Which is certainly a metric of use.

That every few years there's a new crop of mass spectrometer/gas chromatography instruments with 10-100 times the sensitivity of the previous generations of instruments. Of course, one of the very first uses of these new instruments is to find toxic materials in the parts-per-billion range and do a shock story about them. Nevermind that the levels have no know correlation with health hazards other than the toxic materials being toxic.

Nothing of value was lost...

All 3rd world countries sell most any antibiotic over the counter with no prescription.

It's also worth noting that people with suppressed immune systems also take some of the most powerful antibiotics most of their lives.

I'd argue that if the data is used for AI learning, there is no need for it to be more than a couple weeks old. I'd also argue that given the general shitification of Wikipedia, images 10 years old would be more useful than the current data. Alternately, with some programming work, the crawler could look at the Wikipedia page change log and only download pages changed since the last crawl. This would work real time or locally updating from a new image. But then what do I know. I'm just a cynical old fart used to programming in assembly language.

Why you would crawl the on-line copy of Wikipedia when you can download an image of it and crawl it locally?

What about deuterium oxide? 0.015 percent of all tap water. That can't be good.

At Digital Equipment Corporation back in 1973.

And heard a rooster crowing in the background.

Is AI going to take that out too?

The "ruling grade" in railroad parlance.

My plasma cutter can easily run longer than 22 minutes.

I'm 73 years old. "Lifetime" subscriptions are probably not the best value.

Has a slightly different meaning in China than it does in the US.

Resisting the urge to post the canonical "BSD is Dying" rant.

Most here are too young to get it anyway.

Implying that every retraction is due to scientific misconduct. Which is false.