Austrian Economics Wiki:Link repointing

The 2014 corpus this wiki was restored from carried roughly 3,800 mainspace links to mises.org. Very few of them still reached what they were written to point at: mises.org restructured after 2014, and the old URLs now either return a 404 or, more often, redirect to a generic landing page. That second failure is the harmful one. A link labelled Economic Calculation in the Socialist Commonwealth answering with the mises.org front page looks healthy to a link checker and is useless to a reader.

This page records what was done about it, so the automated edits in the page histories are accountable.

What was changed

In August 2026, roughly 3,000 link replacements across 700 articles repointed the dead citations to archived copies of the material they originally cited. Edits are attributed to ArchiveBot and marked as bot edits. Of the 2,843 distinct mises.org URLs in mainspace, 2,088 were resolved and 755 were left alone.

Links were resolved to one of two archives, preferring the first:

• the Liberty Archive — the modern full-text library. It takes priority wherever it holds the cited work. Legacy identifiers made this exact rather than approximate: the archive catalog keeps the same document and product identifiers the 2014 site used, so /resources/1164 resolves to The Anti-Capitalistic Mentality and /store/Americas-Great-Depression-P63.aspx to America's Great Depression. Journal citations resolve to the issue, author listings to the author's page, and citations of individual Human Action chapters to that chapter's reading page rather than to a scan of the whole book.
• mises.freecapitalists.org — a copy of the pre-2014 mises.org site, served from its original database, used where the archive has no equivalent. Mises Daily columns are the main case: legacy paths such as /daily/, /story/, /freemarket_detail.aspx and /misesreview_detail.aspx still resolve there to the specific article that was cited.

What was deliberately not changed

A wrong link is worse than an old one, so the process never guessed:

• Every candidate was fetched and checked before rewriting. A page that answered with a generic index title, or with a body too short to be an article, was rejected rather than rewritten. This matters because the failure being repaired is a page that answers 200 without serving what was asked for.
• No fuzzy title matching was used. Every resolution went through an exact identifier or an exact file path recorded in the original site's own database. Links that could not be resolved that way were left alone — 755 URLs, mostly under /media/, /rothbard/, /store/ and /etexts/, plus 311 to the old Mises Blog, which has no archived copy here.
• No link was deleted. Nothing was removed on suspicion of being dead.
• Identifier spaces that look interchangeable were not treated as such. On the snapshot, /document/3797 and /library/3797 are different items; only the first was used.

Only the URL was touched. Link text, citation templates, and surrounding prose are unchanged, so an edit is easy to read in a diff and easy to revert.

Reproducing it

The tooling is public and version-controlled at wiki.freecapitalists.org: scripts/repoint_mises_links.py builds and verifies the plan, and scripts/repointLinks.php applies it as a MediaWiki maintenance script, doing a read-modify-write against live wikitext so that concurrent edits are not overwritten.