Input lag wouldn't show up on the ping meters. By input lag, I mean that ALL of the game's controls have delayed responses for both the client and the host, that way lag issues can be resolved before they have to be prematurely displayed on-screen. If the networking lag ever becomes greater than the input lag (ping spikes), the entire game will freeze up to let the netcode catch up, then play resumes.
With this kind of netcode, you'll be able to play as if it were LAN no matter how much lag and latency there is as long as it stays below a set limit. There's no host advantage, and there's no predictive system to give players a false image of his opponent (although I hear that some engines use a predictive model on top of an input system to decrease input lag and/or reduce game-freezing lag).
The obvious problem with this kind of system is that if both players don't have a good connection, the lag can ruin the game. Play with more than 2 players is almost impossible with anything less than stellar connections.
This kind of model is useful for fighting games, because when trying to judge by your opponents moves and counter, host advantage and predictive models usually don't work well. Even if it's only incorporated into a more sophisticated model and doesn't bring as much to the table as a full-functioning input lag system does (which it shouldn't), some input lag for online play would make Brawl much more playable online, IMO. |