The point isn't to delete bad content. The point is to delete content that is known by so few people that it is unlikely to get the attention needed to become a good article.

To put it another way, even if someone writes a really, really good article about their neighbor's dog Spike - complete with genealogy down to Rover the fifth, the harrowing near-death encounter when that gallstone removal surgery went wrong, and even a detailed account of that incident last Wednesday when Spike wet the carpet, there are so few people who know anything about that dog that it's impossible to fact-check the article. For all anyone knows it's all made up. So the line has to be drawn somewhere, and there need to be clear policies for it, as otherwise, well, things like this happen.

But Spike is a straw-man (well, dog).

These programming languages aren't non-existant, and can be verified as actually existing. It's not that there are no references to it.

On top of that, i'm still unclear why it would even matter? Things with little interest will generate little traffic. The potential to misinform is pretty low. And if it were to be the case that an article that was incorrect (maliciously or not) were to suddenly see an increase in traffic, presumably that can only serve to encourage people to point out inaccuracies and correct them (treating fictionalized accounts of anything as being an inaccuracy). And wikipedia already has a mechanism for resolving synonymy and name-resolution disputes.

If Wikipedia just accumulated everything it could, we'd end up with a billion articles 90% of which are vandalized, outdated, mistaken crap. The potential to misinform would be huge. They can only successfully scale the corpus at the rate at which they gain volunteer editors (who are far rarer than people who just write one article about their favorite obscure thing, and then leave).

It is frankly easier to get a fake article by the editors than a real one. The editors generally do not check articles for accuracy or quality (beyond what a high-schooler with a dictionary could do) — they just fastidiously apply a checklist of rules.

Also, I think it would be easier to get editors if being an editor were more about making the site good and less about dealing with the head cases who currently edit the site. (Nothing personal intended against any individual editors who may be reading — it's just a general reflection of how things look to people who have casually attempted to try editing.)

This still sounds like a technical limitation. A page that gets ten hits a year should not demand as much editorial oversight as one that gets a million. Who cares if a long tail article gets vandalized if no one sees it?

It sounds like a missing UI. Something like the ability to show recent changes sorted by the pages' popularity. Too simple?

If something is vandalized or mistaken then delete it, if it is outdated then there should be some kind of long term archiving policy but I fail to understand the reasoning behind deleting articles which are factually correct, even if rather obscure. What harm do they do?

Either all articles are high quality and there's no room for esoteric knowledge, or the entire system will collapse under the weight of misinformation? That's an obvious false dichotomy.

You could also just take the article in good faith at face value. I don't think anyone really believes everything on wikipedia is going to be top notch quality. If people are looking up popular things a lot of work will have been done maintaining the article, if they are looking up things of narrow interest it is reasonable to assume that everything there might be from just a single perspective, but could still give value to that person.