Indeed, if your coordinator isn't a replica you obviously need it to fail as well in order to get data loss -- I was thinking of the symmetric case where every replica is a coordinator and W=1 reduces to "store locally, broadcast, and send an ACK".
And yes, of course there are hairy ways to solve the problem; I just would have liked to see it mentioned so that people realize that it exists.
You make a good point. Without additional safeguards/depending on implementation, W=1 can indeed mean "durability of one".
This also depends on your failure model. If your node crashed (RAM cleared) and your durable storage broke, you're in trouble. If the data was durably persisted and you just have to restart the node, it's a better situation.
And yes, of course there are hairy ways to solve the problem; I just would have liked to see it mentioned so that people realize that it exists.