There was a post I read a while ago detailing it but I forgot where. The gist of it is that it’s a custom protocol and format that goes back decades. These days clients try to implement lots of features client side that were not imagined back when the protocol was created so it involves massively inefficient actions.

I forget the exact action described but I believe it was renaming a folder requires downloading every email in the folder and then uploading them all in individual requests which takes minutes. Stuff like this plague the protocol and hold back email clients. Rather than spending dev time building the best client, they waste it wrestling a protocol that has long outlived its usefulness.

JMAP is sane, performant and easy to implement. To check for new email and resync email state with IMAP should make separate call for each folder. Even if the app tries to sync multiple folders simultaneously, the server often has a limit of max 2 connection. With JMAP this can be done in a single call. Microsoft's MSGraph API is similar but riddled with inner HTTP (429) errors in batch calls. This needs extra logic to handle all cases.

IMAP IDLE only allows you to monitor a single folder and it's bad for anything battery-powered - yet people want proper notifications about the mail they receive. Nobody implements IMAP NOTIFY which would fix this part for at least desktop PC's. There's also the silly notion that a single letter has to exist in a single folder (may only have a single label). Plus all the concurrency issues with multiple clients accessing the same mailbox.

If we haven't gotten rid of asking users for postal addresses when there is no chance they will ever be used, we are absolutely never getting rid of email. It's decentralised, there's a paper trail that doesn't disappear when a host goes down or gets compromised, and it's still a competent method of asynchronous communication. Plus, the whole world is never going to agree on the same new thing anyway.