I'm not seeing it

Are there alot that give up for being blocked, relative to the number of hosters? That's the popular wisdom, but there's a dearth of data. The wisdom could be a product of how it's discussed: people who tried and failed being more vocal than those who are successful, similar to restaurants reviews where people with bad experiences are more likely to leave reviews than the bulk of customers who were satisfied, especially for high-volume venues (e.g. McDonalds).

I've been hosting my own e-mail for over 20 years, and it's been at least a decade since I've noticed any delivery issues. Even then it was unclear what the issue was as it was another user on my system using an external client, and they chose to simply stop trying rather than debug. There are plenty of institutions and companies still self-hosting, enough that there's still a market for selling software and hardware e-mail products. And while larger hosters have more resources to attend to deliverability problems, they're also having to deal with users sending messages more likely to trigger filters, e.g. many Cc recipients.

The scope of the problem is just very unclear to me. I don't know what to think. I've switched hosting setups at least a half-dozen times over the decades--colocation, leasing, virtual hosts, etc--and every time I switch I get anxious, but things have worked out pretty well so far. Currently I host from home through a tunnel from Vultr, but before then it was from a private office on a dedicated circuit from Cogent. I also host my own DNS name servers, some domains primary and secondary self-hosted, others primary self-hosted with secondary at EasyDNS. Managing SPF, DKIM, and other records via flat locally controlled zone files feels simpler and more transparent that way, but perhaps for someone new at this game it might seem more opaque than using popular web services and GUIs.

That said, for the few business ventures I've started with others, I just choose GMail. I don't want to get blamed for deliverability problems the sales and marketing side might have. It's like uptime issues--across the years my self-hosted uptimes aren't appreciably worse, and perhaps in some cases might even be better--than is typical for, e.g., AWS-based setups, especially taking into consideration misconfigurations and similar accidents. But when AWS or Google is down, coworkers and customers are understanding; when your self-hosted services are down, you're the unprofessional idiot (classic CYA dynamic--nobody gets fired for buying IBM, Microsoft, AWS, Slack, etc).

Most likely, because they do not fulfill the requirements for larger senders.
- all the classic requirements
- Use SPF; ~all is the way to go. Ideally, SPF would have lost most relevance by now, but it can serve as tool to mitigate the increasing problem of DKIM replay, so it actually gained importance over the last year. But ignore recommendations to use -all: Those were not written by people who understand large mail systems and all the complex, indirect mail flows that people and organizations use. (If -all seems to work for you, this is because 1) you either have no users or 2) feedback from your users doesn't reach you or 3) almost all relevant recipients treat your -all like ~all).
- Sign with DKIM
- Align header From and the signing domain; ideally also "envelope from" for those sad recipients that check SPF, but not DKIM
- declare a DMARC policy of at least quarantine
- implement "one-click unsubscribe" as specified in https://datatracker.ietf.org/doc/html/rfc2369 / https://datatracker.ietf.org/doc/html/rfc8058. In practice, this build on top of DMARC. If your reputation is good enough, it can even lead to a users "report spam" action not having any negative impact on you. Instead, systems will treat this as "recipient wants to cancel an originally legitimate subscription".
- set a List-Id header

You really need all of the above nowadays for any non-trivial list.

A mailing list that accepts mails from its subscribers must also munge the From header. There currently is no realistic way around it (it can be avoided for internal mailing lists were all subscribers are using the same domain, or if you control the authentication settings of all domains sending to the list and then do not modify the mails).
Ideally, also implement ARC. This can serve as a way to distance yourself from spam that might be fed into your list by a malicious or compromised subscriber (to a degree!).

> If it's only UBE, why do so many people give up running mailservers because they can't send emails?

Presumably because they try adjusting to modern requirements by trial-and-error. This does not work, because modern anti-spam and reputation systems are highly dynamic. Internally, they use counters on many aspects of a mail and thresholds of those counters determine how much influence (non-)adherence to requirements has (or if a requirement even applies).
Also, parameters of those systems are typically per recipient. You can't send a mail to a test account and then assume that the same mail would be treated the same when sent to any other account.

Alternatively, because a lot of their person to person communications has shifted to instant messengers and the server isn't as valuable anymore.

Ultimately, running your own small server can actually be more reliable than before, in particular if it serves distinct senders with individual domains, because with all this authentication stuff, reputation moves from sending IP to sender domain. But you will still get into IP-related trouble if the ratio of unwanted mails from your IP is too high.