McClellan's primary concern should have been for his own performance which lacked aggressiveness, speed of action and an excuse making attitude. The two, and now apparently three blah, blah, blah's have totally missed the point; Mac was a failure.
An army on the scale of a hundred thousand men cannot be controlled effectively as a unit. It is
essential to delegate much of the command to subordinate officers; what do you do when many of your subordinate officers, who you cannot replace, are men who your evaluation concludes are poor?
This is a point that needs to be emphasized, I think. Armies are geographically
large and information and orders take time to travel between components of them; even if the Army of the Potomac at that point had all been on one field and had formed a single continuous line of battle, with half the force in reserve behind the first line, it would have been approximately eight miles wide and it would have taken information
an hour to get from one end to the other by horse; if McClellan is in the middle then he may well not hear about the enemy doing something on one flank for thirty minutes, and then his orders to do something about it would take another thirty minutes to arrive.
Notably, McClellan attempts in many of the battles he commands (the actions around Richmond from Seven Pines to Gaines Mill, Glendale, Antietam) to use semaphore and other forms of signalling to reduce that lag time; similarly standing on a hill, if one is available, means that it's at least possible to
see things happening and cut down on the decision loop. But conversely if he's trying to do something operationally noteworthy (which is what he's doing with half the army slated to move to Eltham's Landing while the other half engages in what will become Williamsburg) then no matter where McClellan is some large part of his army is going to be
hours away from him.
This means that it is
completely unavoidable to have subordinate commanders with wide latitude. This is a natural consequnce of the increasing size of armies (a full-strength division deployed in single line of battle is on the order of a mile wide, a full-strength corps on the order of three, and an army of multiple corps just keeps getting wider; furthermore, corps-sized forces may operate
independently and march along different roads entirely, sometimes called
battalion carre and resulting in corps being a full day's foot march apart) and is one of the reasons why corps command exists, but the corps commander having specifically a wide latitude is because the corps commander is likely to be many miles (and consequently upwards of an hour) out of direct contact with the army commander, and things happen on battlefields which require good decision making on a shorter timescale than that.
That is why (as was the topic I was discussing) a key part of McClellan's
reaction to the outcome of the Williamsburg operation was to put at least some of his force under the commanders who he feels he can trust. McClellan originally trusted Sumner to handle Williamsburg while McClellan was doing something else (something equally important to the army, since it was the move of half of it to Eltham's Landing, as planned); since McClellan cannot be in both places at once he delegates one of them to a senior commander. He's actually giving Sumner the task which Sumner should have been better suited to, the fairly conventional pursuit operation not the odd littoral blocking manoeuvre.
As actually happened, things went wrong (Union over-commitment to attacking) roughly until McClellan showed up and reinforced Hancock. But this only emphasizes the need for there to be at least
some commanders who can be trusted to work okay where McClellan is not - and that's what drives McClellan's modifications to the original structure.