Mike Mearls talks about plans for the Wizard in his latest D&D Next article, Balancing Wizards in D&D.
One of his ideas is that using a scoll would require expending a spell of that level to use; so the idea is that scrolls wouldn’t increase overall power, but they would allow for more versatality. This sounds like a great solution for allowing a wizard to get some use out of those corner-case spells that have utility only in specific situations, while still maintaining overall power balance.
This implementation reminds me of an idea Paul and I discussed a couple weeks ago for how to handle the same problem. It could be used in addition or in place of the above system for handling scrolls:
A wizard can cast ANY spell they know by expending a spell slot of that level and increasing the casting time to 10 minutes.
Thus, when you memorize your spells for the day, you can focus more on spells you know are going to be useful, such as offensive or powerful utility spells, since you can always cast the more situational spells, such as Stone to Mud or Disguise Self, out of combat if they are needed.
Of course, there are still going to be some situational spells that you may want to memorize since you’ll be able to cast them in a single round, so it is not necessarily just a matter of loading up on your most obviously powerful spells, especially if you have reason to believe (such as by advance scouting or information gathering) that a certain spell will be useful.
The main advantage of this approach (or of the new approach for scrolls) is that it allows more of those fun wacky spells that rarely see use to come into the forefront when they will really be needed out of combat. When I am playing a wizard in 3.5 or previous editions, there are certain spells I never touch until higher levels (when I have enough lower level spell slots to diversify); it would be fun to see those spells coming into play earlier on.