Bot proposal: fill missing `alternate_names` on long-tail authors using Wikidata + Wikipedia evidence

## Problem

Many Open Library authors who have a real backlog of works on file (5+ works) still have an empty `alternate_names` array. As a result, readers searching using a different script or language form of the author's name cannot find them via the OL search box.

For example, before [today's manual edit](https://openlibrary.org/authors/OL412080A) the entry for "Chih-tsing Hsia" (work_count=29) had no native-script form on file, so a reader typing "夏志清" could not find him on OL.

The backlog exists on the long tail of the catalog — particularly authors whose primary name on file is a romanization while the native-script form is missing. Regardless of how big that backlog is, this bot is rate-limited to a small constant load (see Frequency below).

## Proposed bot

A small bot that, for one author at a time, adds **one** well-evidenced entry to `alternate_names` and writes a short edit-comment citing the sources.

### Evidence requirement (per edit)

Every edit must be supported by at least **two** independent reputable authorities. The strong default pair:

- **Wikidata** — the author's Q-id, with the multilingual `labels` block matching the proposed addition
- **Wikipedia article title in another language** — typically the article in the author's native-language Wikipedia

Other acceptable corroborators: official publisher / university page, national library authority records (LoC, BnF, NDL).

If two independent reputable sources cannot be found, the candidate is dropped. The bot does **not** write speculative or single-source additions.

### Why this isn't a typical LLM-writes-to-OL situation

The bot itself does **not** generate the alternate name with an LLM. The pipeline upstream of this bot is:

1. A discovery crawler reads OL search to find authors with `alternate_names == []` and `work_count >= 5`.
2. For each such author a research worker (which may use an LLM) gathers the two-source evidence from Wikidata + Wikipedia and produces a structured JSON proposal: `{ol_key, proposed_addition, comment, evidence: [{source, url, value}, ...]}`.
3. A QA gate checks the JSON for required fields, that the proposed addition matches the value reported by both evidence URLs, and that the OL author still has `alternate_names == []`. Only proposals that pass the gate (score 100) are forwarded.
4. **This bot** then takes the gate-approved JSON and submits the single addition via the OL API.

Concretely: the LLM proposes; an independent automated check verifies; the bot only commits proposals that survived both. The bot's input is not free-form LLM text — it is a structured proposal whose two evidence URLs the bot itself can re-fetch and compare against before PUT.

### Scope guardrails

- **One** addition per edit (not bulk-replacing the array)
- **Author pages only** (`/authors/OL...A`). No work / edition / subject edits in this bot.
- Only adds to `alternate_names` (not removing or reordering existing entries)
- Skips any author whose `alternate_names` is non-empty (someone else already handled them)
- Skips any author whose `work_count < 5` (not worth a name edit)

### Frequency

Targeting **≤ 8 edits per day** total (well below polite-bot thresholds), regardless of how large the candidate pool is. Each edit is a single PUT plus 2 GETs, paced at 1.5s between requests. Happy to tighten further if reviewers want.

### Edit comment format

Pure factual citation. Example:

> Adding native-script form per Wikidata Q778276 (zh label '苏童') and Chinese Wikipedia article title '苏童'.

No promotional content, no project attribution in the edit comment.

## Account

Bot username: `agenticcommonsbot` (S3 keys configured; currently 403 on PUT pending bot review, as expected).

The S3-key authentication path is wired and a test login + GET against `/authors/OL2630047A.json` returns 200 cleanly. The PUT is the only step pending privilege grant.

## Code

Forthcoming PR will add a directory `AgenticCommonsBot/` to this repo with:

- `wikidata_author_alias_bot.py` — single-file Python 3, stdlib only, S3-key auth via `/account/login`
- `README.md` — usage, dry-run output, evidence checklist
- `sample_proposal.json` — a real two-source proposal for review
- `sample_dry_run.txt` — captured dry-run output against a live (still-empty) OL author so reviewers can see the exact request/response shape

## Maintainer

- GitHub: [@agentic-commons-foundation](https://github.com/agentic-commons-foundation)
- Email: wiki-bot@agentic-commons.org

## Ask

Looking for a thumbs-up on the approach before opening the implementation PR, plus pointers on anything you'd like tightened. Happy to adjust the per-day cap, evidence requirements, edit-comment format, or anything else that helps reviewers.