After a quick look, IMO Watermill is a bit more flexible, because of middlewares and decorators support ;)

And what is unique, Watermill provides some high-level concepts like messages Router or out-of-the-box CQRS support which is really helpful when you are building bigger application.

I'm too lazy to back with some links my following statement, but usually in industry diversity is good. There are many frameworks and libraries, that have overlapping use cases. But with more choice it's easier to get the right tool for the job. Also maybe this framework is solving the same problem in a more clever way. So: yes, it's good to have smart developers contributing instead of starting new projects, but it's also good to have more diverse frameworks to choose from.

Even a cursory glance tells me the scope of this projects plans outstrip a pub sub protocol implementation. Maybe that's why?

not sure exactly what you mean but there is a driver package that can adopt other protocols through the same interface.

https://godoc.org/gocloud.dev/pubsub/driver

Even if it was a different implementation, I wonder if it could align with the open source cloudevents spec (https://github.com/cloudevents/spec)