Panic level: severe
The defense did pretty well, holding Michigan to 4.3 yards per play (only the 5th worse performance they've posted, amazingly) and 10 points (tied with Utah for the second fewest). The point total is particularly impressive given the favorable field position Michigan had all night, with an average drive starting at their own 42; Michigan's touchdown drive started from the Northwestern 21.
As usual, the offense was a mess. Michigan's run defense is excellent and it showed, with Justin Jackson managing only 38 yards on 17 carries. Northwestern did manage to string together a couple of late scoring drives with the quick passing game, but Trevor Siemian's 5.6 YPA, 2 interceptions, and 5 sacks taken (to go with a sixth from Matt Alviti) represented an awful day overall. Overall, NU scraped together a measly 3.1 yards per play and failed to capitalize when they had favorable field position.
Finally, the special teams: Jack Mitchell missed a 36 yard field goal in the third quarter, Chris Gradone was pulled after two punts to be replaced by Hunter Niswander, who mishandled a snap and gave Michigan fantastic field position off of a left-footed desperation boot. To be fair to Niswander, the snap was terrible, as they were all game, so long snapper can be added to the list of specialists that need improvement. Finally, Tony Jones muffed a punt, leading to a Michigan recovery on the Northwestern 21 and their only touchdown.
After doing nothing on offense all game (longest drive before the 4th quarter: 38 yards, turned over on downs; total yardage: 95 yards), Northwestern somehow strung together a 19 play, 95 yard field goal drive and a 14 play, 74 yard touchdown drive in their final two possessions. Down 10-9, Northwestern was faced with a choice: go for two and the win, or kick an extra point and try in overtime.
The decision to go for two has been much agonized over, but it is defensible, at the least: Jack Mitchell is converting 89% of his extra points, so, assuming that you think overtime is a 50/50 proposition, kicking the extra point gives NU about a 45% chance to win. This is the baseline; I would stick with 45% as the bar to clear, but you could convince me to go lower on the OT odds, given NU's struggles with negative plays and in the kicking game.
While the game management decision was reasonable, the implementation was more than a bit dodgy. It did not surprise me to hear that Michigan expected a rollout to the right; while the cosmetics change, variations on that theme have been NU's go-to two point play during Mick McCall's tenure (the most entertaining came in 2009, when the Outback Bowl featured a wide receiver pass variant). In a vacuum, it isn't a terrible goal line option; with Northwestern's recent two-point history, you had better be damn sure it will be perfectly executed, since an opponent would be foolish not to expect it.
This time around, Northwestern starts in an illegal formation, with Dan Vitale as an h-back on the left and no receiver on the line to the right.
Predictably, Northwestern shifts Vitale to the right as a tight end. Michigan adjusts their defense to have four defenders outside Vitale, with three shallow in apparent man coverage and a safety over the top.
After Vitale is in position, Northwestern motions Tony Jones from the right slot across the formation and back again; Michigan sends a defender with him, confirming man coverage, but the rest of the defense has no reaction.
At the snap, Michigan's coverage is flying to the offense's right; they don't plan to let NU beat them outside. On the line, Frank Clark takes an extremely wide rush, and Jack Konopka can't handle it, getting turned sideways instead of widening to make the block.
Clark is able to make contact with Justin Jackson at full speed; that goes a bit better than I would expect, but Siemian is forced to pull up. He falls over, and the game is over.
Even had Siemian been able to stop his momentum, this was not a promising play; the frontside receivers were blanketed, leaving Cameron Dickerson coming across from the left and Vitale leaking out to the left flat, while the backside protection was already beat, meaning that Siemian wouldn't have had time to get the ball to either of those players.
Goal line offense is naturally difficult, since the compressed field makes it easier for defenses to deploy their numbers effectively. But while it will always be difficult to find a good look at the goal line, Northwestern's focus on using a shift and motion to try to gain an advantage over the defense allowed Michigan to commit heavily to plan A, the rollout right. Returning to the presnap formation, we find Cameron Dickerson one-on-one with no help available on the left as well as a reasonable run look in that direction (Michigan has 6 in the box plus Dickerson's defender to take on 5 linemen and the receiver, but the linebackers are far enough inside to make outside zone a hopeful option). One of the chief advantages of spreading the field is that it forces the defense to declare their intentions by positioning; this only works if the offense is willing to take advantage of the information revealed.
Let's just agree not to talk about it.
Panic will stay the same if: the Earth keeps spinning
Panic will decrease if: LOL