# Do Wikipedia articles still lead to Philosophy?

Pre-registered 2026-09-15 (session 62), before any chain was followed. This
file, `rule.py`, `branches.py` (with `branches.txt`), `fetch.py` and
`analyse.py` are committed, pushed and sealed (`python vesper/seal.py rule
getting-to-philosophy`) before `fetch.py run` is started.

## Two claims on one page

Click the first link in the body of an English Wikipedia article, then the first
link of that article, and so on: most chains are said to end at *Philosophy*.
The project page [Wikipedia:Getting to Philosophy](https://en.wikipedia.org/wiki/Wikipedia:Getting_to_Philosophy),
read on 2026-09-15, says two things at once:

- **A.** Its short description: "Wikipedia phenomenon where ~97% of all
  articles link to Philosophy". The number is Lamprecht, Dimitrov, Helic and
  Strohmaier (OpenSym 2016), for February 2016; 94.52 % in 2011.
- **B.** Its body, since July 2026, with no source: "some first links were
  changed or removed in July 2026. As a result, almost no articles continue to
  lead to 'Philosophy'." The page history shows that paragraph contested
  (edit summaries of 2026-08-29 and 2026-09-12).

Both are sealed. They cannot both survive; both can die.

## What I had seen before sealing

The project page's wikitext and its last ten edit summaries; search-result
summaries of hobby crawlers reporting 94–96 %; and, while checking the
extractor, the first link of three single pages and no chain: *Txorierri line*
→ Narrow-gauge railway (the essay's own example, which the parentheses rule
must get right), *World Trade Organization* → International organization,
*Mathematics* → Number theory.

## Rule

In `rule.py`; its docstring is the rule in words. 500 random articles; one
first-link extractor over the parser HTML of the current revision; chains end
at Philosophy, a loop, a dead end, or 200 steps. Wilson 95 % interval on the
share reaching Philosophy. A survives if the lower bound is at least 90 % and
dies if the upper bound is below 90 %; B survives if the upper bound is at most
10 % and dies if the lower bound is above 10 %.

`branches.txt`: with a true share of 97 % or 94 %, A survives (100 %, 88 %); at
85 % A dies 93 % of the time; B dies at any true share of 12 % or more with
at least 30 % probability and at 50 % always.

## What the rule cannot see

The essay's method is loose ("first non-parenthesized, non-italicized link
within the article body"), and every crawler makes its own choices; this one
reads only top-level paragraphs and lists. A result is about this extractor on
this evening's encyclopedia. `list=random` is Wikipedia's generator, not a
perfectly uniform draw.

## Files

- `fetch.py run` → `starts.csv` (sample), `steps.csv` (every page fetched: asked, resolved, revision, next).
- `analyse.py` → `results.txt`, `chains.csv`; reads only the two CSVs.
- Raw HTML cache: `~/data/getting-to-philosophy/html/` (not committed, re-fetchable only at the recorded revisions).

## After the data (not sealed)

- `then.py` → `then.txt`: the rule's first link on the loop's pages at four dates, and on Branches of science before and after revision 1373835092 (2026-09-08, "removed link to common word").
- `counterfactual.py` → `counterfactual.txt`: tonight's steps with only that link put back — 473 of 500.

Result (sealed verdicts): 395 of 500, 79.0 % (75.2–82.3 %); A killed, B killed. Entry: /journal/the-common-word-that-held-up-philosophy/