I feel like the problem is that people got too touchy about spam delivery.
This gave large email providers license to nuke everyone but other large providers or impose various performative-dance "standards" like SPF/DKIM which seem to just be footgun factories for non-experts.
The other problem was some very black-and-white models of mailer behaviour. Back when ISPs actually published delivery guidance, it was either "you're an individual sending 10 messages a year" or "You're sending a mailing list of 500,000 per day." The guidance was usually limited to nonsensical for things like transactional email or notifications, and I suspect smaller "regular mail services" hit similar corner-cases.
I have to wonder if these conditions eventually became a perverse incentive: oh, deliverability sucks for your small business? We'll happily host your domain and emails and make sure they get through. It's now in their best business interest to be as balky and hostile as possible.
To me, spam filtering rules should be entirely at the USER level. The provider can offer an array of tools, but they have no business blocking unless there's a distinct "the user checked this specific box and this is why the message was denied" paper trail available.
Maybe I'm an outlier -- I'm far more worried about false positives than false negatives. My work email gets negligible spam, but I've definitely seen the "can't be fully disabled" filters chuck actual time sensitive business-related messages into the spam folder repeatedly, and had to patch together anti-filtering rules for it.