Velojet, as explained it's just as bad as most cheating in games anyway where players can self-host servers or even MMO's with "anti-cheat". A slightly easier way to fix it is to simply not tell the player they are the host, and to try and pick host by best ping.
Anti-cheat doesn't work - see gunbound/maple story.
rexrainbow, ah awesome, and yes I know it's not MMO quality, but I'm thinking of smaller games that people play with their friends rather than massive multiplayer.
As Ashley said, if he cannot make it work seamlessly with the same quality as the rest of the product, he'll leave it to third party developers (that's us).
Ashley, ah okay I know exactly what you mean, but I think it'd be really important for built-in support for even the most basic of things like online chat, for those who aren't worried about extreme cheating or aren't basing their game around the multiplayer aspect.
It'd still involve a server, and custom server-side scripting, since p2p html5 tech isn't there yet, and even if it was, we'd have no way of identifying multiple instances of our game without a server.
It could be like exporting Windows 8 apps instead though, where you go into the exported files and change a value or two to make the server work specific to your game.
a value or two or thirty or a hundred thousand...
The only way I see this working is an exporter for node.js that skips drawing - as awesome as I think that would be, the work/benefit ratio isn't as good compared to, for instance, new objects/plugins (deep json parsing, rich text boxes, skinnable form controls, etc), more powerful eventing (custom ACEs, callbacks, widgets, modules, etc), IDE options (collaborative editing, search'n'replace, tiled level editor, AI editor, etc), developer possibilities (Hooking into the interface, plugins able spawn dialogs) or even working on the site (improving the store and the arcade, for instance). Thoughts Ashley?