Kessel can backcheck. He's getting a bit better slowly at playing physically along the boards. There were a couple instances where he used his body along the boards to play physically in the offensive zone. I have no doubt that under carlyle he'll be pushed to do the same defensively and we'll see a noticeable improvement.
The baseball-type error system is even worse than the current +\- system, since it makes it more 'subjective' rather than 'objective'.
The fatal flaw with the current baseball-type error system and what would carry over to your proposed system, is that it will not take 'range' into considersation.
What do I mean? Let's take your example. The defender who coughs up the puck will be docked a point, however, he's involved in the play. The player who is not involved in the play will not be docked a point. Therefore, the uninvolved player will be judged as defensively superior.
It happens all the time. Some players in baseball have the ball hit to them over 100 times less than other players at the same position, which decreases their chances of making an 'error'. You want the players to be more involved in the play, and possess more 'range' -- even if they make more mistakes. A player that gets to a ball and makes an error showcases more defensive-ability than a player who does not get to the ball at all.
The baseball-type error system under an NHL format would continue to suffer from this ignorance.
I don't see how it could possibly be any worse than the current +/-. Instead of penalizing everyone on the ice, it chooses to penalize a subset of the players on the ice, and potentially some players who are off the ice. Using the goal the Habs scored on us, and assuming it was an ES goal, Dion would get a mark for chasing the puck behind the net, and losing goalside mark on his man, and Kulemin would get a mark for failing to get the puck out, and for losing his mark, who ended up scoring the goal. The rest of the players didn't have much to do with the play.
It's more like using goals and assists who try to determine the contributors to the goal. A flawed system, that often gives out phantom assists, and fails to reward guys in front of the net screening goalies, and creating havoc. It's a flawed system, but sure as heck better than looking at how many ES goals for a player was on the ice for.
It would be a strictly defensive score, not mish mashed in with some offensive figures. Just because your line's producing offensively, does it really make the line better defensively?
I don't see how the range argument applies to hockey. Hockey players are much more mobile than baseball players. While defensemen/goalies are more likely to make errors that lead to goals than forwards, it's the same for all defensemen/goalies across the league. Therefore, it's a "stat" that could be used to compare players of the same position across the league.