The problem with HOI is that it's far too easy to move units, and move them fast. If I could have my way, I'd make the formula to determine unit speed much much more complex. You can easily put divisions ad infitum in a province or an invasion, when this wasn't really possible. To summarize, the HOI battle system, as complex as it is, is still rather simplistic.
The problem with the front AI is that the it has numerous tasks to keep track of, which often make the AI 'confused' as to what it should be doing. As things changed, the AI is constantly being updated, leaving to 'indecisiveness', where the AI appears to be bouncing the same troops back and forth, or suddenly launching a large offensive, then losing its initiative and stopping.
While and Amphib AI would be possible (and I'm sure it exists to some extent...), to make it a very strong piece of code would take an incredible amount of time... making the AI handle land operations at a mediocre level is hard enough; effectively coordinating between land and sea over great distances would be a nightmare (as evidenced by the HOI AI).
The problem with the front AI is that the it has numerous tasks to keep track of, which often make the AI 'confused' as to what it should be doing. As things changed, the AI is constantly being updated, leaving to 'indecisiveness', where the AI appears to be bouncing the same troops back and forth, or suddenly launching a large offensive, then losing its initiative and stopping.
While and Amphib AI would be possible (and I'm sure it exists to some extent...), to make it a very strong piece of code would take an incredible amount of time... making the AI handle land operations at a mediocre level is hard enough; effectively coordinating between land and sea over great distances would be a nightmare (as evidenced by the HOI AI).