Maybe off-topic? But since Ashley brought up the poor multiplayer plugin... it is actually pretty feature complete, I think people just have trouble wrapping their head around multiplayer concepts. They have some preexisting assumptions about how it should work, and walk away discouraged before realizing the multiplayer plugin is pretty adaptable.
Someone mentioned missing directly connecting instead of via signalling server, but utilizing a signalling server is pretty much superior any way you look at it (users ARE directly connecting to each other, the signalling server just takes the work out of making that connection). For example if you want to connect to someone by IP, use that as your room name instead.
The second big thing is understanding lag and latency and the fact that no two users will ever see the same thing at the same time. The tutorial spends a lot of effort explaining this and the implications AND the solutions, but I think it turns off a lot of people. This isn't something you can just wish away or ignore when developing multiplayer games, but people go in to it not realizing how important it is. This is no shortcoming of the plugin itself, rather the users. The plugin actually has tons of built in behaviors to make things like interpolation and lag compensation easier to handle for the developer.
I think the last thing is that many people envision server-client architecture for their game, which isn't exactly how the plugin was positioned but it is perfectly capable of doing so. Probably could use a specific tutorial to set up such a system (kind of like how there are two basic tutorials for top down shooter and side platformer - one can be foor peer-peer and another could be for client-server). Doesn't mean it is superior to a fully featured backend service, but those are out there already with working plugins.
Back on topic - while the multiplayer plugin might be under utilized, I think it is still a huge selling point, especially for "serious" developers down the line if some of the other issues get addressed. Even as a feature checkbox, I think not having it would be a fair reason for a prospective customer to turn to another engine.
TLDR: Basically wanted to say voting on features is important regardless of utilization, it usually is a fair metric of what people are looking for, regardless of weather or not they are capable of utilizing it.
Since I haven't weighed in on this thread yet, just going to mention in my opinion the two biggest priorities for "serious" developers down the line would be exporting (already/soon to be addressed) and monetization (ad service support/tutorials).
Also I think an official perlin noise plugin/function/support is something a lot of people are looking for and trending recently, but I suppose a voting system would show if it is or not.