the archive. They connected over film photography, collaborated on a zine, and met in person in Berlin in November 2021. The conversation faded after Jane moved to Tokyo in early 2022.

What belongs: Biographical details, chronological arc (summarized not exhaustive), key statistics, links to episode pages, media embeds, source citations.

What doesn't belong: Full voice note transcriptions, raw research notes, detailed retellings of specific episodes (those get their own episode pages).

Blockquote discipline: Only quote when exact words matter more than the information — confessions, turning points, self-descriptions that can't be paraphrased without losing the voice. Let paraphrasing carry the rest.

Episode references: When the chronological arc mentions a story with its own episode page, summarize in one sentence and link out:

On 14 August, Jane described a disastrous shoot at Tempelhof
in a series of five voice notes (see [[Jane and the Tempelhof Disaster]]).

Episode pages

Naming: {Person} and the {Episode Title} (e.g. Jane and the Tempelhof Disaster)

Self-contained page for a specific story, event, or extended narrative. More narrative latitude than person pages, but still third-person and factual. The storytelling comes from sequencing, detail, and well-chosen quotes — not from the writer's adjectives.

Create when: 3+ voice notes telling a connected story, or a sustained back-and-forth that would take more than two paragraphs to tell properly.

What belongs: Full contextual setup, the story with detail, all relevant voice note transcriptions inline, audio/video embeds, surrounding messages, links back to person page and related episodes.

What it should feel like: Reading one should feel like being shown a specific memory. Beginning, middle, end.

Source pages

Namespace: Source (e.g. Source:Whatsapp)

Documentation page for a data source in the vault. wai snapshot <dir> creates an initial page with a {{Source}} infobox and basic file-type inventory. The editor enriches it by exploring the actual data — opening databases, counting records, identifying top contacts, extracting date ranges — and documenting what the source contains and how to query it.

Lead paragraph: What kind of data, whose account, platform, date range covered, extraction date, total size. Factual, one to two sentences.

A decrypted Android WhatsApp backup of [[Jeremy Philemon]]'s account, extracted on May 29, 2021. The export covers nearly four years of messaging history, from June 2017 through the extraction date. It includes the full message database, contacts database, decryption key, and 9,744 media files totaling 5.5 GB.

Structure (include all sections that apply):

1. {{Source}} infobox — snapshot hash, file count, total size (auto-generated by wai snapshot)
2. Lead paragraph — data type, account holder, platform, date range, extraction method
3. Overview — property/value wikitable: account holder, platform, export date, date range, total records, sent/received breakdown, session counts, starred items, call log entries
4. Key files — wikitable of important files (databases, keys, media directories) with sizes and descriptions
5. Content breakdown — message types, media types, or record categories with counts
6. Top conversations — ranked wikitable of most active contacts or groups with display names, identifiers, and message counts
7. Volume over time — monthly or yearly message/record counts showing activity patterns and notable peaks or dips
8. Media files — subfolder breakdown with file counts and sizes
9. File types — extension inventory with counts (richer than the auto-generated version — include all extensions)
10. Context — what period of life this covers, relationship to other sources, notable observations about the data
11. Data quality notes — database schema version, timestamp format, encoding caveats, missing data, known issues
12. Querying — step-by-step instructions for programmatic access: how to locate the database in the vault via snapshot hash, key tables and their columns, timestamp conversion recipes, common SQL queries (top conversations, read a thread, monthly volume)
13. See also — links to related or overlapping sources
14. Categories — [[Category:Sources]] plus year categories as appropriate

What belongs: Statistical summaries derived from the data, database schemas, querying recipes, contextual notes, cross-references to other sources and person pages.

What doesn't belong: Interpretive analysis, raw message content, personal commentary.

Enrichment approach: Open the database or files in the vault directly. Run SQL queries (for SQLite databases) or parse JSON/CSV files to extract statistics. Count messages, identify top contacts, compute date ranges, break down message types. The goal is to make the source page a self-contained reference that another editor can use without guessing.

Editorial standards

Core principles

1. One canonical home — every piece of content lives in one place. Other pages link to it; they don't duplicate it.
2. Prefer splitting to growing — a story that takes more than two paragraphs deserves its own page.
3. Documentary voice on person pages — third person, past tense, factual. Like Wikipedia.
4. Episode pages allow storytelling — still third-person and factual, but more narrative.

Don't interpret for the reader

• Don't editorialize: Replace adjectives with specifics. "They exchanged 1,800 messages in five days, averaging 360 per day" — not "The conversation density was staggering."
• Don't inflate significance: Cut "marking a pivotal turning point" and "reflecting a broader shift." If something is significant, facts demonstrate it without a caption.
• Don't use promotional language: No "vibrant," "rich," "renowned," "groundbreaking," "nestled," "showcases."
• Don't attribute vaguely: No "observers have noted" or "friends describe her as." Cite specific sources.

Prose quality

• Say "is" when you mean "is": Not "stands as" or "serves as."
• Keep sentences short: Split anything over ~40 words.
• Vary rhythm: Mix short and long sentences. Avoid the "rule of three" tic.
• Use punctuation precisely: Don't overuse em dashes as a Swiss Army knife.
• Don't cycle through synonyms: If you said "conversation," say "conversation" again.
• Avoid formulaic transitions: Cut "moreover," "furthermore," "notably," "additionally."
• Don't frame by negation: State what something is, not what it isn't.
• Don't end sections with summaries: No "In summary," "Overall," "In conclusion."

For the full words-to-watch list, see words-to-watch.md.

Quoting conventions

Use direct quotes when:

• The exact words matter (confessions, self-descriptions, turning points)
• The phrasing is distinctive and can't be paraphrased without losing character
• The quote is short (under ~30 words)

Don't quote:

• Routine factual statements that can be paraphrased
• Three quotes in a row saying similar things
• To show off the archive

Integrate quotes grammatically into sentences. Save {{Blockquote}} for extended passages (2+ sentences) that need to stand alone.

Talk page structure

Talk pages use these sections as needed, in this order. Omit any with no content.

1. Active gaps — open editorial questions marked {{Open}}
2. Resolved — closed questions marked {{Closed}}, corrected ones {{Superseded}}
3. Editorial decisions — choices about structure, scope, voice, what to include/exclude
4. Infrastructure — technical issues and their resolutions
5. Agent log — one entry per task: ID, date, what changed, link to task page
6. Research notes — index of raw research materials (what exists, where it is, which pages consumed it)
7. Voice note transcriptions — complete chronological index with inline audio embeds

Active gaps

=== Birth year unknown ===
{{Open}}
Likely 1996-1998 based on contextual clues. Never stated directly in DMs.
Would require external source to confirm.

Resolved

=== Did they meet in person? ===
{{Superseded}}
Previously resolved as one meeting (dinner, Nov 12).

{{Closed}}
Three meetings confirmed via WhatsApp thread (snapshot 3f0390a3...):
dinner (Nov 12), gallery opening (Nov 13), darkroom session (Nov 14).

Agent log

=== Task:0008 — Initial page creation ===
2026-02-15. Created page from Instagram DM research (6,200 messages).
Posted 3 open gaps. See [[Task:0008]].

What does NOT belong on talk pages

• Reader-facing content (goes on person/episode pages)
• Duplicate research indexes

Citation system

Inline citations use <ref> tags rendered via <references /> in a == References == section. This is standard MediaWiki.

Inline citation templates

Cite message — for text messages (DMs, chats):

<ref name="ig-2021-04-15">{{Cite message|snapshot=a1b2c3d4e5f6
|date=2021-04-15|thread=janedoe_12345|note=Family background exchange}}</ref>

Cite voice note — for voice note content:

<ref>{{Cite voice note|number=7|date=2021-06-03|speaker=Jane
|snapshot=a1b2c3d4e5f6|note=Darkroom discovery story}}</ref>

Cite photo — for facts derived from photos:

<ref>{{Cite photo|file=IMG_2847.jpg|hash=...|date=2021-05-20
|snapshot=a1b2c3d4e5f6|note=University ID confirming enrollment}}</ref>

Cite video — for video content:

<ref>{{Cite video|file=berlin_gallery_opening.mp4|date=2021-11-12
|snapshot=a1b2c3d4e5f6|note=Gallery opening footage}}</ref>

All templates include: snapshot (vault hash), date, note (human-readable description).

Bibliography template

Cite vault — for the Bibliography section, describes full vault snapshots consulted:

{{Cite vault|type=messages|snapshot=a1b2c3d4e5f6
|timestamp=2021-03-01/2022-05-15|note=Instagram DM thread with Jane Doe}}

Additional fields: type (messages, photos, video, etc.), timestamp (date range).

When to cite

Always cite: Biographical facts, direct quotes, specific event dates, statistics, claims corrected or disputed on the talk page.

Don't need citations: Broadly sourced observations, information already attributed inline with a date, episode page content drawn from a defined set of voice notes listed at the top.

Named refs for reuse

Jane's mother is from Munich.<ref name="ig-2021-04-15" />
Her father works in Zurich.<ref name="ig-2021-05-02">
{{Cite message|snapshot=a1b2c3d4e5f6|date=2021-05-02
|thread=janedoe_12345|note=Family details, father in Zurich}}</ref>
She has a younger brother named Max.<ref name="ig-2021-04-15" />

Page structure

Every person and episode page ends with:

== References ==
<references />

== Bibliography ==
{{Cite vault|type=messages|snapshot=a1b2c3d4e5f6
|timestamp=2021-03-01/2022-05-15|note=Instagram DM thread with Jane Doe}}
{{Cite vault|type=voice_notes|snapshot=b2c3d4e5f6a1
|timestamp=2021-04-12/2021-06-03|note=47 voice notes, Jane and wiki owner}}

References = inline citations tracing specific claims to specific moments in the vault.

Bibliography = full vault snapshots consulted for the page overall.

Namespaces