Comment by account42
9 days ago
> A non-zero amount.
So you don't know and we can assume its not relevant.
> And what happens when someone in an edit war makes death threats to another editor. What happens when they figure out where that person lives and show up at their house?
That's a matter for law enforcement and organizations shouldn't have their own badly implemented versions of it.
> Its also important to keep in mind, last year at a wikipedia event someone brought a gun and attempted to commit suicide on stage (traumatizig most of the attendees).
They don't need to be organizing events in the first place. That's already itself a completely unnecessary money sink.
> I'm mostly just claiming the amount is not zero. There is a lot of room to debate specifics. However stuff does pop up. Security vulns happen. Software needs to be updated to work with updates dependencies (php has been making a lot of breaking changes lately). The AI boom has made access patterns shift causing caches to be less effective then before. Sometimes servers die and you need to swap out a replacement. Etc etc. There is always something.
Making unauthorized requests 100% static is not rocket science. This doesn't need an ongoing tens or hundreds of millions per year.
> That's a matter for law enforcement and organizations shouldn't have their own badly implemented versions of it.
Theoretically, yes.
Practically? Law enforcement can be ignorant, incompetent or slow as molasses, or a combination of all three of them. Besides, some places like the EU impose specific requirements on what happens with certain kinds of speech, especially anything glorifying or calling for violence, so as an organization, you have to have people and procedures to deal with it.
> They don't need to be organizing events in the first place. That's already itself a completely unnecessary money sink.
It was a core factor, other than Jimbo Wales wanting to distance himself/his company that hosted Wikipedia (Bomis) from it a bit, why Wikimedia got founded in the first place.
> Making unauthorized requests 100% static is not rocket science.
The problem is the bots are not respecting robots.txt and instead do stuff like load the source diff pages. It does not make any sense for AI training and likely pollutes the database, but unfortunately the diff pages are among the most expensive to render, the worse the older the requested diff is.