149
How AI and Wikipedia have sent vulnerable languages into a doom spiral
(www.technologyreview.com)
this post was submitted on 23 Oct 2025
149 points (93.6% liked)
Technology
76361 readers
1325 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related news or articles.
- Be excellent to each other!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, this includes using AI responses and summaries. To ask if your bot can be added please contact a mod.
- Check for duplicates before posting, duplicates may be removed
- Accounts 7 days and younger will have their posts automatically removed.
Approved Bots
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Hmm, the law begins with "Given enough eyeballs". So it's explicitly not about small-language Wikipedia sites having too few editors.
It also doesn't talk about finding consensus. "All bugs are shallow" means that someone can see the solution. In software development, that's most often quite easy, especially when it comes to bugfixes. It's rarely difficult to verify whether the solution to a bug works or not. So in most cases if someone finds a solution and it works, that's good enough for everyone.
In cultural fields, that's decidedly not the case.
For most of society's problems, there are hardly any new solutions. We have had the same basic problems for centuries and pretty much "all" the solutions have been proposed decades or centuries ago.
How to make government fair? How to get rid of crime? How to make a good society?
These things have literally been issues since the first humans learned to speak.
That's why Linus' law doesn't really apply here. We all want different things and there's no fix that satisfies all requirements or preferences.
That's because people generally don't know or understand these solutions. One can't raise above separate human's ability to understand any system. And I mean any. Doesn't mean a housewife has to know electrical engineering, but enough people in any group reliant upon any system should have passive knowledge and understanding.
So, thanks to AI translators and such, I can now link https://kvant.digital/ here.
The very first problem in the very first issue of that journal is very enlightening.
Wikipedia (from my understanding) was built off similar doctrine as Linus' law for iterative improvement where the dedicated and the many culled the misinformation and the outdated.
I wonder what would be a viable alternative for delicate situations like these where "hugs of death" (too many eager users who don't understand the damage they're causing) are occurring due to the current model for niche cultural systems like these. Maybe a council established group of authors and editors based on their background and qualifications?
That's kinda what Wikipedia does. They have a quite elaborate review process before stuff goes live: https://en.wikipedia.org/wiki/Wikipedia:Reviewing
In the English Wikipedia, that process is working quite well. But in e.g. the Welsh Wikipedia or other tiny languages, they might only have a handful of reviewers in total. There's no way that such a small group of people could be knowledgeable in all subjects.
Welsh Wikipedia, for example, has less than 200 total active users, and there are dozens of small language or dialect Wikipedias that have <30 active users.
https://en.wikipedia.org/wiki/List_of_Wikipedias
I don't think there's an actual solution for this issue until AI translations become so good that there's no need for language-specific content any more. If that ever happens.