I've kinda changed my mind about this a few times, but my stable position for a few years is that email is not broken at all. If it looks like it wasn't designed for what it actually does, it is because humans and human societies aren't designed for what they actually do.

The two effects sort of cancel out.

Also email is not an issue for the vast majority of people. It is beyond a first-world problem. It is a first-world CEO type problem. The people for whom it breaks are typically themselves impossibly overloaded crutches for organizations teetering on the brink of collapse. Fix the orgs, email will fix itself.

PG's personal mental model of email as a todo list which should be made officially so is dangerously bad. It works well (or like democracy, better than anythingelse) as a todo list precisely because it was not designed for the purpose. Anything that is specifically designed for the purpose is likely to fail because people think they understand to-dos and put together designs that reflect their own narrow understanding of decision-making. Invariably they fail.

To work better than email for todos, a new protocol would have to be simpler than email. Increased complexity, inevitable with any positive definition of to-dos, is doomed.