I've definitely read some Wikipedia pages on niche technical topics in my field that were clearly written by someone with _some_ degree of expertise (maybe a grad student) but with no understanding of the purpose, standards, and style of Wikipedia. The voice is often all over the place.

Yeah absolutely. I've seen non-wiki-savvy experts do these kind of edits, get reverted, and stop editing wikipedia forever because the whole experience left a sour taste in their mouth. I recently found the essay below targeted to exactly that audience, it's super useful to helping experts who want to contribute.

https://en.wikipedia.org/wiki/Wikipedia:Expert_editors

I tried to upload an image that I took (as part of my graduate studies) but was previously published by media outlets and the whole situation really turned me off wikipedia. I think the image is still there though!

I've seen this too. Sometimes they seem like they are copied and pasted from whatever the person already happened to be writing when they came across the Wikipedia page.

A large number of my larger edits are fixing up voice / tone to make an article read more like an encyclopedia and less like an excited blog post.

What stops me editing more (usually it's a grammatical/formatting error I want to fix, or where I think a link to another page is warranted) is IP blocks. I'm logged in! Why do you care what my IP is! I have a (small) track record but most importantly it's all going against my name and if I'm doing bad things you can just block me!

I understand anon IP blocks, of course. But not logged-in ones. Especially when (afaict) all of Mullvad's (London at least) IPs are blocked.

Why would a math article need citations? A proof is a proof.

Only for experts capable of following the proof. For everyone else, it's a magical formula.

Not to mention, I very very much doubt that those articles prove every property they present (since I've never seen a math text of any kind do that, essentially).

You’ll always need some axioms to build on.

If you allow that, proofs made using proof assistants prove every property they present.

And of course, Principia Mathematica (https://en.wikipedia.org/wiki/Principia_Mathematica) is a non-computer example.

Both get tedious fast if you try to understand a proof from the axioms up. That’s why you’ll rarely find a publication starting from almost zero.

Because that's essentially original research. If it's a known proof, it should exist in written form somewhere else, which you can cite. if it's novel, it belongs somewhere other than Wikipedia.

Because wikipedia is not trying to report the capital-T Truth. Its goal is to summarize human knowledge found by others.

You could cite all of the notations that you introduce because that tends to be a mess.

Aside from the fact that proofs are often difficult to verify, one cannot prove what a monoid or a vector space is. They are definitions, and to make sure the ones in the article match what is used in the mathematical community, you need citations.

And yes, sometimes there are conflicts. In France, we have two competing definition of a limit (relating to wether you include the point in its neighbourhood), one being the traditional one, taught in schools, and one being the one that's become the world standard and used from university onwards.

How do you arbiter that on wikipedia without sources ?

I always chuckle when I see some like basic logical steps slapped with "citation needed".

I similarly chuckle in cases where there's a citation covering the whole sentence but someone felt the need to slap a "citation needed" on a specific clause despite that clause being supported in the existing citation.

Wikipedia's math articles are so bad that I can't even follow the ones about things I already know.

Math is one of the main categories--although there are other especially adjacent ones--where I often think that 1.) You either already know this stuff or you're going to emerge no more enlightened and probably click elsewhere after the first few sentences, and 2.) Some people really love to play with their equation editors.

Wikipedia's math articles can be dense and assume you're an expert, but I've never known them to be wrong.

There's other types of bad than wrongness.

Unfortunately I think 10 page expositions/derivations are the highest quality math and physics pages. I think the wikipedia style is fundamentally incompatible with communicating this kind of information unfortunately.