They tried that with BLOPS2 and it failed miserably. You pretty much contradicted yourself there with...
Whether it was in emergency or you left because you were having a bad game, your stats should suffer.
My issue isn't that. But there are situations where things are beyond your control. For instance, this happens on occasion with my games... I would be in first place often with a very high kill death spread and my game would magically decide to freeze (I suspect this may be due to a glitch in the game if the host decides to rage) or I get booted from the game. I end up taking the loss even though my team was ahead by a lot and basically the game dashboarded me against my will. I get more annoyed at that than other players dashboarding TBH.
What Doc said - sorta.
There was an EXTREMELY long and well-debated discussion about this issue. I loved the way BO2 worked, which was the DIRECT result of that conversation. I may even have the bookmark somewhere (though the overall thread, which discussed more than backouts/dashboards, was far in excess of 300 pages long) if you want to go back and revisit the debate. In fact, everything that Doc mentioned was discussed. It was decided that "connection losses," regardless of the cause, warranted a "probation" for the simple fact that if there are things beyond a player's control, the player needed to stop playing until his/her gaming environment was back within his control. Otherwise, as the OP noted, the game is "ruined" for up t 17 other players (back when GW allowed 18 total players).
The community at-large did not like the probation feature. I did and still do.
The thing is, in addition to what Doc said, it turns out there is something else that plays a major factor in "lost connections," whether dashboards or back outs. It is something that has not yet been directly addressed in any game I am aware of and certainly has not been addressed in the CODiverse. That "something" is party play vs. randoms. The fact of the matter is that most people complaining that other players are leaving the game are players who game with a party and the gamers leaving the match are players who tend to run solo.
And the fact of the matter is that there really is no way, commercially, to address the issue. We've seen that in two seasons of COD with competition-specific game modes available. League play, I think, has been an effort to separate parties from randoms.
Frankly, it just does not seem to be going over very well at all. Part of the problem is that in an MLG-environmnet like League Play, parties that can "own" teams of randoms face two distinct hurdles. The first hurdle is that they are no longer up against disorganized randoms and, as a result, simply cannot perform as well. That removes the fun of seeing your party of six score a combined 600 kills against the other team who scores a combined score of ... 50 kills in a game mode like Dominatin, which is not supposed to be a kill-centered game. The second issue, as we've seen in Ghosts, is that much of the mega-kill streak reward menu is either not present or too hard to achieve when playing against organized parties.
The bottom line is that if most of your gaming is with a party of friends, you're just going to have to accept this aspect of the COD culture.
Personally, I run solo and run into parties that are too good to beat only occasionally. Most parties are not nearly as good as they think the are. Still, they are good enough to score the win. Thus, up against a party, if I don't leave before the match starts, I focus strictly on my KDR and minimizing the deaths my random teammates score and just ride it out until the match is over.
Doing so, I've maintained a +0.500 WLR since at least MW2 and a 1.5+ KDR since BO1.
And I back out what I consider frequently - about 12% to 15% of the matches I'm in. I don't dashboard (except for a couple of months during MW3 strictly to make a point during that discussion I mentioned earlier).