A large number of pages on debian.org websites are no longer
accessible, and internal links on those pages which are accessible are
broken.

This includes the "submit" function of the BTS search form on
www.debian.org/Bugs, so I cannot check whether this problem has been
reported already.

Specific broken links/pages include:

- The BTS search form, as described above
- packages.debian.org/sid/all/yt-dlp/download (from search engine)
- www.debian.org (from www.debian.org/Bugs/pseudo-packages)
- bugs.debian.org/general (from www.debian.org/Bugs/pseudo-packages)
- wiki.debian.org/Teams/Debbugs/ArchitectureTags (from www.debian.org/Bugs/Reporting)

The problem in all cases is the same: Instead of the expected page, I
am redirected to something called *.debian.org/internal/challenge.html.
This shows as a blank grey screen with no content on it at all.

Inspecting the page source reveals that this is because of a piece of
broken javascript embodying various false assumptions:

- It assumes that cookies are enabled, or alternately, that instructing
  the user to "enable cookies" is in some way acceptable.
- It assumes that javascript is enabled, or alternately, that instructing
  the user to "enable javascript" is in some way acceptable.
- It assumes that ES6 is supported.
- It assumes that javascript crypto functions are supported/enabled.
- It assumes that the only cases it needs to cater for are "javascript
  works flawlessly" or "javascript is disabled". It does not cater for
  "javascript does not work". The "noscript" tag is not adequate for
  this.

And most fundamentally:

- It assumes by default that the user is a bot, and that the
  unavailability of javascript and/or cookies is infallible
  confirmation that the user is indeed a bot, whereas any human user
  must necessarily not only have them but have them turned on.

None of these happen to be true.

From the page source I see that the "noscript" tag contains a snarky
message about "AI crawlers break the web". What actually breaks the
web is blocking humans from sites because they don't allow sites to
run code or store state on their machines, on the principle of "guilty
until proven innocent" using a fundamentally broken method of "proof".

Nice.

<https://thelibre.news/foss-infrastructure-is-under-attack-by-ai-companies/>



Am 26. August 2026 02:09:21 MESZ schrieb Pigeon <pigeon@pigeonsnest.co.uk>:

This is now breaking search engines and causing debian web pages to
become effectively unsearchable.

For example, at the time of writing, the snippets on duckduckgo for at
least wiki.debian.org/manpages.debian.org, wiki.debian.org/ManualPage,
wiki.debian.org/InstallingDebianOn/Microsoft/Windows/SubsystemForLinux,
wiki.debian.org/FrontPage/, and no doubt others, say nothing more than
"I Challenge Thee" for both the page title and the page content.

These are merely chance discoveries: I was not searching for anything
to which those pages relate; they just happened to appear because the
word "debian" was one of my search terms.

Evidently the real page content is no longer being indexed and
therefore anyone making a search query which the affected pages would
be expected to match will not in fact find them.