In the original Final Fantasy there was a mechanic that meant that if one of your characters were selected to attack an enemy that was either killed or had fled before it came for the character to take his turn then the character would simply miss. In later games then this was altered to make it so that characters would automatically select a new target.
This was rather a pity. The original mechanic had many strengths including the relative minor ones such as reducing the influence of luck; good fortune that kills an enemy prematurely won’t necessarily allow you to get a start on the next enemy. More importantly it can increase the closeness to pen and paper games as playing optimally requires, well, pen and paper (or a spectacular memory).
With auto-targeting the player can simply rapidly press the confirm button. Battles are rather time wasting affairs, and increasing levels are to be celebrated and made happier as a reward for it. Without auto-select then one has to know the optimum attack selections for an enemies health, this trivial mathematical operation, requires knowing your attack range. Going up a level influences tactics, it becomes embedded in your thinking. Auto-selection separates the player from the game.
The ease of being able to make that argument is one of the hardest parts of solo game design*. This is for something that has actually annoyed me as I played, and that I had no hand in making. For stuff that I have made and that haven’t annoyed me, the temptation is even bigger. In commercial coding, there are a variety of testing mechanics that one can utilise to keep from letting that self-love from getting the way. However, games are fluffier things. Iterative design also gets in the way of a pre-plan to compare the product to.
I guess that’s why the present is a present of teams. Apparently, that didn’t solve the problem for Square.
*The actual hardest part is of course art. The hardest art style being that of geometric squares, no where to hide and demanding perfection.